Table of Contents

• AutoTOC

Intro

Intel is finally trying to unbend its CPUs, despite having to be on a bender to buy a $630 285K right now. Today, we’re using our laser scanner to look at the deflection in the CPU heat spreader from the different loading mechanisms, including these scans of the 285K (read our review) and 245K with different coolers installed. Today’s testing also includes specialized pressure scanning to produce pseudocolor images of pressure distribution across the IHS surface, very brief thermal testing to look at the differences with Noctua’s LBC (Low Base Convexity) flat coldplate, and we’ll look at the mechanical aspects.

Editor's note: This was originally published on November 4, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing, Host

Mike Gaglione

Camera, Video Editing

Vitalii Makhnovets

3D animation, Camera

Andrew Coleman

Writing, Web Editing

Jimmy Thang

---

• 

• 

• 

• 

• 

Unfortunately, Intel’s new and better ILM is optional. It didn’t force motherboard manufacturers to use it, so they can still cut corners if they want to save a few pennies. The new ILM is called the RL-ILM, or Reduced Load ILM, with the old one being referred to as the “Standard” ILM (indicating an assumption of it being the default). Our Z890 Hero ships with the RL-ILM, as do most high-end boards, so we used it as a test platform to swap to other official LGA-1851 ILMs for comparison.

Let’s get into it.

Differences

We’ll get some basic education in and go over the differences:

CPU sockets are one part mechanical and one part electrical. Intel uses what is called an Independent Loading Mechanism for its socket. Some people include the ILM when referring to the socket. On a technicality, the literal socket is the Land Grid Array with the carrier that actually holds the CPU. The loading mechanism is the mechanical part of the socket.

Intel is shipping 2 types of ILM, RL-ILM and Standard, and it is using at least 3 different suppliers that we’re aware of to manufacture these. Our Z890 Hero came with an RL-ILM by Lotes, which is a long-time supplier of ILMs. We also have the same-brand ILM of the Standard variant, plus the two other suppliers you’ll find on boards.

RL-ILM vs. Standard ILM

Here’s a CAD render of the socket. The standard ILM has an angle that increases the force application along the edges of the CPU. That’s the real difference here. This is what causes the depression we’ve seen in previous 3D laser scans we performed. These scans are from our past content: You can see how the ILM causes significant bending and forms a central concavity with the heat spreader, leading flatter cooler coldplates to be worse on Intel despite being better on AMD. You can learn more about that in our previous coverage here and here. 

Back to the CAD model, the RL-ILM is basically just flat. This is the biggest change, as the force should be reduced. This is also why Intel requires a higher force heatsink to be installed in order to ensure contact.

The RL-ILM also has one other difference: There’s an additional adhesive spacer on the underside, which can be thought of as similar to the washer mod that Noctua now ships with its NH-D15 G2 coolers as an option. The additional spacer goes underneath the existing black spacer, meaning that the ILM "leg" component probably was taken from existing Standard ILM stock, then retrofitted with effectively a sticker.

3D Animation

In our original Thermal Grizzly contact frame benchmark, we showed how the ILM clamp appeared to apply very slightly higher pressure to one side of the socket. This was exaggerated by the fact that the ILM has some play in it, where it can shift side-to-side and be repositioned and we saw that still happens on the RL-ILM. 

• 

• 

• 

Here’s our 3D render of the standard ILM: With the CPU dropped into the socket, the standard ILM uses a hook that’s attached to the lever to centrally press down on the ILM lid that clamps directly to the CPU IHS. With the lever fully down and secured, the ILM is now secured at 3 points: 2 on the bottom of the ILM and 1 at the top. All of this is the same on the new RL-ILM. 

• 

• 

As for the CPU, the ILM has two wings that press down on the IHS at the borders, and with that curvature we showed in the CAD model, the force application at these points is high enough that a highly precise gauge can show how light is able to shine through despite the CPU being relatively flat when unclamped and looking flat to the eye.

We’ll refer you to our Thermal Grizzly Contact Frame benchmark from 2022 to learn more about this older style of ILM.

For the new version, clamping the CPU in the socket functions mechanically identically for the end user, with the lever pulling down to hook under a securing latch and clamp the ILM at 3 points, with 2 main contact points at the wings of the IHS. However, the lack of a bend in the ILM reduces the load. Intel still has to keep the force high enough that the CPU’s pads make contact with the socket pins, but has to be careful that it’s the right amount.

Too much or too little force can cause boot issues and high clock memory stability.

And that’s really it for these ILMs.

Pressure Scans

• 

• 

• 

Noctua just got done spending literal years developing its new NH-D15 G2 and shipped it with 3 different coldplates, which makes it a unique candidate for pressure testing. 

For pressure testing, we take the different ILMs and apply a special pressure paper between the CPU and the coldplate. We then take that and scan it in with a specialized pressure scanner to create pseudocolor images of the pressure distribution.

Pressure Scan Noctua Results - HBC on RL-ILM vs. Standard

Here are the results for the two ILM types with the HBC cooler.

The new Reduced Load ILM with the high base convexity Noctua coldplate yielded low pressure at the outer edges, but especially toward the top of the board near the VRM and EPS12V cables. The pressure centrally remained high; however, because the CPU should be flatter with this ILM, the Noctua cooler ends up with less evenly distributed pressure because it’s designed for a different scenario.

The standard ILM with HBC cooler scans reinforce this: The HBC cooler ends up with more evenly distributed pressure at the top and bottom edges of the CPU IHS.

Pressure Scan Noctua Results - HBC, LBC, Standard

And here’s only the RL-ILM with the 3 Noctua cooler cold plates.

The RL-ILM pressure distribution was the most evenly distributed with the standard and LBC solutions. The two are mostly indistinguishable for distribution, although the precise pressure centrally will influence the results in thermal testing.

The LBC cooler had slight gaps at the left and right edges, but consistently square distribution at the top and bottom corners, with good pressure across the entire center. The standard cooler had less consistent pressure at the top and bottom edges and similar gaps to LBC at the edges. Ultimately though, these two basically look the same for contact.

Laser Scan: Noctua Coldplates

Our Noctua NH-D15 G2 review went into depth with laser scans of the cooler’s coldplates, and that hasn’t changed. We’re showing the LBC, Standard, and HBC scans again briefly here just to help recap the impact because when we’re looking at pressure and how it is affected by the ILM, the cold plate is a part of that equation. 

And now we’ll scan the new Intel CPUs to see how their shape matches the pressure scans we saw earlier. 

285K & 245K Unsocketed Laser Scan

• 

• 

• 

Here’s a look at the plain Intel 285K when it’s just flat in the laser scanner. The CPU isn’t in a socket at all here, so this is as simple as it gets. Even in our 2D screenshots of the 3D scan, we can see the letters -- the CPU IHS is so flat that the very slight indentation for the text is visible.

The IHS itself has a few higher points, one just off-center, one along the right edge when oriented in a legible orientation, and one just off the left edge of the CPU.

Magnifying it 100x, that coloration grows to form just a few high points. Overall, it’s flat, but at high magnification, some small deviations appear. One thing that is clear though is that there is no substrate curvature, which makes sense since it hasn’t been socketed.

Let’s create a grid with the 285K and add the 245K to it. The 245K (read our review) follows a similar pattern: Centrally, it’s a little higher, then just off-center right it’s also slightly higher at 1x. Adding 100x to the grid, there’s a similar pattern as with the 285K.

Finally, we added our unsocketed 12900KS from the golden sample coldplate story. It’s still flat when unsocketed, but the difference in IHS design is also slightly showing through.

Socketed Testing

• 

• 

And now, we’re going to throw this Z890 Hero with the new ILM into the scanner and socket the CPUs in it. We obtained this package of ILMs to test. The ASUS board uses the Lotes RL-ILM, so we’ll start with that one.

Standard vs. Low Pressure Socket - 3D

• 

• 

Here’s a one-to-one 3D visualization in Blender taking STL files from our laser scanner, showing the 285K with the standard ILM first. As usual, 1x magnification doesn’t show much, but bringing that to 100x quickly shows a deep concavity centrally, just like we saw with LGA 1700 CPUs.

• 

• 

Switching over to the new reduced load socket, we can see that the 1x to 100x magnification shows less of a pronounced curvature of the IHS itself. It’s still curved, but much less, with the CPU maintaining a more consistent height instead.

Socket Comparison - Grid (285K)

Here’s a grid comparison of the different ILMs on the same motherboard, tested with the same CPU -- starting with the 285K.

You can see that the Standard ILM at 100x magnification shows a huge deflection centrally, as we’ve seen before, with higher pressure on the far ends of the mechanism. While we can sort of see the slight ridgeline down the middle of the CPU, the bigger issue is how deeply it indents.

The reduced load socket is significantly flatter, with less of a central deflection. The ridgeline in the CPU IHS becomes more pronounced in the graphic because it is more consistently the highest thing in the image. Remember that this is at 100x magnification, so the differences are exaggerated intentionally.

Socket Comparison - Grid (245K)

Unveiling the 245K results in the grid, we see the same patterns: The standard ILM deeply indents the CPU centrally, deflecting and deforming it in a way that coldplates with matching convexity will cool it the best. The reduced load socket is flatter and more consistent, though is still slightly deflected centrally.

Laser Scan: ASUS Cooler

• 

• 

We need a laser scan of the cooler coldplate before moving to the pressure maps, as the cooler and IHS alike contribute to the pressure distribution. 

This laser scan shows the coldplate of the ASUS Ryujin liquid cooler, which is what Intel sent with its CPUs to reviewers. Other coolers would fit, but we wanted to test what Intel officially endorsed.

At a 1x scale, the ASUS Ryujin coldplate looks relatively flat, but still shows a protrusion dead-center, gradually reducing height towards the outer edges. When we did our in-depth testing on Intel performance with varying custom-made coldplates from Scythe, we found that this pattern often did well for Intel.

Scaling it 100x, we get this almost comical tower protruding from the coldplate. This helps us see the steep slope as ASUS applies massive pressure dead-center with its coldplate design. This is sort of a hamfisted approach and version of what Noctua did more precisely with the D15 G2 for LGA 1700, except Noctua had more nuance in the exact shape of the convexity, which will better align with the concavity in LGA 1700 CPU heat spreaders previously. It’s similar to what we saw in the $60 Thermalright liquid cooler, which managed to brute force its way in performance thanks to similar protrusions.

Pressure Testing Results

Time to look at some pressure scans of the ASUS cooler with the new Intel ILMs.

These images show a new pressure scan of our 14900KF with the ASUS Ryujin cooler and the old (or “standard”) ILM. In this scan, you can see the 14900KF has narrowing pressure on the left and right sides, with most of its pressure centered. That’s where it should be, and most of that is thanks to the comically protruding ASUS coldplate, but fuller coverage is ideal. The older IHS also is a little bit different shape than the new one. The second column represents the pre-installed reduced load ILM using the 285K. Looking at the third column, adding the standard ILM with the 285K, doesn’t look too different. The pressure profile appears to be distributed taller and narrower. There’s still some of that slimming effect going on when we get to the left and right sides but not nearly as pronounced as with the last gen IHS design and ILM. 

Ultimately, what we see is that ASUS’ older brute force approach gets a better pressure distribution on the prior LGA 1700 socket than on the new ARL 285K socket, which is thanks to the massive central protrusion. This is the approach Thermalright took with its $50-$60 liquid coolers previously as well. It’s relatively hamfisted, but works, whereas the more carefully shaped approach of air coolers like the D15 G2 and the Scythe FUMA 3 are technically a better pressure match; now, that said, a 360mm liquid cooler is still “better” (with regard to capability) overall, and it will cool better, but the Ryujin could improve with more purposeful coldplate shaping.

Thermal Test Setup

Thermal testing is up now. Full transparency that we’re keeping this really simple this time, mostly because it doesn’t take much testing to verify if there’s a difference at all.

We’re only running the comparison thermals with one cooler this time. The ASUS cooler is so heavily deflected that we’re not sure the comparison would be that useful, so instead, we approached it with what should be a worst case scenario: The Noctua NH-D15 G2 LBC, or low base convexity, which is the flattest of Noctua’s options. In theory, this should be the worst on the more deflected standard ILM+IHS combination and the best on the flatter IHS from the RL-ILM.

Other coolers could have more or less impact. Coolers with higher force application centrally and with more convexity would continue to compensate for design problems of the standard ILM, but we want to just run a quick evaluation on one of the uncompensated scenarios.

We are also not testing anything below the minimum spec Intel declares for the socket, which is a 35 lb. force from the cooler. Anything high-end that’d be paired with the current CPUs will meet or exceed this requirement anyway.

Thermal Results

Here are the results from a simple A/B test. For this testing, we did two full mounts and at least 3 passes to average the numbers. This allowed us to check for variance mount-to-mount. All our other CPU cooler standards and methodologies apply, like manually spreading paste, controlling the fan speeds, and fixing the voltages and frequency. We disabled all power and thermal limits and set a fixed voltage with fixed frequencies. We have a known power draw down the EPS12V and 24-pin ATX12V through the 4 phases that it has (without PCIe slot power). That allows us to get these numbers.

The result was 61.8 degrees delta T over ambient for average P-core temperature with the standard ILM and 59.6 dT with the RL-ILM, or the improved one, meaning about a 2.2-degree reduction when accounting for  ambient. Without the deltas, we were running the 285K in the 80s to low 90s because we disabled all TVB 70-degree throttle controls. Running the CPU hotter allows us to see more of a gap between the results. A CPU consuming less power with a stronger cooler would likely not show as big a gap.

Checking briefly with Der8auer as a peer review, we learned our results are roughly in-line with his own. The differences are aligned with cooler and heat load differences.

We observed a slightly lower core-to-core delta with the new ILM, but it was within error. The AVG all-core temperatures were not significantly different from the P-core temperatures in this one due to the proximity of the P-cores to E-cores in this architecture (combined with our adjustments in BIOS). 

So, as short as possible, the RL-ILM is about 2.2 degrees better at this heat load with this cooler.

Tutorial to Remove and install the ILM

Before we move to the conclusion, in case you buy a motherboard with a standard ILM and want to move from the high pressure standard one to the low-pressure RL-ILM, we’ll walk you through how to do that. 

If you are going to swap the socket, we recommend sticking with the same brand for the replacement if possible. In our case, we used Lotes. 

To begin, we recommend starting with the CPU installed to protect the pins underneath to mitigate the risk of dropping, say, a loose screw down into the socket. 

From there, unscrew the 4 screws. We used a regular T20 Torx screwdriver. 

• 

• 

Removing the screws frees the top and bottom pieces of the socket. It also frees the backplate. When you’re installing the backplate, it’s important to get the orientation right and to ensure that the plastic sticker side is touching the bottom of the motherboard and not the exposed metal side. The backplate also features a notch that aligns with the triangle that’s on the corner of the CPU.   

• 

• 

We found that it’s easier to install the lever arm piece first with its 2 screws. Once that’s in place, it’s time to secure the other side with its 2 screws. You don’t need a lot of torque for the screws. We recommend that you tighten them in a star pattern to evenly distribute pressure.  

Conclusion

We regularly see people online saying that some cooler is 3 degrees lower than some other cooler, so just a reminder here on how all of this works: That 2.2 degrees is specifically at the power load we tested and with the cooler we used, under the conditions we employed. It will be higher or lower based on how these parameters change.

As an example, we did one round of tests with all the Intel throttle controls still enabled and saw less than 1-degree of difference -- but that’s because it was just throttling itself to regulate the temperature.

The new RL-ILM is definitely an objective improvement in both the curvature of the IHS and substrate and of the temperature in our brief testing. The pressure distribution depends on the cooler more than anything and it isn’t always clearly better, but the thermal result tells us that the net result is positive.

Frustratingly, this is optional. Intel is not at a stage where it should be making clear, simple, easy improvements “optional” for motherboard vendors. 

Although we don’t want Intel or AMD to force certain lock-downs, like taking away overclocking features, we do think both companies should enforce a default or baseline configuration that is in the best interests of the consumer, with the option for the consumer to tweak as their motherboard allows once exiting default settings.

In this situation, we do think Intel should just bite the bullet and force the better solution. It may be a situation where board partners had already purchased millions of these older mechanisms. Regardless, Intel has at least improved its mechanism. It is technically slightly more expensive than the original ILM, but since we’re talking pennies, we’d like to see this forced in the next generation as the standard ILM since it is just better. Intel needs to stop taking a soft-handed approach to its partners and taking the small victories when it can get them.

This doesn’t kill the contact frame market, though: That’ll still provide uplift, as the RL-ILM remains a mid-step improvement without going full flat like the prior contact frames we’ve tested.

That’s it for this one. We probably won’t do a ton of Arrow Lake follow-up testing since it doesn’t make any sense to buy right now, but we may explore a few other features.

---

Table of Contents

• AutoTOC

Intro

Today, we’re rounding-up some of the best CPU coolers that we’ve tested in recent years, including air and liquid coolers.

This is a continuation of our favorite series each year, where we get to have some fun comparing only the best-of-the-best products we looked at and avoid the usual disappointments. But there was some huge news this year: DeepCool got banned and is now gone from the US market. With that, a power vacuum has formed among cooler manufacturers, and Thermalright’s former nemesis has been replaced with Arctic and ID Cooling aggressively trying to fill it. We can’t say that was on our bingo card.

Editor's note: This was originally published on December 15, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Camera, Video Editing

Vitalii Makhnovets

Camera

Andrew Coleman

Writing, Web Editing

Jimmy Thang

---

These tests will include results from our new 250W and 200W Intel heat loads in addition to our 200W and 123W AMD heat loads. The categories this year are for Best Overall CPU Cooler, Best Budget CPU Cooler, Best Value CPU Cooler, Best Noise-Normalized Thermals, and Best Mechanical Design. Let’s get started.

Setting Expectations

These Best Of articles are intended to get newcomers up to speed quickly, so we won’t go into the same crazy depth we normally do for the standalone reviews. Each of the coolers that has a review will be linked in the description below so that you can find the full details if you want them, including all the 3D laser scanning, pressure maps, and downsides. This article will be more of a flyover.

We’re testing both air and liquid coolers for this one. We’ve added a lot of coolers to the charts that aren’t in our prior reviews, specifically tested for this round-up, so there’s new data mixed with old in here.

Additionally, keep in mind that our noise testing methodology changes between charts. For the AMD platform, we use our old room-scale noise testing.

For our Intel platform we just added, we’re testing in our hemi-anechoic chamber with a lower threshold of 25dBA normalization in a noise floor around 14 dBA.

As always, for cooler round-ups, we can only discuss coolers we’ve directly had hands-on time with. There are hundreds of coolers out there. We have a pretty good cross-section of the big options, but if you don’t see exactly what you’re looking for, we’d encourage you to check other content and expand your research.

You can find the full charts on our CPU coolers Mega Charts page. As a reminder, gamersnexus.net is a free website that is totally free from third-party ads and funded by the audience. We just got done running another 18 cooler tests on a 250W and 200W Intel platform with noise-normalized results in our hemi-anechoic chamber.

That’ll set us up. If you want to check out our Best CPUs and Best Cases round-ups, those are already live. All the review links and product links are in the description. Let’s get into it.

Overview: Best CPU Coolers of 2024

Category	CPU Cooler	Review
Best Overall CPU Cooler 2024	ID Cooling A720 on Amazon
Best Value CPU Cooler	Thermalright Frozen Prism on Amazon on Newegg	Thermalright Strikes Again: $56.90 360mm Liquid Cooler | Frozen Prism Review
Best Budget CPU Cooler	ID-Cooling SE-214-XT on Amazon
Best Thermals (Noise-Normalized) CPU Cooler	Liquid Freezer III on Amazon on Newegg Noctua NH-D15 G2 on Amazon on Newegg	The New Best: Arctic Liquid Freezer III 360 & 280 CPU Cooler Review & Benchmarks Noctua NH-D15 G2 Review & Benchmarks, HBC & LBC Comparison, & Best CPU Coolers
Best Mechanical Design CPU Cooler	Noctua NH-D15 G2 on Amazon on Newegg	Noctua NH-D15 G2 Review & Benchmarks, HBC & LBC Comparison, & Best CPU Coolers

Best Overall CPU Cooler: ID Cooling A720

Amazon

Runner-Up: Arctic Freezer 36

Original review | Newegg | Amazon

Up first, our category for Best Overall CPU cooler this year. This category requires the pricing to be competitive, as it considers the value, the build quality and assembly features, ease-of-installation features, the thermal and acoustic performance, and everything else. Last year, we gave this to the Peerless Assassin 120, which is definitely a GOAT and remains one. This year, we’re giving it to two Thermalright competitors, depending on CPU: Overall, the ID-Cooling A720 gets the award this year. For our Intel bench specifically, we’re giving a tied rank award to the Arctic Freezer 36. 

• 

• 

• 

Let’s start with the Intel results and the Freezer 36, since that one is isolated as a winner to just this bench.

The most heavily weighted element in favor of the Freezer 36 (read our review) is its price. Arctic said it’d increase the price within just months of launch, but seemingly never did. Even today, the Black model that we tested is about $28 -- an incredible value considering its performance, at least on our Intel bench, and fiercely competitive with the Thermalright Peerless Assassin 120 that won our award last year.

Our new Intel 250W noise-normalized charts also include the Peerless Assassin 140, which was eliminated because we discovered quality control issues and variance between our two units that resulted in a 2-degree spread of results, which is actually huge. Unfortunately, one of the Assassin 140 units we have just doesn’t have a good pressure scan or laser scan, while the other is fine. We were able to prove the source of the difference was a combination of the coldplate and mounting hardware.

You can see the two entries for the PA140 here. It’s not because the white model is better, but because the other unit has contact issues.

Anyway, that eliminates the new PA140 even if only the best entry were here. Of the remaining coolers, the Freezer 36 is the best performer after Noctua’s D15 G2 HBC. This is impressive with the Freezer 36’s $28 price. The G2, on the other hand, is $150. Some of this comes from Arctic’s mandate of a contact frame for the Intel version of its Freezer 36, but considering that’s all included in the price, this is a strong positioning. 

One other note here: The D15 2023 model we have here performs a little better than our original D15 from a decade ago, which is a result of minor tweaks or refinement in manufacturing processes along the way. We talked about this in the G2 review. If you’re on an older D15 from around when it launched, it’s likely it is marginally different in performance today.

• 

• 

• 

• 

On to the A720: In our Mega Charts for 200W testing previously, the Freezer 36 fell behind compared to its Intel results with the frame. This reveals our choice that’s tied for the category: The A720, which was also among the top two performers for Intel and is the best performer behind the D15 G2 (read our review) in this 200W, AMD noise-normalized test.

The combined chart-topping performance in both our AMD and Intel test benches is what leads the A720 to the Best Overall rank, with its relatively high build quality and moderate price securing the position. 

At $55, with occasional drops to $50, it’s competitive with the best (like the $150 G2) and manages a 55.2 degree over ambient result on our 200W AMD bench. This improves on the GOATed Peerless Assassin by 1-degree, which does have a little bit of an advantage in price. 

The Intel result improved on the Peerless Assassin 120 (read our review) by 1.6 degrees.

We recently tested the Phantom Spirit as well, but it didn’t outperform our Peerless Assassin 120. The Phantom Spirit EVO has a slightly smaller tower than the Assassin and also has a disadvantage with noisier fans, found in our chamber testing. The Spirit’s fan RPM is higher than the Assassin, but that doesn’t matter when normalizing for noise.

Overall, for the A720, we think it has surprisingly good build quality. The fans are well-fitted to the cooler, with the central fan sinking to benefit VRM cooling as well. We like the simplicity of the black model that we have. The mounting solution is overall straight-forward and the bulked-up 7-heatpipe design helps with coverage across the IHS. The cap plates are a nice touch without overdoing the branding. The central fan also levels-out nicely with the top plates and each fan has rubber bumpers on the corners to reduce vibration, which benefitted it in noise normalizing. Its biggest downside is size, where the 163mm height makes it about 6mm taller than a Peerless Assassin 120. This could limit some cases.

We think ID Cooling and Thermalright both are worth seriously paying attention to in 2025, especially with DeepCool’s disappearance from the US and the Freezer 36 is also worth considering if you happen to go Intel.

Best Value CPU Cooler: Thermalright Frozen Prism

Original review | Newegg | Amazon

Our Best Value award this year goes to the Thermalright Frozen Prism. Value is judged by a combination of performance for the price, not just pure price. We have a Best Budget category as well for what is simply one of the best, cheap coolers.

• 

• 

• 

• 

• 

Thermalright’s Frozen Prism is an enigma: It’s $52.71 at the time of this writing. This is somehow even cheaper than when we reviewed it. It’s cheap enough that we remain skeptical of whether it may have some sort of defect down the line. In some ways, it seems too good to be true. Concerns we have might involve gunk build-up, but we have no evidence of any such issues at this time. What we do have evidence of is impressive performance: The Frozen Prism (read our review) was a competitive liquid cooler in our testing, brute-forcing much of its performance with a hugely protruding copper coldplate, almost comically so. Our laser scans made this protrusion even more clear, with our pressure maps showing how the cooler leverages the design to brute-force high pressure dead-center, which is where the silicon often sits. 

Even on AMD, which has less-centered silicon than Intel’s prior monolithic CPUs, the pressure spreads wide enough that it’ll catch everything.

In our 200W heat load thermal benchmarks from the review, the Frozen Prism 360 ran at 49.7 degrees Celsius over ambient when noise-normalized, whereas the best liquid cooler we’d tested at the time, the Liquid Freezer III, ran at 46.7 degrees Celsius over ambient. That’s a big difference. At the same time, achieving such close performance at under $60 was previously unheard of.

In our prior test bench configuration for Intel, using a 250W heat load on the 12900KS, the Thermalright Frozen Prism was a chart-topper when at 100% fan speeds. If you look at the 53.2dBA noise levels though, it was incredibly inefficient compared to the Liquid Freezer III 360, which ran at 39.8dBA. This is over a 2x increase in perceived noise to the human ear for the Frozen Prism, yet it drops only a couple degrees. That’s an inefficient trade-off, and although fan speed could be lowered to compensate for noise, it’s clear the cooler isn’t some mastery of thermal and acoustic engineering.

What the Frozen Prism actually is, though, is a mastery of cost engineering. 

• 

• 

The Frozen Prism doesn’t have any real quality-of-life features to speak of or anything abnormal. It delivers what it promises, which is cheap cooling. Our only criticism of the physical construction, beyond some cheaper plastic-y feel of some parts, was the stiffer tubing. This can be worked around.

Thermalright has become known for flooding the market with countless options, to the point where it’s overwhelming even for its own staff to remember them all at tradeshows. The company is saturating listings and driving prices down, which is ultra-competitive and in some ways good for consumers. It’s also pretty cutthroat, and is why we now have name-brand coolers also dropping in price. Following-up its success with the Peerless Assassin 120 in years past, Thermalright is looking for a repeat in liquid. We’ll see how the Prism ages, but so far, it’s competitive, especially in thermals. That’s why the Thermalright Frozen Prism 360 gets our best value award.

Best Budget CPU Cooler: ID-Cooling SE-214-XT

Amazon

• 

• 

• 

The next award is in the same vein, but simpler: Our Best Budget award goes to the ID-Cooling SE-214-XT ARGB and it’s challenging the GOAT in Thermalright. ID-Cooling is vying for Thermalright’s strategy. Last year, we praised it for its overall value -- though it has been shifted into the pure budget category with the Frozen Prism’s addition to our charts.

The SE-214-XT ARGB has dropped in price and is now somehow $15.19 at the time of writing. We normally wouldn’t cite the pennies, but when 19 cents is over 1% of the total price, it suddenly becomes relevant. The SE-214XT ARGB is a simple, 4-heatpipe cooler that revives the approach of the old Hyper 212. It’s dirt cheap, its quality is flimsy, its plastics feel like they’re from McDonald’s toys, its coldplate is spartan, and yet somehow, the thing can handle moderate heat loads. It won’t handle a 250W CPU in our test suite with any level of satisfaction, but it’s good enough for cheaper and lower power CPUs with users on more extreme budgets. This would also be a good consideration if you’re buying a used CPU to save money and need something that ticks the “good enough” box.

In our 123W heat load on AM4 last round, the ID Cooling SE-214-XT ARGB held 58 degrees delta T over ambient when at 37.9 dBA, which had it more noise efficient for the result than the lower-ranked Hyper 212 RGB cooler. It was bordering on Noctua’s aged NH-U12S (watch our review). The Thermalright Assassin Spirit (watch our review) ran at the same noise levels and a little more than 1 degree cooler and is likewise a cheap cooler, but at $16.79, somehow, and this is really weird to say, it’s almost 11% more expensive. What a bizarre world of coolers we’re in.

In our 14900KF 200W heat load and with 100% fan speeds, tested instead in our hemi-anechoic chamber that purchases from the store help fund, we landed at 60.6 degrees Celsius delta T. That had the SE-214-XT ARGB as the worst performer on the chart, but still somehow capable of holding a stable operating temperature. In other words, it’s fine. The next lowest performer is the Scythe FUMA 3 (watch our review) with its combination of two fans, a massive 9 degree improvement.

Our 25dBA noise-normalized testing still has it at the bottom of the charts for this heat load, but even at this slightly reduced speed, it’s still a capable performer.

We wouldn’t call the SE-214-XT “good,” but we do think it’s one of the best in class at its seemingly impossibly cheap price, considering it also has to sit in freight to ship to wherever it’s sold. If you need a cooler to just get a system going, this is a good value. ID Cooling seems poised to challenge Thermalright in the future.

Best Thermals (Noise-Normalized) CPU Cooler: Liquid Freezer III, D15 G2

Liquid Freezer III original review | Newegg | Amazon

D15 G2 original review | Newegg | Amazon

Runner-Up: ID-Cooling Frozn A720

Amazon

This is for the Best Noise-Normalized Thermals, and because we have both liquid and air coolers in the charts for this round-up, we’re assigning the award to one liquid cooler and one air cooler.

The Arctic Liquid Freezer III firmly receives the Best Noise-Normalized thermals award both overall and for liquid, but air can’t be expected to compete at the same level of liquid and large radiators (especially when dropping fan RPM). For air coolers, the award goes to the NH-D15 G2 HBC for Intel (or LBC for AMD), with a runner-up award for ID-Cooling’s Frozn A720. We’ve listed the latter because it’s $100 cheaper than the NH-D15 G2.

This is a thermal category, so we’ll focus on charts.

• 

• 

• 

Starting with the Liquid Freezer III (read our review): Arctic’s revision of its Liquid Freezer II tried a number of new things, like shipping a mandatory contact frame with its Intel variation. This complicates matters and isn’t always for the best. 

We found that Arctic’s contact frame was worse than other options on the market, but its cooler design was such that other options couldn’t be used instead. It was still better than Intel’s ILM, though.

For thermals, the Liquid Freezer III is just as impressive as its predecessor, and its price is similarly competitive. In 200W testing on AMD with our older noise-normalized approach, the Liquid Freezer III was the clear chart topper when we limited the pump speed to 70% (which helped reduce noise and allowed higher fan RPM instead). Running the pump at 100% and sacrificing some of that noise budget for it, it was still tied for second place, behind only itself and matched with a cooler that’s now banned in the US. The next closest non-Arctic option is the Trinity Performance from Lian Li, which is specifically performance-focused in its design. 

In our Intel 250W heat load thermals on a 14900KF (read our review), also noise-normalized, the Liquid Freezer III with its mandatory frame climbs to the top of the chart. It’s ahead of the Frozen Prism by enough to not be margin of error or test variance. It’s also ahead of the Light Loop 360 and predictably ahead of all the air coolers.

The Liquid Freezer III remains a top recommendation of ours, though we do find its contact frame solution frustrating for Intel. On AMD, it’s much simpler.

• 

• 

Moving to the air coolers, the victor of the noise-normalized category on pure performance is the Noctua NH-D15 G2 HBC for our Intel 250W bench or LBC for AMD. Previously, we found that the G2 HBC outperformed even its closest competitor, the A720, by a couple degrees. This gap alone is impressive, as finding more than single-degree differences between air coolers is rare. Noctua’s work on the high base convexity cooler really worked for Intel -- but it’s also the recipient of our Best Mechanical Design category, so we’ll save that discussion.

On AMD 200W testing when noise-normalized, we also found Noctua’s D15 G2 LBC (or its flatter model) to be the current best noise-normalized air cooler result, behind only a huge stack of liquid coolers.

As some honorable mentions for runners-up here, since most are in the market for something cheaper, the ID Cooling Frozn A720 Black takes a clean second place in both our Intel and AMD CPU cooler testing. The Frozn A720 well balanced between the 2 platforms. It’s a relatively large tower that may have some clearance issues in some cases, but its $56 price-point makes it one of the more affordable, competitive coolers, and it’s $100 below the D15 G2.

Best Mechanical Design CPU Cooler: Noctua NH-D15 G2

Original review | Newegg | Amazon

• 

• 

• 

The next award is for Best Mechanical Design. This covers the total execution of every physical feature of the cooler: Pressure distribution, laser scans, ease-of-installation features, a highly usable design, aesthetics, and the cooling itself. This year, we can easily give it to the Noctua NH-D15 G2. Like last year’s winner for this category, which was the Assassin IV, Noctua’s NH-D15 G2 isn’t a good value cooler (it’s $150, which is insane for an air cooler) -- but this category isn’t for value. We can still appreciate the engineering.

It’s extremely well-built and takes careful consideration of factors often ignored by other cooling solutions. The NH-D15 G2 was secretly developing something similar to what we showed with some custom-made Scythe cooler coldplates that we had Scythe make early this year, which was trying to game the coldplate convexity to better match the CPU heat spreader’s surface concavity or flatness. 

Noctua was working on this for years and launched three models: The standard G2, HBC (or high-base convexity), and the LBC (low base convexity, sorry, Long Beach City) models subtly modify the convexity or flatness of the coldplate to better pair with deeply deformed Intel 13th and 14th Gen CPUs or with the relative flatness of AMD’s AM5 and AM4 CPUs with the LBC. Standard is meant to work on anything. We used our 3D laser scanner that we bought for cooler testing on the D15 G2 and discovered that the names really match the curvature.

We found that this wasn’t a gimmick and that there were actual, measurable and repeatable differences in benchmarks.

Noctua also lands on this list for its careful attention to detail on the fans, which it pairs and matches with slight RPM offsets intentionally in order to avoid a potential beat frequency phenomenon that could be annoying for some users. We have an interview with one of Noctua’s team members to talk about the engineering topics behind this. 

• 

• 

The D15 G2 also had excellent pressure distribution as a combination of its mounting hardware and the coldplate, shown in our mix of pressure maps across AM4 and Intel. 

An included washer mod added some further fine-tuning and small touches, though it was also clearly an attempt to try and bulk-on a value-add with the high price.

• 

• 

• 

Thermally, the D15 G2 didn’t blow away any of the other coolers by massive margins. It’s a good cooler, but spending $120 more than competition doesn’t mean it’s suddenly competing with a 360mm liquid cooler. The G2 with the HBC solution and a washer mod ran at 52.5 degrees delta T in our review results for the 250W Intel heat load, which was better than coolers like the Peerless Assassin by several degrees (and is impressive), but predictably behind a high-performance liquid cooler. The physics just won’t support beating water and the huge surface area of a radiator. Being realistic about performance expectations is healthy, though, and Noctua never claimed that.
We appreciate what Noctua has done with its mechanical and thermal engineering. The company may move slowly, but thus far, it has moved with purpose. As we said in our review, this is the type of thing you buy if you have your heart set on it and can afford it. You are buying Noctua’s name with the G2 and, likewise, its support. It has already delivered one free update for owners of G2s with a relatively minor rattle complaint.

---

Table of Contents

• AutoTOC

Intro

Today we’re reviewing the Light Base 600 by be quiet!. It’s a big deviation from the company’s monolithic towers with foam padding and is their first dual-chamber case. Be quiet! has gotten attention for designing this case with three presentations in mind: Standard, flat in traditional, old-school “desktop computer” design, and inverted. All of these have been done before, but be quiet!’s approach is simple and overall brilliant: To go flat, you remove the feet by turning them (like screws). To go inverted, you swap the feet to the other panel, which uses a slot-and-groove solution to notch them into place. This is fast enough that you could change it just because you’re bored of it or want to present the glass to a different side of the room.

Editor's note: This was originally published on November 13, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing, Writing

Patrick Lathan

Camera, Video Editing

Vitalii Makhnovets
Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

• 

• 

• 

• 

The Light Base 600 is a dual-chamber with four SKUs, ranging from $150 to $195 in price. Our variant is $195 because it’s white and comes with case fans. The Light Base 900 is a larger variation on this. We’re reviewing the 600 since testing for these reviews is a big commitment, but the general concepts apply to both, just scaled up on the 900.

Light Base 600 Specs

Motherboard compatibility	ATX, M-ATX, Mini-ITX
Case Type	Midi-Tower
PSU	PS2 ATX (not included)
Dimensions (L x W x H in mm)	450 x 305 x 435
Material	Steel (SGCC), ABS, Glass
Light Base 600 LX Weight (kg)	12.1 (net) / 14.6 (gross)
Light Base 600 DX Weight (kg)	11.5 (net) / 14.0 (gross)
I/O Panel	1x USB 3.2 Gen. 2 Type C, 2x USB 3.2 Type A, HD Audio (combined), ARGB controller, Reset button, Power button
PCI Slots	7
Drive Bays	up to 4x 2.5" (2 included)up to 2x 3.5" (1 included)
Light Base 600 LX Preinstalled fans (mm) / (rpm)	Side: 3x Light Wings LX Reversed 120mm PWM Rear: 1x Light Wings LX 120mm PWM
Light Base 600 DX Preinstalled fans (mm) / (rpm)	--
Maximum fans (mm)	Top: 2x 140 / 3x 120 Side: 3x 120 Bottom: 2x 140 / 3x 120 Rear: 1x 120
Optional radiators (mm)	Top: 120 / 140 / 240 / 280 / 360 Side: 120 / 240 Bottom: 120 / 140 / 240 / 280 / 360 Rear: 120
Maximum Dimensions (mm)	CPU cooler: up to 170 PSU: up to 200 GPU: 400
ARGB connector	3-pin
MSRP	Light Base 600 DX (Black): $150 Light Base 600 DX (White): $160 Light Base 600 LX (Black): $185 Light Base 600 LX (White): $195

Specs copied from manufacturer materials, please read review for our own measurements and opinions

Light Base 600 Alternatives

be quiet!’s key competitors for this include basically everyone, since dual-chamber cases have been so popular in recent years. If you’re thinking of buying this case, other cases you should know about include this quick list:

• Lian Li’s O11 series, including the Vision, Evo series (watch our review), and other variants, which jump-started this entire case archetype in the current generation
• Hyte Y60 & Y70, which give a new take on the concept with a cut-corner front that has been picked up by NZXT
• NZXT’s H5 (watch our review), H7 (watch our review), and H9 series of cases vary in their similarities, but share some concepts
• Montech’s King 95 and King 65 series, but especially the 95, for affordable dual-chamber options
• Antec’s C8 cases, which we’ve found to be relatively good thermal performers for the price
• Corsair’s 6500 Series cases, which we wouldn’t recommend overall (read our review)

And tons of others. We’ve left a lot out, but you get the idea: This is a crowded segment. A lot of these cases are in the $150 to $220 price range. A few are cheaper, like the empty Montech King 95 cases at $90 to $120 and the empty Antec C8, similarly priced.

It’ll be a big fight, but be quiet! does at least immediately differentiate itself with some unique features, one of which is their giant light bar that wraps the case. It was done this way so be quiet! could achieve the rotation without losing symmetry when you move the case around. In fact, all of this case had to be designed with symmetry in mind. That’s what gives the cheaper flexibility than more mechanically-intensive inversion processes from other cases -- but it’s also limiting in some areas.

Light Base 600 Basics

For the basics: Our version of this case includes 4x 120mm fans pre-installed as side intake and rear exhaust when it's in its traditional configuration. You could also shift these to the bottom as intake in a traditional orientation or, if rotated flat, they could be used as a side-to-side flow path. This on its own is pretty interesting and is one of the tests we ran. 

Other small fit-and-finish issues include a panel gap between the glass front and side, which is large enough that you’d need a taxi to get from one side to the other... The glass panels also diverge as they approach the bottom, with the bottom edge protruding slightly despite a flush top edge. We’ll come back to the fit-and-finish detail later though.

Getting into the big details:

The Build

We've criticized be quiet! in the past for its complicated case inversion processes. When we reviewed the Dark Base Pro 900, it took about 45 minutes to figure out disassembling and reassembling the case and sorting through the ridiculous variety and amount of screws and hardware. be quiet! must have taken that criticism to heart (even if 7 years later), because the Light Base 600 can be inverted in under ten seconds with no real screws at all (the design doc says 30 seconds, but that's with a system inside). 

The case feet can be pulled off by rotating them 90 degrees, then either placed on the top panel to invert the case or on the side panel to flip it on its side. It's not perfect: the feet don't lock and they go crooked as you slide the case around. They also feel fragile enough to damage by hand if positioned in the wrong spot. The ease of use outweighs the downsides, but the downsides were avoidable in design. Either way, once it’s done, these concerns go away.

• 

• 

This case also includes one of the simplest vertical GPU conversion systems we've seen. You take out all the horizontal slot covers, screw a bar across the back of the case, and adjust the built-in GPU support down a couple of notches.

The horizontal orientation is the most unique in the current era of computing, though it definitely used to be the standard. This old style is so liked in the retro scene that SilverStone made an April Fool’s case that accidentally became popular enough that there’s some real demand to make it.

The downside is that the Light Base 600 isn't a particularly small case, and because of its dual-chamber design, it's more than 32cm tall on its side. The 900 has even larger dimensions.

Backed against a wall, the case requires a minimum of 50cm up to the front edge of the case, requiring a deeper table to comfortably accommodate a keyboard and mouse if placed directly in front of the case. 

The dimensions make it impractical to use in the old IBM PC layout, but if you want to try it anyway, we recommend using a monitor arm rather than resting weight on the glass panel. We wouldn’t recommend setting it up with a monitor directly on it, especially since tempered glass can shatter without impact in some situations.

With the total monitor arm and monitor configuration, the setup could be made to look pretty unique -- we like the concept like this. Most of our desks are either 24” or 30” deep, with a few at 36” for filming. With that old-school setup with the computer right in front of the user, the 30” deep desks would only have 10” (or 25 cm) to spare for the keyboard and mouse. Plan your desk sizing appropriately. It’d obviously not be a problem if the system is off to the side of the keyboard and mouse. 

The desk in the image above is 30” deep and the lowest monitor can conflict with the top of the case, but this gives you an idea for sizing. You could definitely build a good-looking retro setup with the right color lighting and furniture, not that the 600 is confined to that style, but it would look good. You would just need a deep table. 

• 

• 

Fittingly for a case called the Light Base, there are dual LED and fan hubs in the back chamber. These have a total of 12x 4-pin fan and 12x 3-pin ARGB connections. The 600 LX's four stock fans occupy four of each type of plug (no daisy-chaining), but that still leaves eight unused pairs. Each hub has its own power and control inputs, allowing for two separate zones.

Our key focus is performance, but when LEDs are done, we still look for quality. We think the lighting is executed excellently; the diffuser bars work well and the LED animations are smooth. It looks good in a night-time or lights-out setup. 

Out of the box, all LEDs are set to be quiet! orange. The manual provides a full table of built-in lighting effects and instructions for syncing with external input. We thought the manual was well-written here. The cabling for the stock ARGB fans is a little overwhelming, but the light bars and their associated wires are tucked completely out of the way. 

Our one complaint is that the LED button and the reset button are exactly the same size and have no labels - though you’d probably only make that mistake once.

The Light Base 600 is another case without a support pillar between its two glass panels, giving us a top panel that could theoretically bend down. In practice, the top and bottom panels are interlocked with the front panel, keeping it solid as long as the case is assembled.

The steel side panel is fully ventilated, although the side intake vent and the PSU are the only areas where that's necessary. The side filter is removable for cleaning or additional airflow, which is an advantage over related designs like the HS 420 (read our review) and HYTE Y70 (read our review) where the side filters are glued in place. We’re happy to see that flexibility from be quiet!. Unfortunately, the top and bottom filters aren’t as flexible and are built into their respective panels: if you set up a horizontal build, some fans will exhaust through a filter that isn’t intended to be removed, which will create unnecessary impedance.

Primary radiator support in the case’s traditional orientation is at the top and bottom of the case, which both support up to 360mm sizes. 360mm radiators aren't officially supported in the side mount and most won't fit at all with the 39.6cm of clearance between the top and bottom of the case.

• 

• 

• 

The Light Base 600 has cutouts for back-connect boards, but normal ATX boards overlap all the cutouts along the bottom edge. Small fan and I/O cables can be snaked through the back-connect holes anyway, but we expect clear, premeditated cable routes. The storage behind the motherboard is deep, as is typical for dual-chamber cases, but it can be difficult to work in once that space is occupied by cables. The cable tie points are hard to reach once PSU cables are in the way, with the cable cover, HDD bracket, and fan and lighting hubs contributing to a cramped feeling. We ended up just wadding all the cables into the channel and snapping the cable cover shut on top. It’s still easier than the Lancool 207, at least (read our review).

We'd like to see a way to screw the cover down, but the cable channel is deep enough that it's unlikely to be necessary.

Drive storage is a complicating factor for cable management. The stock Light Base 600 only supports 2x 2.5" drives and 1x 3.5" drive, with the two smaller drives mounted to the cable cover, which requires leaving enough slack to open it. The stock bottom bracket can only mount a single 3.5" drive, but it has room for a be quiet! HDD Cage 2 (purchased separately) which fits either an additional 3.5" drive or two 2.5" drives. Using the bottom bracket at all reduces access to cable cutouts and one of the two fan and lighting hubs.

There's a gap between the two glass panes; it's a matter of personal taste as to whether that's a problem, but it's definitely intentional and known since the case shipped with a strip of cardboard in that gap. 

The gaps where light bars meet (top and bottom) definitely aren't intentional, though, and be quiet! didn’t do anything to minimize that.

This is the most brightly-lit, attention-grabbing area of the case, and it needs to look perfect. The gaps don't belong on a $195 case the way our case is set up, and the deviation at the top and bottom edges of the glass makes for a messy execution where other areas are done to a higher standard. The panel gap isn’t great for looks, but could contribute to dust ingress, especially in a horizontal orientation where it would be exposed directly up. We all know how dust collects on glass surfaces.

• 

• 

Some smaller points we want be quiet! to address in future iterations: First, the PSU fitment was unnecessarily tight, to the point that we had a harder time than necessary getting it installed. 

Second, even though the GPU support still works when the case is inverted, it may be weaker since it can't be flipped. 

Light Base 600 Thermal Benchmarks

Our model came with 3 reverse blade fans on the side and a traditional fan on the back.

We have a few key configurations of the Light Base 600 that we’re testing:

• Stock, which is standing in its traditional orientation with the included fans where they come
• Stock without a side filter
• Inverted, which is self-evident
• Flat with the fans left where they were, which becomes bottom intake. For the horizontal case testing, we removed the rubber bumpers in the side panel and we installed the feet in those slots. If we were testing the case without the feet there and just on the rubber bumpers then the bottom intake would be completely suffocated and far worse than what you’re going to see in our tests
• Flat with left intake and right exhaust

CPU Thermals: Full Speed (Light Base Only)

We’ll start with a chart of just the Light Base results.

With all case fans at full speed, the Light Base 600 LX averaged 48 degrees Celsius above ambient all-core and 52 on the P-Cores in its default configuration. Removing the optional side filter reduced the all-core temperature by 1.7 degrees, which is a large drop for a side filter only. It’s relatively restrictive.

Fully inverting the case didn't change temperatures enough to exit the margin of error, which is encouraging since, for the most part, it shouldn’t. The GPU would be most prone to change here.

Flipping the case on its side definitely hurt CPU thermal performance, with an all-core average of 51 and P-Core average of 55. Left in their stock positions, the three intake fans are obstructed by the desk surface when the case is put on its side. We intended to fix that with the left intake/right exhaust horizontal configuration, but with a focus on GPU thermals, so CPU thermals didn't change significantly versus the first horizontal result. You can see that here.

Flipping the case on its side with the intake fans against the desk surface also raised the noise level up to 32.1 dBA, versus 30.9 dBA in the stock or inverted configurations. This is actually noticeable. It’s also really cool: Previously, our less precise testing approach to acoustics would not have been able to surface this difference in numbers even though you’d be able to hear it. That’s because we were limited by the noise floor and measurement tools. Now, with our hemi-anechoic chamber that the audience has helped us build, we can actually detect with tools what the human ear can already hear. That’s why we have this chamber. People forget that ears are incredible at their job and that representing it with a microphone is very difficult, so when they ask why we’d build such a chamber, this is exactly the reason. We can get closer to measuring what you actually perceive.

CPU Thermals: Full Speed (Competitive)

Here’s the full chart.

It's fitting that be quiet!'s case is among the quietest on the chart when at 100% speed, bracketed by the HYTE Y60 at 30.1 dBA and Fractal Pop Air (watch our review) at 31.3 dBA, but there are no obvious noise-damping features other than the use of Light Wings fans.

The Y60 (watch our review) may be equally quiet, but it's also much hotter with just its stock fans, averaging 51 degrees all-core versus the 600 LX's 48, so that is an advantage for the LX. We know based on past experience that the Antec C8 is the dual-chamber case to beat in terms of overall thermals, 46 degrees in this particular test (albeit with a higher noise level of 37.1 dBA), while the (glass-fronted) King 95 Pro (read our coverage) averaged 43 degrees (at 36.8 dBA). We'll get to noise-normalized thermals in a moment.

GPU Thermals: Full Speed (Be Quiet!)

GPU full speed is up now: 

The Light Base 600 LX kept our GPU at 52 degrees Celsius above ambient with the memory temperature at 60 and hotspot at 68. Removing the side filter improved temperatures by 2.2 degrees for the GPU, which is a huge climb for just a filter. be quiet! has room to improve this area. 

Inversion had a mildly negative effect, raising the GPU average to 54 degrees. The horizontal configuration was worse still at 55 degrees for the GPU, 63 for the memory, and 71 for the hotspot. Keeping the case horizontal but shifting to the left intake/right exhaust configuration dramatically lowered temperatures, down to 41 GPU, 44 memory, and 53 hotspot. This is the equivalent of a bottom-intake configuration in a case that's oriented normally, but with unobstructed intake.

GPU Thermals: Full Speed (Competitive)

Here’s the competitive chart. All of the results with the fans in their stock locations are among the weakest on the chart, with the baseline out-of-the-box result falling behind even the Y60. The left intake/right exhaust configuration performed significantly better, tying the C8 ARGB's 41 degree GPU average and significantly outperforming the King 95 Pro's 46 degree average. As a reminder, this was done with just the included fans. All we did was move them. We didn’t want to add fans because it starts to become arbitrary and potentially unfair, but moving them really helped here. The horizontal configuration has the greatest potential for cooling, since a clear unidirectional airflow path can be created without obstruction. You’re not dealing with angled intake from the side, so there is a huge amount of potential here.

CPU Thermals: Noise-Normalized

The Light Base 600 LX's low noise at full speed means it has a shot in our noise normalized test, where all case fans are tuned down to hit our 27 dBA threshold. We normalize these in our hemi-anechoic chamber to get the granularity needed for a fair test.

50 degrees Celsius above ambient all-core and 54 P-Core ties the Antec C8 ARGB (read our review), but the King 95 Pro averaged 47 all-core, and cases with simple front intake like the Antec Flux Pro (read our review) and Lian Li Lancool 207 continue to dominate the top of the chart. 

GPU Thermals: Noise-Normalized

GPU thermal performance was already weak with the case fans at full speed, and with reduced fan speeds, the 600 LX is the hottest case on this chart other than the King 65 Pro. This is a strong argument for the horizontal left-to-right airflow configuration. These were all at the same noise levels. We’ve normalized them for this test. This is not a good result for be quiet!.

Standardized Fans: GPU Thermals

Because we were forced to mount the 140mm fans in the bottom of the case for our standardized fan test, we have an opportunity to see the case's bottom intake performance. 

An average GPU temperature of 44 degrees Celsius above ambient is more competitive than the other GPU thermal results we've seen from the 600 LX so far, close to the King 95 Pro and C8 ARGB's tied 43 degree averages, but that still only places the LX in the middle of the chart. This is also effectively a test of the Light Base 600 DX, since we would have performed all tests on that case using our standardized set of fans.

VRM Thermals: Noise-Normalized

Finishing off with VRM and system memory thermals during the noise normalized test, the 600 LX's performance is again unimpressive. 35 degrees above ambient for the VRM is warmer than the C8 ARGB's 33 degrees and the King 95 Pro's 31 degrees, and although the stack is ordered differently for memory temperatures, both the C8 ARGB and King 95 Pro were cooler than the 600 LX's 27 degrees. We also know from our full speed results that these sensors didn't respond strongly to the horizontal left intake/right exhaust reconfiguration.

Light Base 600 Conclusion

The Light Base 600 LX is on the expensive end of dual-chamber cases with included fans, especially when taking into account the thermal performance of the Antec C8 ARGB and King 95 Pro. Both of those cases also have fanless SKUs that are cheaper than the Light Base 600's fanless SKUs. 

The primary reason for buying the Light Base 600 over any of those other options (including the NZXT H9 Flow, which we haven't tested) is the wraparound light bar, so we're at least happy to see that the light bar looks good in person, with bright, evenly-diffused LEDs. It was critical that they executed on this feature well and they did do a good job on it. There were a couple caveats surrounding the fit and finish, like the panel flushness.

The horizontal option is also a unique selling point, but it's one that few people have the desk space to try. But this is a well-liked presentation layout, it ends up looking better than you might expect when looking at it vertically, and it’s a unique feature that is also executed overall well. It’s also mechanically simple. You’ll just have to plan your furniture more than typical, which isn’t so bad once you’re aware of the sizing.

If you do buy a Light Base 600 and you want to maximize thermal performance, we suggest using the horizontal configuration with side intake fans pushing directly into the GPU and exhaust fan(s) pulling hot air out of the opposite side. 

A few things to be careful of: If using a closed-loop liquid cooler, or “AIO,” with a pump in the block, you wouldn’t want to bottom mount it when the case is on its side. You should never bottom-mount a liquid cooler with the pump at the highest point in the loop, which that would do. Mounting the radiator to the side -- which was the “top” before going flat -- would be best. This will still put one side of the radiator tank at the highest point, which can create some bubbling noises in loops with less liquid or as they permeate over a 5-year period, but wouldn’t lead to catastrophic failure like a bottom-mount will.

All of this changes if you have the pump somewhere else in the loop, like the radiator.

It's good to see some movement from be quiet!, but although back-connect compatibility is fairly new, dual-chamber cases and ARGB LEDs are years-old trends. It'll take some time for the company to become a leader in cases again, rather than a follower. The company has also deviated from some of be quiet!’s expectations, which isn’t a bad thing. Fractal deviated as well and it worked out well for them. 

We like the case overall for its unique presentation. It is mechanically simple and it works well. Aside from that, there are no revolutionary changes to the build process or features. 

But once cases clear a bar for acceptable performance and for functionality, what matters most is that you like the way it looks. This case clears those bars. If you’re a big fan of the layout, we’re not seeing any major detractors to advise against a purchase.

---

Table of Contents

• AutoTOC

Intro

Intel’s Battlemage B580 GPU launches at $250. In our testing, the card has improved massively over the company’s original Alchemist GPU launch, so it’s already in a better spot than previously. By price, the B580 is positioned to compete with the NVIDIA RTX 4060 $300 GPU and the AMD RX 7600 at $250. Both NVIDIA and AMD have new GPUs launching in early January, but until they’re out, we can’t evaluate their performance.
What we can say is that, out of the gate, the B580 is far more reliable than the Alchemist cards were at launch. Intel has definitely improved. It still has issues and it was still shipping driver updates until the last minute of testing, but the company has objectively improved.

Editor's note: This was originally published on December 12, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Testing, Editing

Mike Gaglione

Chart QC

Jeremy Clayton

Camera

Tim Phetdara
Andrew Coleman

Writing, Web Editing

Jimmy Thang

---

Frametimes have largely been smoothed over, but remain a potential pain point for Intel in some games. Idle power consumption is also a pain point; however, overall value and throughput are competitive, and ray tracing performance is particularly competitive, especially against AMD.

We also found that the B580 scales extremely well with higher resolutions, seeing the most uplift over the 4060 or 7600 at 4K and 1440p, and often losing that ground at 1080p.

It’s an interesting review packed with a ton of new information, including a complete overhaul of our GPU test bench to an overclocked 9800X3D platform.

Let’s get started.

Intel B580 Overview, Pricing, & Specs

Here’s a quick refresher on the current market for GPU prices, but remember that the RTX 50 series is rumored to launch next month, followed by rumors of new AMD GPUs. 

GPU Price Comparison | GamersNexus | Early December, 2024

	Newegg Price	Amazon Price
Intel B580 (MSRP $250)	$250	NFS
Intel A770	$230	$280
Intel A750	$240	$200
Intel A580	$170	$170
NVIDIA RTX 4060 Ti 8GB	$400	$400
NVIDIA RTX 4060	$300	$295
NVIDIA RTX 3060	$280	$270
NVIDIA RTX 3050	$170	$170
AMD RX 6600	$190	$200
AMD RX 6600 XT	$240	$260
AMD RX 7600	$250	$252
AMD RX 7700 XT	$400	$420

For today: The B580 has an MSRP of $250 and was available for preorder on Newegg at this price. The prior Intel A770 is available for $230 to $280 for the 16GB model, with the A750 at $200 to $240 and A580 at $170. NVIDIA’s RTX 4060 is a main competitor, typically at $300, alongside the RX 7600 at price parity with the B580. These will be the two key comparisons.

One step up, the 4060 Ti 8GB and 7700 XT are both around $400 to $420 right now, making them primary upgrade options another class up.

Intel B580 Specs

We already covered the B580 specs in a video, but here are the basics again:

The card launches ahead of the B570 (watch our coverage), which launches next month. The B580 is a relatively small 20 Xe2 core configuration on the Battlemage architecture, making it smaller than the A750 or A770 in configuration size. Intel hasn’t yet announced 700-series cards in the Battlemage series.

The B580 has a 190W default TDP, runs on 1x 8-pin power connector, and is cut down to PCIe 4.0 x8, which is a potential downside for anyone on old platforms. We’d have to look at that separately in the future.

Test Bench Overhaul

We have a brand new test bench for all of this. We’ve moved from an overclocked 12700K to an AMD 9800X3D, which marks the first time we’ve used an AMD CPU as our permanent GPU test bench solution in the nearly 16 years we’ve been running now. The 9800X3D gives us far better headroom and scaling to see the impact of high-end GPUs that should be coming out soon. We have it overclocked to 5.4GHz all-core and under a 420mm Arctic Liquid Freezer III.

GPU Game Test Suite | 2025 Methodology | GamersNexus

Game	Resolution & Settings	Release Year
Black Myth: Wukong Benchmark	1080p/High 1440p/High 4K/High *Upscaling is not used. 100% for all tests. Game forces selection, selected FSR. Full RT disabled.	2024
(Ray Tracing) Black Myth: Wukong Benchmark	1080p/High Raster/Medium RT/FSR Quality 1440p/High Raster/Medium RT/FSR Quality 4K/High Raster/Medium RT/FSR Quality *FSR selected as upscaling method for all tests. Scaling set to "FSR Quality" equivalent. Full Ray Tracing set to Medium.	2024
F1 24	1080p/High 1440p/High 4K/High	2024
(Ray Tracing) F1 24	1080p/RT Ultra 1440p/RT Ultra 4K/RT Ultra *Based on Ultra High preset.	2024
Dragon's Dogma 2	1080p/Max 1440p/Max 4K/Max *All settings maxed except for RT (this is not an in-game preset).	2024
(Ray Tracing) Dragon's Dogma 2	1080p/Max RT 1440p/Max RT 4K/Max RT *All settings maxed (this is not an in-game preset).	2024
Final Fantasy XIV: Dawntrail Benchmark	1080p/Maximum 1440p/Maximum 4K/Maximum	2024
Resident Evil 4 (2023)	1080p/Prioritize Graphics 1440p/Prioritize Graphics 4K/Prioritize Graphics	2023
(Ray Tracing) Resident Evil 4 (2023)	1080p/FSR Quality/Max RT 1440p/FSR Quality/Max RT 4K/FSR Quality/Max RT *Based on "Maximum" preset.	2023
Baldur's Gate 3	1080p/DX11/Ultra-Custom 1440p/DX11/Ultra-Custom 4K/DX11/Ultra-Custom *Custom settings.	2023
Starfield	1080p/Ultra 1440p/Ultra 4K/Ultra	2023
Cyberpunk 2077: Phantom Liberty	1080p/Ultra 1440p/Ultra 4K/Ultra	2023
(Ray Tracing) Cyberpunk 2077: Phantom Liberty	1080p/RT Medium 1440p/RT Medium 4K/RT Medium 1080p/RT Ultra 1440p/RT Ultra 4K/RT Ultra *Based on Ray Tracing: Ultra and Ray Tracing: Medium presets.	2023
Dying Light 2 Stay Human	1080p/DX12/High-Custom 1440p/DX12/High-Custom 4K/DX12/High-Custom *Based on "High" preset but with DX12 and associated features.	2022
(Ray Tracing) Dying Light 2 Stay Human	1080p/DX12/FSR Quality/RT High Quality 1440p/DX12/FSR Quality/RT High Quality 4K/DX12/FSR Quality/RT High Quality *Based on "High Quality Raytracing" preset.	2022
Total War: Warhammer III	1080p/Ultra 1440p/Ultra 4K/Ultra	2022
NOTES	Dynamic resolution, upscaling, VSync, and VRS are disabled unless otherwise noted, even if they would normally be enabled by the in-game preset. RT features are disabled in tests not prefixed with (Ray Tracing). Games are run in exclusive fullscreen mode if available.

Our games list includes these. Not all of these will be shown in each review, but they are all tested. We’ve added Dragon’s Dogma 2 from 2024, Black Myth: Wukong from 2024, Baldur’s Gate 3 from 2023 -- which will be too easy for high-end GPUs, but is a good match for the B580, Starfield from 2023, Final Fantasy 14: Dawntrail from 2024, Cyberpunk: Phantom Liberty from 2023, and all the others you see above.

Cards Tested

Card	Time Tested
NVIDIA RTX 4090 Cybertank	December, 2024
AMD RX 7800 XT Ref	December, 2024
EVGA RTX 3060 XC Black 12GB	December, 2024
ASUS RTX 4060 Dual	December, 2024
Intel Arc B580 Ref	December, 2024
Sparkle Arc A580 Orc	December, 2024
EVGA RTX 3050 XC Black 8GB	December, 2024
XFX RX 6600 CORE	December, 2024
PowerColor RX 6600 XT Red Devil	December, 2024
XFX RX 6500 XT Black	December, 2024
Sparkle Arc A750 Titan	December, 2024
AMD RX 7600 Ref	December, 2024
Acer Arc A770 BiFrost 16GB	December, 2024
EVGA RTX 3060 Ti FTW3	December, 2024
XFX RX 6700 XT MERC Black	December, 2024
NVIDIA RTX 4060 Ti FE 8GB	December, 2024
EVGA RTX 2060 KO	December, 2024
EVGA GTX 1060 SSC 6GB	December, 2024
XFX RX 7700 XT Black	December, 2024
NVIDIA RTX 4070 FE	December, 2024
Sapphire RX 7900 XTX Nitro	December, 2024
EVGA GTX 1070 SC	December, 2024
Sparkle Arc A380 Elf	December, 2024
Colorful RTX 3070 BiliBili	December, 2024
EVGA GTX 1650 SC Ultra Gaming	December, 2024

We’ve tested about 25 GPUs for this launch. That list does not include the RTX 4080, 4080 Super, 4070 Ti or Super, 7900 XT or GRE, and similar high-end cards. You can find scaling data in our prior benchmarks for those.

We added the 7900 XTX and RTX 4090 for a ceiling. 

The reason for the cuts is time -- rolling into the end of year with team scheduling and our own travel, we wanted to allocate more testing time to older hardware or lower-end stuff to represent users considering an upgrade. That means we cut the high-end stuff for the GTX 1060 (watch our revisit), 1070 (watch our review), RTX 2060 (watch our review), GTX 1650 (watch our review), and similar cards. We’ll add higher-end GPUs in for the January launches, but they’re out of the price range of the B580.

As far as cards with multiple VRAM configurations, we ran the “default” or first-launched VRAM configuration wherever an option was present. 

Finally on the test bench side, we’ve added power efficiency testing and idle power consumption testing with calibrated tools.

We have a lot more to test, but this article is huge already and we have a busy time ahead of us with some travel.

Let’s get into the benchmarks.

B580 Gaming Benchmarks

FFXIV - 4K

Final Fantasy 14: Dawntrail is up now at 4K first.

At 4K, the Intel Arc B580 ran at 47 FPS AVG, with lows exceptionally close to the average. The B580 has a staggering lead of 28.6% over the RTX 4060, which is elevated as compared to its lead at 1440p and 1080p. The RTX 4060 (watch our review) is behind in lows also; although they remain consistent with its average.

The B580 also leads the RTX 4060 Ti at 15% ahead, the A770 by 16.5%, A750 by an impressive 27.9%. The RX 7600 is around the same price as the B580, but struggles at 4K in this test at 31.5 FPS AVG.

Users of the GTX 1060 6GB card would experience an uplift of about 193% moving to the B580, with the 1070 seeing 110% and the 3050 similar.

As we saw in our 4060 review, the 3060 Ti (watch our review) has a massive memory bandwidth advantage over its successors and embarrasses the 4060 and 4060 Ti cards here.

From what we’ve tested lately that’s better than the B580. That list includes the $450-$480 7800 XT at 22% better and the RTX 4070 non-Super at 27% better, among the flagships.

FFXIV - 1440p

At 1440p, the gap narrows in some places. The B580 is running at a completely playable 86 FPS AVG with good lows. Its lead over lower cards has decreased: 3.4% against the 4060 Ti (watch our review), 9% against the A770 (down from 16.5%), 19% over the A750 and RTX 4060 (from around 28% before), and 31% over the RX 7600 (watch our review).

The RX 7800 XT jumps to 40% from 22% and the 4070 jumps to 36% from 27%. The $400 RX 7700 XT (watch our review) leads at 98 FPS to 86 FPS AVG on the B580, but is more expensive.

The B580 looks good here despite a reduction in its lead from 4K.

FFXIV - 1080p

1080p squishes everything together and reduces the B580’s advantage.

The B580 at 124 FPS AVG is now just 10 FPS ahead of the A770 and RTX 4060, or about 10%. The B580 went from a 28.6% lead at 4K to 19% at 1440p to 10% over the RTX 4060. 

At 1440p, the RX 6700 XT (watch our review) was marginally below the B580 and is now ahead at 1080p. The B580 also had nearly a 15% lead over the RTX 4060 Ti at 4K, but now is behind. The 4060 Ti leads at 133 FPS, or an advantage of 7.6%. That’s a huge swing.

Starfield - 1440p

Starfield is up now, tested at 1440p.

The B580 isn’t as strong in this one as it was elsewhere. Starfield is also a game that was not functional on Arc at its launch, so running at all is already better than last time -- though that’s a low bar.

The B580 ends up behind the RX 7600, RTX 4060, and 4060 Ti in this benchmark. It’s beating the A770, A750, and A580 cards from Intel’s last gen.

The 4060 leads by about 17%, with the 4060 Ti in an insurmountable 42% lead. The 7600 is closer, at 5% ahead. 

The biggest downside here is the frametime pacing. Intel’s B580 struggles with 0.1% lows and 1% lows, which are indicative of a deeper per-frame problem that requires a frametime plot to investigate. We’ll explore that later in this review, closer to the conclusion.

Starfield - 1080p

At 1080p, the B580 ran at 51 FPS AVG, giving the RTX 4060 a 10 FPS AVG lead, but an even bigger lead in lows. Although the 4060 has an advantage of 21% in average framerate, the real difference is in frametime pacing that we’ll see closer at the end of this review.

The RX 7600’s lead is now 4%, with the 4060 Ti at 44%.

Intel’s B580 leads the A770 by an impressive generational 16%, despite a lower core count configuration. That’s a huge win for Intel over its own hardware. Its lead over the A750 is 27%. Despite these gains, the B580 is not that competitive in Starfield.

Resident Evil 4 - 4K

Resident Evil at 4K is pretty interesting. The B580 roughly matches the 4060 Ti, with a marginal deficit on 0.1% lows. The lead over the 4060 is massive, benefiting the B580 by 31% in average FPS. It also holds an advantage in lows. The lead over the 7600 is 24%, with the generational lead over the A580 and A750 alike impressive.

This is a strong showing for Intel’s B580.

Resident Evil 4 - 1440p

At 1440p, the B580 maintains a competitive position: The card is fully playable at 85 FPS AVG with overall OK lows, albeit lower proportionally versus the neighboring A770 and RTX 3060 Ti. With the resolution drop, the 4060 Ti now leads the B580 by 7% from its rough equivalence before. B580 benefits from its memory bandwidth at 4K.

The lead over the RTX 4060 is reduced to a still-noteworthy 22.5%, down from 31%. The 7600 lead is reduced from 24% at 4K to 18% at 1440p.

Resident Evil 4 - 1080p

1080p weakens the B580’s position overall. The 4060 Ti now moves to 10% ahead. The B580’s lead over the 4060 is reduced to 19%, although this is still a big lead. Most notably though, we see the frametimes deviate more from the average: The 128 FPS AVG is accompanied by 74 FPS and 63 FPS 1% and 0.1% lows with the B580, whereas proportionally and based on neighbors, these figures should be in the 100-110 territory. The drop jumps out as different from the rest.

Baldur’s Gate 3 - 4K

Baldur’s Gate 3 is up now, one of the most played games in the last year. Intel needs to do well in these types of critically acclaimed titles. Last round, we saw that Intel did better with Vulkan in Baldur’s Gate 3 than it did with Dx11, but was still disadvantaged in both APIs.

With Dx11 and 4K/Ultra, the B580 ran at 45 FPS AVG. The lows are reduced versus its immediate neighbors in the 4060 and 7600, both of which would be better experiences by frametimes alone, despite all looking approximately equal in average FPS. This is another for our frametime charts list at the end.

The B580 improves on the A750 by 6% and the A580 by 21%, falling behind the A770 in this one. The Arc cards are generally less competitive in frametime consistency in this game; however, they have significantly improved in their performance from a year ago.

Baldur’s Gate 3 - 1440p

Here’s 1440p.

The B580 ran at 74 FPS AVG, which is completely playable in this game. The lows are also acceptable in an objective sense -- it’s not a “bad” experience, but it’s also not as good as you’d get for frametime pacing from the RTX 3060 (watch our review), RTX 4060, or RX 6600 XT (watch our review).

The 6600 XT’s average framerate holds a 14% lead, with the 4060 at 8% ahead, breaking rank from a roughly tied average at 4K. The 7600 also climbs ahead, now matching the 6600 XT rather than the B580 AVG FPS as we saw at 4K. The 4060 Ti is 35% ahead of the B580 with its 100 FPS AVG.

Generationally, Intel improves over the Alchemist A750 by 7%, with a notable leap over the 62 FPS AVG of the A580.

Cyberpunk: Phantom Liberty - 1440p

In Cyberpunk: Phantom Liberty at 1440p, the B580 performs overall well. This is one of Intel’s strongest titles. 

At 1440p and with Ultra settings, without RT, the B580 lands at 53 FPS AVG in our in-game test within one of the cities. The B580 roughly matches the RTX 3070, although is lower in 0.1% lows and frametime consistency. The B580 leads the RTX 4060 in AVG FPS by 36%, the RX 7600 by a similarly large gulf, and the RTX 4060 Ti 8GB model by 12%. The 4060 Ti has stronger 0.1% lows, but not in a way which we think is worth the cost tradeoff.

Cyberpunk: Phantom Liberty - 1080p

At 1080p, the B580 slips in the ranks below the 4060 Ti, which also carries a large advantage in 0.1% lows. 

The B580 technically leads the RTX 4060 in AVG FPS by 18%, but the 4060 pulls ahead in overall frametime consistency. The RX 7600 is in a similar boat.

Black Myth: Wukong - 1080p

Black Myth: Wukong is next. This is one of the few games where we use the built-in benchmark. It’s new to our test suite and it’s been pretty consistent. Dropping settings would obviously increase framerate, but we’re interested in this from a scaling standpoint.

At High settings, broadly speaking, the B580 still has a slight deficit in lows, but a far less noticeable one than some other games. It’s close enough to the RX 7600 and RTX 3060 that the experience would feel overall similar, although technically worse by measurement.

The RTX 4060 holds a significant lead in this test, up at 19% and 54 FPS AVG to the 46 FPS on the B580. Frametime consistency is also better. The 3060 Ti leads the 4060 and B580, with the 4060 Ti at 65 FPS AVG, or 42% ahead of the B580.

The B580 at least improved on the last generation, at 29% over the A750, and 38% over the A580. Intel is making big generational strides against itself here.

Black Myth: Wukong - 1440p

At 1440p, we’re bordering on synthetic test territory for the B580. But what matters most to us is the scaling, not the raw FPS.

At 35 FPS AVG, the B580 allows the 4060 a lead of 6.6%, down from 19% at 1080p. This is a huge swing in favor of the B580, which continues to scale more favorably at higher resolutions.

The 7600 falls below the B580 in average FPS and is roughly the same in lows. The B580 also leads the prior generation flagship from Intel, the A770, which was at 30 FPS AVG. We’re going to pick up the speed here now that patterns are establishing.

Dragon’s Dogma 2 - 1440p

Dragon’s Dogma 2 is up next. For our CPU testing, we benchmark inside the city to create an NPC workload on the CPU. For GPU testing, we test outside of the city and in a GPU-heavy area with fields, water, and structures.

At 1440p, the B580 ran at 44 FPS AVG with good lows. The GPU does much better here than in some of the prior tests and ends up keeping pace with the RTX 4060. The 7600, 4060, and B580 are all functionally equivalent. The 3060 Ti (watch our review) benefits from its memory bandwidth and leads this grouping.

This is a much better positioning for the B580 than the prior games.

Dragon’s Dogma 2 - 1080p

At 1080p, things shake-up to knock the B580 loose from its rough equivalence with the RX 7600 and RTX 4060. Now, the RTX 4060 holds a 13% lead, with the RX 7600 in an 11% lead. The B580 retains overall typical frametime pacing, which is good for Arc. It manages to hang in there, but it’s seen some losses as the resolution comes down.

Dying Light 2 - 1440p

Dying Light 2 is next. This one looks good for Intel Arc, including frametime consistency: The B580 runs at 63 FPS AVG, which has it ahead of the 4060 Ti, 4060, and 7600. It’s also ahead of all of these devices in 1% and 0.1% lows, in what’s a major victory for Intel considering its challenges in some other games.

The B580 ends up between the RTX 3070 and RX 6700 XT. The $400 7700 XT holds a lead in AVG FPS of 12%, despite being 60% more expensive. This is one of the stronger showings for Intel.

Dying Light 2 - 1080p

At 1080p, the B580 continues its good performance by outmatching the A770 and RTX 4060; however, it loses ground to the 4060 Ti that it had outperformed at 1440p. The 87 FPS AVG is well into playable territory and carries good lows. The die size is much smaller than an A770, so outperforming it is key here: Intel may be able to reduce some of its cost per card, which will help it compete long-term.

Intel B580 Ray Tracing Benchmarks

We’re moving on to ray tracing benchmarks. These are not comparable to the rasterized charts.

Ray Tracing: Resident Evil 4 - 4K

Resident Evil 4 is up first. This is a relatively light workload for ray tracing and isn’t as distorted for one vendor or another as some other games.

This first chart is at 4K with Max RT, but with FSR set to quality for all devices. We have validated other upscaling solutions and found their performance to be roughly identical, so matching for just FSR agnostically allows us to control for image quality as well. Because we are using FSR, it is not native resolution. We don’t use FSR in our rasterized tests.

Here’s the 4K chart. The B580 ends up directly competing with the RTX 4060 Ti in the average framerate and in frametime consistency, with lows overall indicative of a comparable frame-to-frame interval to the 4060 Ti. It’s not far off from the RTX 3070. Intel’s B580 leads the RTX 4060 by 21%, at 50 FPS to 41 FPS AVG. The improvement over the A750 is similar. We need to run the 7600 back through this one still, but the 7700 XT and 6600 XT give an idea for AMD’s positioning.

Ray Tracing: Resident Evil 4 - 1440p

In this test, the Intel B580 runs at 75 FPS AVG with overall good lows. The 1% and 0.1% metrics are similar to what we see on the 3060 Ti, which outperforms the B580. Intel’s B580 manages to outperform the RTX 4060 by 6%, including proportional frametimes. It also outperforms the A770’s 71 FPS AVG, which was already relatively good for RT. AMD’s RX 7600 ran at 65 FPS AVG, giving the B580 nearly a 16% advantage while costing about the same.

The 4060 Ti’s 91 FPS result outdoes the B580’s 75 FPS by 21%, including scaling in the lows. The 7700 XT pushes further ahead, up to 106 FPS AVG. Both devices cost about 60% more than the B580, at around $400 for the 4060 Ti 8GB and 7700 XT alike.

Ray Tracing: Resident Evil 4 - 1080p

At 1080p and with the same FSR and RT settings, Resident Evil 4 positions the B580 now below the RTX 4060 -- though they’re functionally tied. It’s losing the advantage that we’ve now proven it typically has at higher resolutions. These two and the RX 7600 are closest in price, with the 7600 down at 85 FPS AVG and allowing the B580 a slight lead of 9%.

The 4060 Ti leads the B580 by 29%, with the 7700 XT 40% ahead of Battlemage.

Overall, Intel’s performance in this lighter RT test is competitive, especially for the price and in comparison to AMD. 

Ray Tracing: Dying Light 2 - 1080p

Dying Light 2 is a heavier RT workload and is tested with FSR Quality. 

At 1080p, the B580 ran at 65 FPS AVG and held consistent frametimes. It’s a little better than the A770 and similar in average to the 3060 Ti. The 4060 Ti leads the B580 with its 73 FPS AVG result, a gain of 12%. The B580 leads the 4060 by 13%. AMD falls behind in this test, with the 6700 XT now below the B580 and RTX 4060. The B580 ends up with nearly a 30% lead over the 6700 XT, which is a precarious spot for AMD. The lead over the RX 7600’s 45 FPS AVG is 44%.

Ray Tracing: Dragon’s Dogma 2 - 1080p

In Dragon’s Dogma 2 with Max RT and no upscaling, the B580 ran at 44 FPS AVG with lows slightly lower than where they should be. The A770 leads the B580 in lows and ties in the average. 

AMD’s RX 7600 does better here than in Cyberpunk and Black Myth, up at 53 FPS AVG and outranking the B580 by 20%. The RTX 4060 also outdoes the B580 (and also outperforms the 7600’s frametime pacing), landing at 54 FPS AVG.

Battlemage is doing OK, but does not have the same advantage it has elsewhere.

Ray Tracing: Dragon’s Dogma 2 - 1440p

At 1440p and RT with Dragon’s Dogma 2, the B580 ran at 36 FPS AVG and had dips in lows compared to the neighboring prior-generation A-series GPUs. The 7600 leads the card now by about 10%, reducing its lead from at 1080p. The RTX 4060 is in a similar position, but with better lows.

Ray Tracing: Cyberpunk - 1080p Medium

Cyberpunk with Medium RT settings is less brutal on AMD than Ultra, something we showed in our last round of reviews, but remains one of the heaviest RT loads and disproportionately hard on AMD devices. This is tested without any upscaling.

The B580 did overall well here: Its average framerate was competitive with the RTX 3060 Ti, despite lower 0.1% lows than it. The 4060 struggles in this one with frametime pacing and is erratic and unpredictable with these settings. The B580 is a much better experience, with the 3060 Ti and up better still. The 4060 Ti also struggles with frametime consistency, illustrated by the 0.1% lows that draw our attention to the problem. This appears to be an issue with memory bandwidth, especially as contrasted to the 3060 Ti, which likely compounds with the VRAM capacity.

In a massive loss for AMD, the B580 runs 62% ahead of its RX 7600 in average framerate while also maintaining superior lows and frametime pacing. AMD’s closest card is the RX 7700 XT, which is significantly more expensive. This is a repeat of what we saw with the Alchemist cards last time, where the A770 and A750 alike outperformed AMD’s 6700 XT and 7600.

We tried testing RT Ultra for Cyberpunk, but unfortunately, the system hard locked with the B580 card. This behavior did not occur on any of the other 18 plus devices we tested with these settings and appears to be related to Battlemage.

Ray Tracing: Black Myth Wukong - 1080p (Experimental)

Black Myth: Wukong is up next. 

This is an experimental chart, which means we are still researching the performance and vetting it. Our confidence is lower in data for experimental charts, which means they should be weighted less; however, we publish them when we feel we are ready to begin sharing, as this allows us to continue learning and advancing. We just like to be clear in our disclosures of data confidence.

This one is brutal on anything that isn’t NVIDIA. Intel also had a last-minute driver update for RT performance in this game.

At 1080p and with FSR set to quality, using high raster settings and medium RT settings, the B580 ran at 34 FPS AVG, with lows where they should be. Intel’s last-minute driver change did nothing to our results -- they were nearly identical, so we’re just showing one set here. The B580 ranks alongside the RX 7800 XT (watch our review), as AMD has serious difficulties with this game. Wukong is similar in performance to Cyberpunk in its performance characteristics on AMD, where even a 4060 outperforms a 7900 XTX.

For Intel, it at least is doing better comparatively than AMD, but NVIDIA has a stranglehold here.

Frametime Consistency Inspection

We’ll move to some quick zoomed-in shots of the frametime charts for some good and bad scenarios for the B580.

Frametime Consistency: Baldur’s Gate 3 (4K)

We’ll start with Baldur’s Gate 3 at 4K. This chart shows the frame-to-frame interval on the vertical axis and the frame on the horizontal axis.

Lower is better, more consistent is best. Although the B580 is at times lower than the RTX 4060, the card experiences occasional but massive excursions from the previous frame. We’re seeing some jumps upwards of 20ms, which is noticeable, but not necessarily game-breaking -- it depends on the frequency of them. In this case, you’d notice. Arc has improved substantially from Alchemist in this specific regard, but it’s clear that Intel still has some work to do in at least Dx11 with this game.

Frametime Consistency: Dying Light 2 (1440p)

Dying Light 2 rasterized is one of the better scenarios for Arc, so we’ll balance by also looking at that as well.

In this one, the B580 had better frametimes across the entire bench pass. It’s lower overall, but also doesn’t experience any major unexpected deviations. Unlike the Baldur’s Gate chart, the B580 GPU is exceptionally consistent in its frametime pacing here. The 4060 ends up the worse of the two tested devices.

Frametime Consistency: Phantom Liberty RT (1080p)

In this next one, NVIDIA and its RTX 4060 are a victim of its own creation. In Cyberpunk with RT Medium, the RTX 4060’s framebuffer and memory bandwidth cause it to struggle, resulting in frametime spikes that, in one case in this chart, exceed 120ms. That’s 1/10th a second that you’re looking at the same frame. This massive spike would result in a stutter in-game. These stutters were repeatable during play. There’s another large spike later in the run. They also happen frequently in our later runs. 

The B580 did not exhibit this behavior in this test and would be a better experience. Intel’s memory choices benefit the company here; bandwidth is one of them but capacity also contributes.

Frametime Consistency: Starfield

Starfield had some major consistency issues run-to-run for Intel so we’re actually only showing Intel here.

This shows 2 test passes from Starfield at 1080p for the B580. In this one, you’re seeing wildly inconsistent frame pacing with a total range from 13ms to 53ms. There’s no pattern to it and frame delivery is overall erratic, making the experience less consistent and allowing more microstutter behaviors to emerge. Dying Light 2 looked much better for Battlemage.

Frametime Consistency: FFXIV 1440p

Final Fantasy 14 is next. In this one, B580 did well overall but had slightly reduced 0.1% lows versus the 3060 Ti and 6700 XT; however, it did better than the RTX 4060 across the board. In the frametime chart for Final Fantasy, we see an incredibly flat line with only one major spike that would hardly be noticeable since it isn’t repeating.

Let’s move on to efficiency.

Intel B580 Power Consumption & Efficiency (Experimental)

Now, we’re getting into power consumption and power efficiency measurements. For this testing, we’re using a PMD2 hooked into a PCIe slot interposer and the PCIe cables. This PMD2 has been personally calibrated by Elmor, from ElmorLabs, for our work upon request and has been validated against other measurement devices, including current clamps. We are isolating for just GPU power, eliminating other test bench power draw. This allows us a clean and isolated feed of GPU behavior.

We haven’t yet tested our full lineup of GPUs as this is all brand new testing, so we only have some of the most relevant cards. You’ll have to check back as we add more.

Power Consumption: Idle

This chart is for idle power draw on Windows desktop and without any GPU tasks active.

The Intel B580 pulled about 35W when idle at desktop, which is a lot less than the Alchemist GPUs pulled (up at 46W, 45W, and 43W), but it’s still more than everything else. Intel still has an idle power consumption challenge to overcome.

The RX 6600 (watch our review), for example, pulled only 5W through the PCIe slot and PCIe cables when idle, which is impressive. The 6600 XT was similar, at 6.2W. The 4060 is a relevant comparison and far lower in power consumption idle, down at 11W to the B580’s 35W or so.

This is an area Intel will eventually need to improve, but no one will care about its idle power draw if the performance isn’t good enough to warrant a purchase first, so we’ll see if this improves subsequently.

Idle GPU power has been requested by our audience for years, so we’re happy to finally start delivering it with this first look.

Efficiency: Baldur’s Gate 3 (1440p)

The next chart is for Baldur’s Gate 3 efficiency at 1440p, looking at rasterization performance. This is measured in FPS/W, which can be thought of as frames per joule since it’s just frames per second per joules per second. Higher is more efficient here.

FPS/W is easier to understand for most people. In this test, the RTX 4060 Ti has the best FPS/W performance, up at 0.71. The 4060 Ti is about 37% more efficient, meaning its FPS/W score is 37% higher than the Intel B580. This is still a massive generational improvement for Intel, where its A580 previously was way down at 0.31. The B580 scores 68% higher in FPS/W than the A580 and A750. Its total power draw is about 141W here, the same as the 4060 Ti, and so the reason its FPS/W is lower is because its framerate is simply lower. 

The 4060 pulls 126W in this test and has a slightly higher framerate, benefiting on both sides of the equation. We don’t yet have the RX 7600 in this lineup.

Efficiency: FFXIV (1440p)

Now we’ll look at one of the B580’s best case scenarios to balance the last two.

This chart is for Final Fantasy 14 at 1440p, where it did well. The B580 ends up 4th on our list so far at 0.51 FPS/W, ahead of the 7800 XT and behind the 4060.

The RTX 4060 pulled 124W here, putting it at 0.58 FPS/W and almost 14% more efficient than the B580. That’s a good spot for the B580, despite being less efficient and pulling more power at 170W.

The 4060 Ti pulled just 129W, so it didn’t leverage its total power budget, and was 27% more efficient than the B580.

Against the last generation, the B580 is significantly more efficient than the RX 6600 XT, RX 6600, RTX 3060 Ti, and Intel’s own Arc cards. The jump over the A750 is huge, especially since the Titan model has a boosted power budget.

Efficiency: FFXIV (1080p)

At 1080p, the 4090 and 4060 Ti appear the same because the 4090 (watch our review) is completely CPU-bound. It can’t perform any better, but remains in a higher power draw state.

The B580 again ranks as less efficient than the 4060 and more efficient than the 7800 XT and below. The B580 pulled 161W here.

Efficiency: Black Myth Wukong (1080p)

In Black Myth: Wukong at 1080p and without ray tracing, the B580 scored 0.3 FPS/W and pulled 146W, which has it significantly behind the RTX 4060 at 0.4 FPS/W and 121W. The 4090 would rank higher if it could fully stretch its legs.

The B580 is about equal to the 7800 XT again.

Cyberpunk: Phantom Liberty (RT, 1080p)

In Phantom Liberty with Ray Tracing and at 1080p, the B580 manages to hold onto its 4th place positioning in our current limited lineup. The lead over the 7800 XT is now a massive 29%, thanks to AMD’s performance issues in this title. The 4060 maintains an advantage of 15% over the B580. What’s clear is that Intel has drastically improved on its A-series performance.

Intel B580 Conclusion

Intel Battlemage is way better than our experiences with Alchemist. The launch of the A380 and subsequent A750 and A770 cards was plagued with experience-ruining bugs that were unfixable in many cases. This time around, we ran into a couple driver issues, but the only one that completely prevented play was an issue with Cyberpunk and RT Ultra settings, where it’d hard crash. Everything else ran more or less without issue.

This review is long already, so we’ll keep it simple. Here are the key points:

• We observed that the B580 scales well as resolution increases, especially at 4K and the games that aren’t heavy enough to eliminate it as an option. Unfortunately, the card isn’t so powerful that 4K is playable in a lot of these games, but the behavior was consistent with 1440p as well. 1080p saw a loss of advantage or sometimes a flip with the RTX 4060 or RX 7600.
• Intel’s frametimes are overall much better and mostly avoid experience-ruining issues in almost all cases. The Battlemage drivers still have some frametime pacing problems in some games and tests and are not overall superior, but have improved enough that, in combination with a low enough price point at $250 and below, they could be overlooked depending on the games you play.
• Also as a downside, Intel’s idle power consumption remains high. It has improved massively, but is higher than everyone else at idle for the most part.
• Intel’s efficiency in actual gaming scenarios, at least when those games perform well for it, has improved massively and is competitive with the RX 7800 XT and RTX 4060 in some scenarios.

As the A750 went on massive fire sales previously, we’d sometimes recommend it with heavy caveating. The main one was that we never felt comfortable recommending it for the general mainstream audience as we feared users without troubleshooting experience (or even those who have it, but don’t have secondary GPUs) would find it frustrating when encountering games that simply don’t work. Starfield, for example, just didn’t work at launch on Arc. 

Battlemage has improved on this front. We feel much more comfortable recommending at large. However, we still have one major caveat, which is that this is still only the second true generation of Intel’s modern dGPUs. We fully anticipate that Battlemage will have issues given the thousands of games and millions of hardware and software combinations, and we can’t possibly vet them all. That’s why it’s important to check other reviews. We can only speak for our experiences: In the games we tested and with our software, we encountered only one game-breaking bug, and that was RT Ultra in Cyberpunk, which is likely not particularly playable anyway. It is a valid issue, though. We also had one system hard reset at some point.

Now, on NVIDIA and AMD, such issues would almost never happen at this point. On Intel, this is an improvement from dozens of problems with the Alchemist launch. The issues were so many that we published a bug report detailing over 30 major problems. 

The B580 poses considerable value as compared to the RTX 4060 and RX 7600. It is not always the best and trades with these cards, but is a serious competitor. You have all the numbers. We’d recommend looking through our results and determining if it’s a good fit for your games, but overall, Intel’s improvement is obvious.

The next big thing will be seeing what happens in January with Intel’s competitors.

---

Table of Contents

• AutoTOC

Intro

Today we’re reviewing the new $100 Montech King 65 Pro. The coolest feature of the case is its side-mount fan tray, which can be released with two screws and pivoted out for easy cable routing or fan and radiator installation. It’s well-designed so that the tray is loosely secured by the cable grommets, which use their chamfered edges to guide it into place in a way that almost feels like there’s a guide wheel.

This dual-chamber case comes with 3 pre-installed fans: 2x 140mm reverse blade ARGB side fans and 1x 120mm ARGB rear fan. Looking around the case, we can find remnants of the King 95. That’s because a lot of this case is the same, but cheaper.

Editor's note: This was originally published on October 14, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Mike Gaglione

Camera, Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara
Andrew Coleman

Writing, Web Editing

Jimmy Thang

---

• 

• 

• 

But for those of you who have used the King 95, a lot of this case will be familiar: It’s intended to be a cheaper variation of the theme. There are some paneling changes, like the top panel, but a lot of the interior is familiar. 

Because we never formally reviewed the King 95 as it was during our case benchmarking rework, we’ll treat this as the first look at both the 95 and 65 features and review it fully. Since there’s shared tooling, some of the discussion will apply to both.

Let’s get into the review.

Pricing & Alternatives

We’ll start with the competitors in this price range.

Currently, the King 95 Pro is $140 to $150, including its 6x ARGB fans. The non-Pro variant is the same case, but drops to 0 included fans and is currently around $110. At $100, the King 65 Pro swaps some panels but adds 3 fans back. 

Somewhat confusingly, we found Montech’s own King 95 on Newegg for $90 at the time of writing this. They’re calling this a “holiday sale” -- it’s October 10th as we write this and we have absolutely no idea what holiday that is for, maybe “Newegg Needs Money Day,” but the point is that $90 makes it cheaper than the brand new King 65 Pro and may lead to some product cannibalism. Adding 3 fans wouldn’t be that expensive.

In price and ignoring form factor, the King 65 Pro competes with the Lian Li Lancool 207, the $80 case we just reviewed. The Lancool case is a more traditional design, while this is a dual-chamber fishtank, so they have different looks. Lian Li’s Lancool is an excellent thermal performer with some cable management challenges.

The SilverStone 514X is another case around $100 that the King 65 competes with on price, but is totally different on style.

The Phanteks XT Pro Ultra is $80 and comes with 4x 140mm fans in a standard form factor case as well.

As for actual style competition, the Lian Li Vision is currently $130 to $140 and doesn’t include fans, but would match for the general design.

Montech King 95 & King 65 Differences

The King 65 Pro adds a bolted-on front plastic piece to square out the front/bottom of the case. This is in contrast to the curved design of the King 95. 

Aside from the glass, the top panel also features curved edges as compared to the King 65’s straighter-edge design. The King 65’s top and back side panel also features a more standard mesh ventilation as compared to the King 95’s Hyte-inspired sweeping design. The King 95’s ventilated side panel also features a built-in dust filter, whereas the King 95’s solution does not. 

• 

• 

There are also differences between the fan hubs of both cases. The King 65 features 6 total connections whereas the King 95 Pro has 10 total ports. 

Another difference includes the fact that there are rubber grommets within the King 95 Pro, which are absent from the King 65 Pro. 

Montech King 65 Pro Build Quality & Panels

Those are the key differences. Now we’ll get into build quality and panels of the King 65. A lot of this discussion will also apply to the King 95.

Overall, panel construction quality is good on the King 65, but there are some caveats.

• 

• 

At Computex, we noticed that the front panel was very slightly misaligned with the decorative plate for the I/O side. This is somewhat of a nitpick, but that’s because it’s rarely an issue. 

The cases that we got resolve this issue and have the two front panels level, so that was nice to see. 

The front decorative strip is actually just plastic with a brushed look to sort of fake an aluminum appearance, serving as a cheaper gimmick to emulate the O11D and 6500D (read our review) look.

As a downside, the top metal strip for the supporting backing of the glass front panel was bent up on our black King 65 Pro. We always include damage in our reviews, whether that’s quality control or shipping. This particular damage looks like something that happened at the factory, not in shipping. It could be easily pressed back down with a flat edge.

As a positive, the panel gap between the front and side panel is flush, which wasn’t the case with the show model we saw 5 months ago. Cheaper cases often have problems with panel alignment.

• 

• 

The side panels are secured with thumbscrews, which we always like to see just as a security feature when moving the computer around for cleaning and maintenance. The front panel pops out easily once the side panel is removed.

Like we saw on the Lancool 207, the King 65 uses a reinforced rectangular piece of steel to support the snaps to the frame.

The front corner of the top panel is also relatively strong, unlike the Tryx case, because Montech designed it properly. The Tryx case top panel was caving in not from shipping damage, but from design issues, and we want to make sure that plot doesn’t get twisted because it’s important. The King 65 is relatively stout, which helps reduce the top panel length to provide strength, and is also supported by a more significant anchor to the deeper secondary channel, meaning a lengthier attachment point to serve as a counter support on that front edge.

The top panel itself is of sturdy build and has butted-together steel around the edges to give clearance from the fan rails while adding structural rigidity, but lacks the mesh we saw in the King 95. 

The final side panel is a little floppy from lack of thicker supports, but has no functional downside. It’s also ventilated for air intake through the side. 

• 

• 

Removing that panel, there’s a cable management cover tray with 2x 2.5” SSD mounts in it. Two screws can be removed to hinge the door outward, giving access to the drive mounts and two additional fan mounts with rails in the door. And this is where we get into an area that Montech could improve.

• 

• 

Both sets of fan mounts have heavy obstructions. By accommodating both 120mm and 140mm sizes, Montech needs two sets of mounting points. Unfortunately, in the process of creating these, it has covered up to about 20mm per side of the fan blade, or 40mm total horizontally obstructed. This is a huge amount of intake area loss. Montech could still support two sets of rails, but at least cutting a hole right in the middle would be better -- or even running the rails the entire length of the fan so that there’s some room for air. This is an oversight that likewise extends to the King 95. The King 95 Pro did OK in most of our tests, but falls behind other cases with similar fan count. This is a large contributor to that.

Montech King 65 Pro Cable Management

• 

• 

For cable management, the King 65 continues the trend of including velcro cable ties that secure behind the motherboard tray, which itself is already deep enough to provide functionally limitless cable space. Although we’ve since built in the cases, we liked how the fans were pre-routed and arranged out of the box.

• 

• 

The power supply sits elevated atop a rubber-damped pad, providing clearance on all sides while hopefully absorbing vibration that could cause noise. Although the exterior of the hard drive cage has rubber dampers, the drive sleds themselves do not. If supporting 3.5” drives, we’d like to see rubber dampers at the contact points where the drive meets the plastic or metal. 

Access is through the rear for drive installation, which mostly makes things easier. They aren’t hot-swap cages with pre-wired backplanes or anything fancy like that, but the rear access does make things easy to get to.

As for radiator, GPU, and fan support, it’s the same as the King 95 Pro: Montech’s GPU support is up to 420mm long, so basically any video card, and radiator support is up to 280mm or 360mm top, 140 or 240mm side, and 175mm tall air coolers.

Montech King 65 Pro Thermal Benchmarks

CPU Thermals: Noise-Normalized

We’ll start with noise-normalized CPU thermals under a full torture workload.

Unfortunately for the King 65 Pro, it’s the new worst result on the charts. These fans just aren’t powerful enough to provide the cooling necessary, and likewise, extra noise created by paneling choices, such as the sharper angles and the obstructions, leads to a need to reduce speed to normalize for noise.

The King 65 Pro lands at 57.6 degrees over ambient for average P-core temperature, putting it behind even the Hyte Y60 (watch our review) in the baseline setup.

The Antec C8 runs about 3 degrees cooler, which is significant, although it carries a $50 premium. The same-priced Silverstone 514X (read our review) runs about 6 degrees cooler with its own stock fans. The Flux Pro (read our review) does well here, but costs $80 more than the King 65 Pro, so it’s not a fair comparison. We know, that’s hard to believe since they’re both pros.

The Lancool 207 is $20 cheaper than the King 65 Pro though and manages to completely dominate it on this chart. At 44.5 degrees, it’s not even close. The King 65 Pro is just not a great thermal performer.

The King 95 Pro with the glass front ran at 50 degrees over ambient, which is over a 7 degree reduction without even using mesh. That’s a lot of impact from the extra fans and panel changes.

GPU Thermals: Noise-Normalized

Here are the GPU thermals for the same noise-normalized test.

The King 65 Pro gets destroyed here. It is not only at the absolute bottom of the chart, but it’s several degrees worse than the previous worst: The FARA 515XR set the floor at 54 degrees over ambient previously, with the King 65 Pro at 57 degrees over ambient. For GPU testing, that’s a big swing. The GPU is not normally as sensitive to the case changes as the CPU in our tests.

The $150 Antec C8 (read our review) is an impressive 15 degrees better than the King 65 Pro. The $80 Lancool 207 is also about 15-16 degrees better than the $100 King 65 Pro.

Standardized Fans: CPU Thermals

The next test is where we pull all the included fans and swap them for 3 standardized fans. Some cases are worsened by this, as we’re removing more abundant or better fans; however, this is useful for evaluating changes to the chassis structure itself. 

As a perfect example: The King 95 and King 65 are suddenly exact equals when we swap to the same fans and positions. They’re within 0.4 degrees of each other for P-core average performance, which is impressive. That’s a testament to our new methodology and is great to see.

The Antec C8 with comparable side intake lands at 46 degrees, outperforming the King 65 and King 95 alike. Because we’ve normalized to the same fans and in the same positions, we can confidently state that the King 65 and King 95 have paneling that obstructs the fans enough to worsen performance. Montech could boost itself by changing its porosity and all the small obstructions to airflow that add up, like those earlier rails we discussed.

Standardized Fans: GPU Thermals

GPU thermals when using standardized fans are also within roughly 1-degree ranges between the King 95 and King 65. The curvature of the glass theoretically should help improve air access to the GPU, but this gets into areas where we’d need CFD to know for certain. Regardless, at about 1-degree difference, they’re functionally the same.

The Antec C8 with side intake again climbs ahead, up at 42 degrees for GPU temperature against 44.6 on the King 65.

CPU Thermals: Full Speed

Back to the included stock fans and at 100% fan speeds, the King 65 Pro landed at 37.8 dBA as tested in our hemi-anechoic chamber, aligning it most comparably with the Antec C8 ARGB, which was 37.11dBA. That’s the same noise level.

The C8 outperforms the King 65 even with both at full fan speed, at 49 degrees to 54 for the average P-core temperature. The cheaper FARA 514X is quieter than both, at an impressive 35 dBA and 47 degrees over ambient. The Lancool 207 sounds more like a vacuum cleaner at 41.6 dBA when running at full speeds, but pushes to 41.6 degrees over ambient.

VRM Thermals: Noise-Normalized

Noise-normalized VRM and RAM thermals position the King 65 Pro at 36 degrees for the VRM and 26 for the RAM. These results have it toward the bottom of the chart, though better than the Hyte Y60 with its stock configuration. Both the VRM and RAM thermals are completely acceptable, it’s just that it isn’t as competitive.

Montech King 65 Pro Conclusion

Thermally, the Montech King 65 Pro is just not impressive. It can’t overcome the limitation it has on its panel and case design. Its 3 moderate fans are just not sufficient for it to get anywhere impressive on our thermal charts and it’s at the bottom of some of those charts.

In terms of build quality, in some places, it’s pretty good. The actual core of the chassis is good overall. There’s a lot of space to work with the cabling. There’s also some cool features like the side-mount fan tray, which is also on the King 95 Pro. Overall, the build and panel quality are fine. It’s just that at $100, it has to be really competitive and one of the things it's competing against is the King 95 Pro without any included fans. 

This makes it hard to recommend the case in any one category. Overall, the King 65 Pro is okay. We don’t hate it, but against standard form factor cases like the Lancool 207 or even the FARA 514X, which we weren’t thrilled about but we thought was fine, those are better options at this price. They are also better thermal performers, though the King 65 Pro is certainly a lot easier to build in than the 207. 

In terms of other competition, the Antec C8 is similar in design to the King 65 Pro, but it’s wider and its non ARGB variant without fans is similarly priced.

Within $20 plus or minus of the King 65 Pro, the case kind of gets its ass kicked in either direction thermally.

---

Table of Contents

• AutoTOC

Intro

We’re looking at the best cases available right now. 

We broke back into case reviews this year and have added to our case collection. This round-up highlights the best cases we think are available in 2024 based on our thermal benchmarks, acoustic tests, build quality evaluation, and cable management testing.

This year was packed with excellent cases -- but of course, a few that were really problematic as well.

Editor's note: This was originally published on November 21, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Host, Writing

Steve Burke

Writing

Patrick Lathan

Video Editing

Vitalii Makhnovets

Writing, Web Editing

Jimmy Thang

---

Setting Expectations

Welcome back to the Best Of Round-Up series. These are our favorite pieces to run annually and help you quickly get back into PC components if you haven’t been paying attention, but also give us a reason to go back through all the cases we’ve tested recently and remember what was the best. 

These are meant to be more of a fly-over story in the round-up series to help simplify and highlight the key points. If you want the full in-depth reviews with all the nuance, check the links in the description below. We’ve linked each case review and have also provided all the usual affiliate links with them.

We already published Best CPUs and the same thing applies here: Sometimes, the best case or CPU in 2024 might be something that launched in a previous year. We look at the total market and what the best options are now, not just what launched this year. As a reminder as well, we took about a year off of case reviews in 2023, so we launched back into it heavy this year and we’re excited to be back to the Best Of series for something positive.

Let’s get into it.

Overview: Best PC Cases for 2024

Category	Case	Review
Best Overall Case 2024	Fractal North on Amazon on Newegg Fractal North XL on Amazon on Newegg	Excellent Design: Fractal North Case Review & Benchmarks Fractal North XL Case Review & Benchmarks: Wood Panels & Mesh
Best Mechanical Design	HAVN HS 420 on Amazon	A New Type of Computer Case | HAVN HS 420 Thermal Benchmarks & Review
Best Sub-$100 Case	Lian Li Lancool 207 on Amazon on Newegg	Lian Li Lancool 207 Airflow Case Review | Cable Management, Build Quality, & Benchmarks
Best Mid-Range Case	Antec C8 ARGB on Amazon on Newegg	Surprisingly Good: Antec C8 ARGB Case Review, Thermals, Cable Management, & Noise
Best Out-of-the-Box Case Thermals	Lian Li Lancool 207 on Amazon on Newegg Antec Flux Pro on Amazon on Newegg	Lian Li Lancool 207 Airflow Case Review | Cable Management, Build Quality, & Benchmarks Best Case of 2024 So Far: Antec Flux Pro Review & Benchmarks
Best Noise-Normalized Case Thermals	Antec Flux Pro on Amazon on Newegg	Best Case of 2024 So Far: Antec Flux Pro Review & Benchmarks
Most Innovative Case	Meshless AIO	A Better Computer Fan - Sometimes: Cross-Flow Meshless AIO Case Benchmarks & Review
Biggest Disappointment Case	Corsair 6500 Series on Amazon on Newegg	Corsair Forgot How to Make a Case: 6500D Airflow & 6500X Review

Best Overall Case 2024: Fractal North & North XL

Fractal North: Original review | Newegg | Amazon

Fractal North XL: Original review | Newegg | Amazon

Awarding the best overall case is difficult this year, which is a good problem to have -- that means cases don’t suck. This category requires a strong offering on all fronts: Value, thermal performance, acoustic performance, aesthetics, build quality, ease-of-installation features, and cable management.

• 

• 

• 

• 

We’re giving it to the Fractal North & North XL. Although other cases are superior in some individual situations, like the Antec Flux Pro (by a slim margin) in thermals or the HAVN HS 420 in mechanical aspects, the North and North XL are masterful executions of a standard computer case.

This is bolstered by prices that have been lower in recent weeks, typically down around $125 to $140 for the North (watch our review).

The Fractal North was the first case to successfully kick-off the wooden trend at scale, with everyone else in the industry scrambling to follow. InWin gets credit for initially attempting this years ago, but it failed to find a way to bring it to market and execute. Fractal pulled it all together and cleaned up the idea.

• 

• 

The North comes with either a glass or mesh side panel, which includes options for additional side mounts to further boost cooling performance. Actually, in our recent tour of Wendell’s offices, you can see such a configuration for one of his machines.

Fractal’s fan support on the XL is also effectively whatever you want it to be: The company cleverly uses standard hole spacing for its rear ventilation (and even PCIe slot covers and top of the power supply shroud) so that you could mount a fan basically anywhere you want in the case. The cooling options are really opened up because of this. 

Ease-of-installation options are also great: Because the top panel so easily slides off to remove, and because Fractal doesn’t put a ton of other metal in the way, you can get clear access through the top of the case for builds. The extra width in the XL also makes larger radiators easier to work with up top, reducing board clearance issues.

• 

• 

The North XL (read our review) is what we prefer to the North, mostly for its increased size for easier building and for our style of build, but both are good takes on the theme. Visually, the wood look doesn’t look like the slapped-on Vinyl style we see in other cases. It looks more purposeful. Fractal also does great work mixing different tone wood coloring with different panel coloring, including matching the black case with gold accents and the white case with silver accents.

And finally, even in cooling performance, the North XL is routinely at the top of the charts for thermal benchmarks. It’s not the best -- that’ll go to two other cases this year -- but it’s regularly at the top, and that plus all the other features, plus the value, get the North cases the Best Overall award.

Best Mechanical Design: HAVN HS 420

Original review | Amazon

Runner-Up: SilverStone Alta D1

Original preview | Amazon

The next award is for the Best Mechanical Design, which we give to cases that may not have won in other categories, but truly innovated for the physical and mechanical elements of the case.

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

This year, it goes to newcomer HAVN and their HS 420. The HS 420 (read our review) exhibits a mastery of mechanical features.

The HS 420 doesn't include fans, it's not fully airflow-focused, and it isn't cheap at $200 for the horizontal GPU model. That makes it a tough fit for our price- and performance-focused award categories, but it needs to be recognized, because it's one of the most interesting and downright good cases that we've reviewed this year.

Mechanically, it’s excellent. The curved tempered glass panel is the most significant feature. The panel is installed by sliding it back and down into the case on plastic wheels, where it then comes to rest on the bottom edge of the chassis. It takes some getting used to, but it's a clever and secure way of getting the huge glass pane seated. It's also proven to be durable, surviving multiple build/unbuild cycles through the filming process without serious wear.

There was a lot of thought put into airflow, something we illustrated in animations. The HS 420 VGPU includes a curved glass deflector that's intended to direct incoming air around and into vertically-mounted GPUs. It was a clever idea and a fresh approach, which we appreciated, but it did not help performance in our test system. That’s also why we're specifically giving the award to the $200 base model, which is cheaper and sheds the hardware that didn’t do much. 

Even with the base model, though, HAVN planned for an unusual airflow pattern with intake at the bottom of the rear panel and exhaust at the top of the same panel, and included 120mm plates on each mount that are isolated with rubber vibration damping. In fact, every fan tray is padded with rubber, including the side mount that doubles as drive storage. The drive storage alone is pretty cool. It creates a spine-like structure that can be removed for more fans.

The case's construction is clever and precise in areas like the tightly-spaced storage bays at the front and the rounded edges of the PSU shroud and motherboard tray. The simple GPU support that ships with the non-VGPU case variant is one of the sturdiest we've seen, cast from solid metal and braced against the PSU shroud. The cable channels are not only labeled, but color-coded in what is an excellent feature, as it helps new builders understand where to route cables.

There are a ton of other small attention-to-detail elements, like the fact that HAVN matched the structural reinforcement for the panels to align with fan hubs, or the dead zone, rather than obstructing air intake. You can check our review for all the other details, including our full charts with 7 or 8 different fan configurations.

• 

• 

• 

• 

• 

• 

• 

• 

Before closing out this category, we want to give some recognition to SilverStone's ALTA D1. The ALTA D1 is an $800 halo product that's finally hit retail following our first encounter with it at Computex 2023 and again at Computex 2024. The price is ridiculous and very few people will buy the case, but the mechanical design hopefully inspires some more affordable cases: It has modular bays on rails everywhere, including 5.25" drive support, socketable radiators, a slide-out motherboard tray, SSI-EEB support, 11 expansion slots, large fan mounts, repositionable power supplies, and a true home server setup. It's beyond what we could recommend in a serious review if only for the price, but we may come up with a fun project for the ALTA D1 in the future and we respect its mechanical design.

Best Sub-$100 Case: Lian Li Lancool 207

Original review | Newegg | Amazon

We’ve had a category for years for the “Best Budget Case,” but the amount of quality cases in the $60 price point and under has really declined in recent years. Honestly, at or below $60 right now, we think used cases are typically a far better deal than something new at the same price.

• 

• 

• 

• 

• 

• 

• 

So this category has become Best Sub-$100 case, which we’re giving to the Lian Li Lancool 207. 

There were a lot of cases to consider for this category: When we were deciding this year's winners, although there are a lot of sub-$100 cases, there just aren’t any that defeat the Lian Li Lancool 207 (read our review). In the current market, $80 is dirt cheap for a name-brand case: there are cheaper cases available from companies like SAMA and DIYPC, but we're confident that none of them come close to the Lancool 207's frequently chart-topping thermal performance.

When we reviewed the SilverStone 515XR, we were aware that SilverStone had pulled out all the stops to offer a case with four fans at $68, but the performance gap between it and the 207 is massive. There’s also differences in quality-of-life features. 

The same goes for the Phanteks XT Pro Ultra and the Montech XR. Both cases have been added to our charts recently. The Montech XR is supposed to be closer to $60, but has been out of stock lately and often more expensive when it is available.

The Lancool 207 is a solidly-built case with sturdy toolless panels, which is unusual at this price point. The most unusual aspect, though, is at the bottom of the case, where two 120mm intake fans are positioned directly underneath the GPU, while the PSU is mounted at the front with an extension cord. The shroud is ventilated on both sides as well as the rear in order to provide an airflow path for the bottom intake fans (although adding drives and cables to the case reduces that ventilation, but the ample intake through the rear counters this). In addition to the two bottom intake fans, 2x 140mm ARGB front intake fans are included with the case. 

They may not sound significant individually, but the fact that the 207 has a full suite of basic quality of life features like reusable (and bridgeless) expansion slot covers, rubber grommets for cable cutouts, a USB Type-C port, ARGB fans, built-in velcro straps, and toolless panels is impressive in combination with its price and thermal performance. It's rare to see that full combo in a case, especially one which is also a top thermal performer. The fact that the 207 also wins our objective, numbers-based Best Thermals category is also part of why it lands here.

We have two big issues with the 207: first, the toolless side panel won't stay on without meticulous cable management. There's no way around this, but sticking to flat cables and minimizing the number of cables used will help. There's also room to improve PSU cooling in the case, as most PSUs will be oriented fan-side-down with barely any clearance above the table surface. The cable management is the one to be aware of. You can watch our review for more on that.

Still, at $80, Lian Li is being brutally competitive.

Best Mid-Range Case: Antec C8 ARGB

Original review | Newegg | Amazon

• 

• 

• 

• 

• 

• 

Next up is the award for the Best Mid-Range case. We’re looking for something that isn’t too wild or out there. We want relatively standard, but a good price-to-performance and overall build quality. We’re also looking for something a little bit better than the budget case. For this, we’re giving it to the Antec C8 ARGB.

Antec has had a strong year. It had 2 showings this year in our list we have today, but it was the Antec C8 that first signaled to us that the company was worth paying attention to again after a decade of fading into irrelevance. 

MSRP is $150 for the ARGB variant with fans included, but it's been long enough since the case's June launch that we're beginning to see discounts. The fan-less variant is regularly available with promo codes around $100. 

This mid-range award category gives us some room to consider quality of life and aesthetic features, not just raw performance-per-dollar, and the C8 is well-balanced in those categories.

The C8's dual chamber layout isn’t new, but Antec impressed us with the level of care that it put into details like molding text into the fan frames, color-matching all the components in the white SKU, and the accessory kit. Small touches, even ones that aren't functional, add up to create a product that feels worth the price. The C8's appearance reminds us of the clean designs that made Corsair's 4000D Airflow (and 5000D, and 7000D, et cetera) so appealing, but we much prefer Antec's case to the directly comparable Corsair 6500D Airflow.

Functionally, the usual benefits of dual-chamber design apply: there's plenty of space to work with in both chambers, including storage space for cables and drives. Dual chamber cases frequently have cable channels so deep that it's difficult to reach tie points at the bottom and effectively bundle cables, but the C8's removable "air duct partition" solves that problem.

The centerpiece of the C8 ARGB (read our review) is the bottom fan mount, containing two unusual 160mm x 35mm reverse-blade intake fans (with an additional conventional 140mm exhaust fan at the rear of the case). These two 160mm fans are placed over a completely open vent, meaning that the removable bottom filter is the only obstruction to incoming air. Antec was aware enough to not obstruct this intake path, unlike many other companies. 

• 

• 

This makes GPU cooling in the C8 ARGB excellent, although CPU cooling with our test setup was predictably lackluster due to the lack of front or side intake.

We prefer the C8 ARGB to the fanless base model, but the less expensive SKU may make more sense for liquid-cooled builds since the ARGB's 160mm stock fans won't match any normal radiator. The case can fit up to 3x 360mm radiators simultaneously, most easily in the side and bottom slots. For air cooled builds, the $20 upcharge for the ARGB variant with the extra fans is worth it. Many of the dual-chamber cases we've reviewed have followed in the footsteps of the O11 Dynamic by not including any stock fans, so having a choice is a welcome change.

Best Out-of-the-Box Case Thermals (Lian Li Lancool 207, Antec Flux Pro)

Lian Li Lancool 207 original review | Newegg | Amazon

Antec Flux Pro original review | Newegg | Amazon

Our next two awards are for thermals. The next one is for Best Noise-Normalized Thermals, while this one is for Best Out-of-the-Box Thermals. The difference is that noise-normalized is focused on one category, whereas Best Thermals looks for the best performer in all tests we ran and at all fan speeds.

The raw all-fans-maxed stock performance is a tie between the Lian Li Lancool 207, represented in the sub $100 category, and Antec Flux Pro, which is represented in the next category. These cases have similar layouts, each with front intake fans behind a mesh panel and bottom intake fans mounted on top of a ventilated PSU shroud. Our test bench hardware uses air cooling with a flow-through GPU cooler and a CPU tower cooler, which we've chosen to be as broadly representative as possible and as controllable as possible. This hardware benefits from simple, direct airflow: cool air from the front for the CPU, cool air from the bottom for the GPU. These two cases are the ideal realization of that airflow pattern.

The Flux Pro and Lancool 207 both averaged 38 degrees Celsius above ambient all-core and 42 degrees on just the P-cores. Fractal's Torrent and North XL are both within one degree of this result, but the Flux Pro (read our review) and 207 pull slightly ahead in other categories.

Moving to GPU thermals, both the Flux Pro and the Lancool 207 averaged exactly 38.2 degrees above ambient, although the 207's memory temperature averaged slightly better at 41 degrees versus 43, while the Flux Pro's hotspot temperature was a little lower. The Flux Pro's stock configuration is set up to mount the PSU rotated 90 degrees in Antec's "iShift" mount, but we found that mounting the PSU in a more typical orientation with the plug at the rear of the case slightly lowered overall GPU temperatures for a best-case 37 degree delta above ambient.

The 207, on the other hand, logged our best temperatures so far from the VRM and RAM sensors at 21 and 17 degrees above ambient respectively. These two cases are extremely close in performance, and we can't declare just one winner between the two of them.

Best Noise-Normalized Case Thermals: Antec Flux Pro

Original review | Newegg | Amazon

The next award is for Best Noise-Normalized Thermals. This is a pure performance-driven category, but it combines 2 metrics, which is acoustics and thermals. Fortunately, its winner is also just a good standard case to build in: The Antec Flux Pro.

• 

• 

• 

• 

Noise-normalized thermal tests are the most important performance benchmarks we run on cases. With the current iteration of our bench, we place the case on a table in our hemi-anechoic chamber with a mic pointed directly at the center of the front panel from one meter away. Then, we lower the speed of the case's stock fans in tandem until we hit our chosen threshold of 27 dBA SPL. This allows us to level the playing field so that the cases with the loudest fans don't automatically land at the top of every single chart by brute force. It creates a control.

The Flux Pro's front panel doesn't block much noise, but it's so well-ventilated that the three 140mm front intake fans don't need to work hard and don’t generate much noise to begin with. There isn’t much resistance and so they can spin slower, while the 2x 120mm shroud-top fans assist GPU cooling. The Flux Pro's PSU shroud is open on all sides, including the bottom, which maximizes airflow without directing significantly more noise towards our mic.

CPU thermals are tied at the top of the chart with the Lancool 207 at 41 degrees Celsius above ambient all-core and 45 degrees on just the P-cores. Some of our other high-performing cases like the North XL and Torrent come close, and the Lian Li Lancool 216 (watch our review) is tied in this one, but the Flux Pro is the best overall when taking all results into account. 

In the noise-normalized GPU thermal test, it outperforms the Lancool 207 and ties the Torrent (watch our review) for GPU die temperature. 

In VRM and DDR5 memory thermals, the Flux Pro is the chart leader by a technicality, but again is functionally tied with the 207. Both the Flux Pro and 207 offer excellent cooling at the noise levels, but the Flux Pro technically takes more lead spots than the 207. But the 207 is well-represented already.

We also know from our full speed results that it's possible that the Flux Pro could perform even better with its PSU mounted in the typical orientation rather than the stock rotated iShift mount.

The Flux Pro was an excellent return to form for Antec and offers a standard layout case for strong fundamentals. It is deserving of the Noise-Normalized victory while still offering the basics and foundation required for a good standard PC build in a non dual chamber case.

Most Innovative Case: Meshless AIO

Original review 

Honorable Mention: InWin Dubili

Newegg | Amazon

• 

• 

• 

• 

• 

• 

• 

• 

• 

Our Most Innovative Case award goes to the case that has the newest and most different design, even if it doesn't work perfectly, because innovation pushes this industry forward. This goes to the Meshless AIO Mini-ITX case.

Despite its generic, boxy exterior, the Meshless AIO (read our review) is one of the most innovative cases we've ever reviewed.

The Meshless AIO (now apparently listed with a formal launch name as the “MD280”) is built around a cylindrical crossflow fan, like a smaller version of the fan in a mini-split wall unit. We found the case design curious enough that we built this 3D animation to help explain and educate on how these work. The basics are that the crossflow fan shoots hot air out at the upper edge of the side panel, and theoretically pulls cool air in everywhere else: towards the GPU fans from the side panel, through the radiator, and over the motherboard and PSU from a tiny strip of ventilation along the bottom edge of the case. In practice, we found that the GPU was better cooled by taping a couple of normal 120mm case fans to the top of the case, but these had other downsides in other tests. 

• 

• 

Thermal performance was at least adequate, and considering there was a single fan handling everything, overall impressive for what was done on a single fan that you might not normally find in a PC case. We were concerned about the PSU's lack of direct access to cool air, but otherwise, everything was getting fed air.

There are some other creative aspects to the MD280’s construction outside of the fan: the main portion of the case is a single piece of extruded aluminum, and other elements like the fan, radiator, glass side panel, and handle slide into the open ends. This is an excellent and compact design that works well. Another cool feature is the lever that’s accessible from the rear of the case, working to slide the PCIe riser cable up and down. It presses the slot into place to improve accessibility. It's an ambitious and creative project, especially since Meshless Design informed us that it's an independent outfit driven by one man: Hank Lin. This is the work of an engineer who turned an idea into an actual product with overall good execution.

We need to issue a buyer beware warning here: it looks like the Meshless AIO's price has been increased to at least $450, a big jump from what we were expecting, and we have no way of vouching for the company's ability to deliver on pre-orders. Unfortunately, that price kills the case for us, as there are simply too many other good competitors in ITX at half the price. 

But it still gets an innovation award.

• 

• 

• 

We have an Honorable Mention for this category as well: InWin’s Dubili ATX case. Dubili is an anagram of iBuild, an abbreviation for InWin's DIY "iBuild iShare" brand, as well as an abbreviation of "Do Believe." The instruction manual comes in the form of a dedicated single-purpose app with animated 3D renders, which is absolutely required if you order the flat-packed build-it-yourself DIY edition. We like the Dubili for its unique look, high-quality build and materials, and the modularity of the paneling. That it flat packs is also a cool aspect to the case, creating a fuller DIY experience. The color scheme is also different: A sharp silver-and-orange or a Noctua-style cream-and-brown with gold screws. The case doesn’t win over the Meshless AIO, but we like it enough that it deserves a mention.

Biggest Disappointment Case: Corsair 6500 Series

Original review | Newegg | Amazon

The Biggest Disappointment is up next. There are a lot of options to consider for this category: Between the GameMax HYPE (read our review), the Tryx LUCA L70, the Corsair 6500, and more. We had our work cut out for us.

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

The award goes to Corsair for the 6500 series of cases. Although Tryx’s case was a mess, it ultimately doesn’t have the same expectations attached to it as Corsair. To be a disappointment requires some level of expectation.

Corsair wins the award handily: Broadly speaking, Corsair has continued to disappoint ever since it took on huge investment. Like the assets of its corporate overlords, its disappointments are diversified: They bought Origin, which sold us a $6,600 computer that had the CPU underclocked by 1GHz out of the box; they made the A500 air cooler, which had heatpipes so unlevel they sheared our pressure paper. But the company also shipped some excellent products in recent years, and those came from its case division.

The Corsair 4000D (watch our review) was commonly $80 and one of the best sub-$100 cases available for a while. The 5000D (watch our review) was an OK follow-up to that, scaling-up the useful features to a larger size. Corsair was finally doing well with cases again, following a long drought of basically nothing.

But then it launched the 6500 (read our review) series this year, attempting to jump onto a hype train that had already left the station, and doing so with substandard build quality. The Corsair 6500D and 6500X were $200 without fans, both featuring superbly bendy panels without any reinforcement -- which we found totally unacceptable for a $200 so-called “Premium” case. There were lots of plastics, misaligned panels, and build quality issues. 

The 6500D had a number of slapped-on features that didn’t work together. They tried to accommodate back-connect motherboards, but doing so required punching what seems like a completely random hole into a drive cage, which was then entirely left out of the original manual. If you used one of Corsair’s own SHIFT power supplies, you’d have to remove the drive cages completely and move to only NVMe drives. We found build quality issues, design oversight, and a high price to be not only disqualifying from any considerations on our lists, but also deserving of the Biggest Disappointment after falling from the grace of the 4000D.

Corsair is a big company and has the resources to fix this, but it’s a matter of whether they choose to.

Since our critical review, Corsair has dropped the price of the 6500D closer to $150. It’s still worse than an Antec C8 ARGB, which often costs the same or less.

---

Table of Contents

• AutoTOC

Intro

We’re looking at the best CPUs of 2024, following what are probably the last CPU launches for the year. 

We’ve benchmarked most of the major CPUs out right now and have a huge lineup for you today, but there’s a remaining, weird factor from last year: And that’s the fact that prior generation parts are sometimes the best options. It wasn’t always the case that 3 generations of parts existed new-in-box from both vendors. They’d typically cleared some of those out by now, but with AMD and Intel alike, there are actually good, older parts available in some categories. They’re everywhere.

Editor's note: This was originally published on November 16, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Video Editing

Vitalii Makhnovets

Writing, Web Editing

Jimmy Thang

---

We’re going over the Best Overall, Best Gaming -- which is one of the easiest categories to judge, Most Balanced for a mix of workstation and gaming performance with price in consideration, best upgrade, most efficient, and more.

Setting Expectations

Welcome back to the annual Best Of round-up series. We run these at the end of every year to get people back up to speed efficiently. A lot of you only check-in on PC part performance every few years for a new build, so this article will serve as a quick-start point for you to figure out which CPUs to pay the most attention to in your research. It also helps us get back up to speed on our own test data. As you find CPUs you’re interested in from this article, you can check out the in-depth reviews linked below for each of those components. That’ll get you everything you need to make a decision.

As usual, these round-ups don’t go as charts-heavy as our usual reviews. We’ll scatter them to support the points, but the goal is to provide a quick-and-easy recap. 

Let’s get into it.

Overview: Best CPUs for 2024

Category	CPU	Review
Best Overall CPU 2024	AMD Ryzen 7 9800X3D on Newegg	RIP Intel: AMD Ryzen 7 9800X3D CPU Review & Benchmarks vs. 7800X3D, 285K, 14900K, & More
Most Balanced CPU 2024	R9 7950Xon Amazon on Newegg Runner up: i9-14900K on Newegg on Amazon	95°C is Now Normal: AMD Ryzen 9 7950X CPU Review & Benchmarks Intel's 300W Core i9-14900K: CPU Review, Benchmarks, Gaming, & Power
Best Gaming CPU 2024	AMD Ryzen 7 9800X3D on Newegg	RIP Intel: AMD Ryzen 7 9800X3D CPU Review & Benchmarks vs. 7800X3D, 285K, 14900K, & More
Best Upgrade Value CPU 2024	AMD R7 5700X3D on Newegg on Amazon Runner-Up: Intel i7-14700K on Newegg on Amazon	New AMD Ryzen 7 5700X3D CPU Review & Benchmarks vs. 5800X3D & More Intel is Desperate: i7-14700K CPU Review, Benchmarks, Gaming, & Power
Most Efficient CPU 2024	AMD R5 7600X3D	AMD's Silent Launch: Ryzen 5 7600X3D CPU Review & Benchmarks vs. 7800X3D, 5700X3D, 9800X3D
Best Mid-Range GPU	i7-12700KF on Newegg on Amazon	Intel Core i7-12700K & 12700KF CPU Review, Benchmarks, & Efficiency vs. AMD Ryzen
Best High-End Desktop CPU 2024	Threadripper 7970X on Amazon	Crazy Efficient: AMD Threadripper 7980X & 7970X CPU Review & Benchmarks
Best Gaming CPU Under $100 in 2024	R5 5600on Newegg on Amazon i3-13100F on Newegg on Amazon i3-12100F on Newegg on Amazon	AMD Ryzen 5 5600 vs. i5-12400 & 5600X CPU Review & Benchmarks Budget CPU Battle Royale: Intel i3-13100F CPU Review & Benchmarks Budget King: $130 Intel Core i3-12100F CPU Review & Benchmarks
Biggest Disappointment CPUs 2024	Intel Instability	Intel's Biggest Failure in Years: Confirmed Oxidation & Excessive Voltage

Best Overall CPU: AMD Ryzen 7 9800X3D

Original review | Newegg

The first award category is for the Best Overall CPU for 2024. This one goes to the AMD Ryzen 7 9800X3D CPU, which also takes our Best Gaming CPU category later. We’ll save the gaming discussion for this CPU for that category in a little bit.

The 9800X3D (read our review) gets Best Overall for a few other key reasons: First, its $480 price makes it a better buy than several of the other high-end CPUs that have come out recently, such as the Intel 285K at $630 launch pricing. 

Secondly, the 9800X3D manages chart-topping gaming performance -- which we’ll save for discussion in our Best Gaming category -- while also managing overall acceptable production and workstation application performance. It’s not the best in these workloads, but it’s still capable. 

In Blender, the 9800X3D doesn’t benefit from its extra cache, but still manages to at least outperform the lower power budget Zen 5 components that are 8 cores and fewer, like the 9700X. 

In Photoshop, the 9800X3D ended up as our best performer. Photoshop likes Zen 5 in general and benefits from its architectural changes, and that really shows on the 9800X3D.

The 9800X3D is also an excellent overclocker and is fully unlocked, something that wasn’t true for the prior X3D parts. It’s such a good OC part that we even took it to stream to use liquid nitrogen to overclock it at -140 degrees Celsius, yielding some ridiculous performance scaling in real games that we might not have expected.

AMD moved the cache die closer to the substrate, with the core complex getting the cores right against the inside of the integrated heat spreader for this generation. This has helped improve the thermal situation, which is what gives AMD more clock headroom. AMD designed Zen 5 ground-up for X3D this time. It’s also a single CCD part, which keeps things as simple as possible to set up. 

We already know the 9800X3D is king in gaming, but it’s all these other reasons that make it the Best Overall CPU for 2024. We can highly recommend this one for new, high-end gaming builds especially, but it’s still an overall competent performer in workstation.

Most Balanced CPU: i9-14900K & R9 7950X

Original review | Newegg | Amazon

The next award is for the Most Balanced CPU, which is a category we’ve used for years to ensure credit is given to CPUs that perform well both in workstation or production applications and in gaming. This weighs all aspects of GN reviews, so it considers price, gaming performance, and production performance. 

We’re giving this to the AMD R9 7950X and secondarily to the Intel i9-14900K as a runner-up, now that its price has dropped. There are two reasons that it shares the category with the 7950X: First, Intel’s poor handling of its instability issues that has caused us to question the company’s successful resolution of all its problems; second: Intel’s efficiency is through the floor and power is through the roof.

• 

• 

• 

Currently, the 7950X (watch our review) has fallen to $480 and under, with the 14900K now at $440. These CPUs both land on this list for their impressive workstation and production application performance, balancing good-enough gaming with high throughput performance in use cases like file compression and decompression, rendering or code compile, and Adobe applications like Premiere for video editing.

The 285K had some impressive results in production, but it just has so many other faults and its price is so high that it doesn’t get a place here. 

The 9800X3D is a great CPU, but its limited core count is a significant disadvantage in these non-gaming tests we run. 

• 

• 

The 7950X, particularly with ECO Mode enabled, has some of the best efficiency we’ve seen in our CPU efficiency charts.

• 

• 

The Threadripper parts can outperform it due to their incredible performance, but for a balance of price and performance, the 7950X remains strong. The 9950X (read our review) also does well in these same categories, but is significantly more expensive right now.

• 

• 

• 

The 14900K (watch our review) doesn’t do well in efficiency (actually, it’s one of the least efficient), but its gaming performance is a little higher in most places compared to the 7950X and it trades places with the 7950X in some of our production tests. With the new lower price, it’s worth a second look. It’s just that the used market for these CPUs has been turned into a minefield with the stability issue parts potentially getting dumped secondhand. 

But ultimately, the 7950X has the advantage of being on a platform that will live for many more years. You may one day be looking at your own Best Upgrade CPU in our 2027 or 2028 round-up for AM5, and that’s the advantage AMD holds here.

Best Gaming CPU: AMD Ryzen 7 9800X3D

Original review | Newegg

Next up is the award for Best Gaming CPU. This one is extremely simple: We’re strictly looking for the best possible gaming performance and ignoring all other factors, including cost. This is a simple numbers-based category for framerate.

Fortunately, the best gaming CPUs these days are often cheaper than their non-gaming flagship counterparts. The AMD Ryzen 7 9800X3D at $480 gets this one with a clean sweep of the gaming charts. It isn’t even close in some situations, surpassing even last year’s winner, the 7800X3D, by sometimes large margins.

Let’s go by the numbers for this numbers-based award:

From our 7600X3D review, you can see how AMD has completely taken over the gaming charts.

The 9800X3D has an incredible lead in Baldur’s Gate 3, up at 160 FPS AVG to the 7800X3D’s 126 FPS AVG. We spent some time in our 9800X3D review explaining and exploring this one, as a 26.9% uplift was uncharacteristic. Ultimately, it replicated and we had data to support the finding. 

Dragon’s Dogma 2 is a 2024 title that’s extremely heavy on the CPU. This one had another break-out success, where the 9800X3D doesn’t even take a second to identify with how it breaks away from the rest of the stack. The CPU holds a 16% lead over the 7800X3D, which was already the best gaming part before that. Intel isn’t even close here.

F1 24 also has the 9800X3D advantaged. Even though it’s becoming limited by other components, it still holds a lead. 

Dawntrail is another like this: Gains are lower, but ever-present.

Starfield posted an impressive uplift for the 9800X3D as well, jumping to 169 FPS AVG from 145 FPS AVG on the 7800X3D, which was already a good result.

Frametime pacing is overall strong with the 9800X3D. There are some categories where its 1% and 0.1% lows improve disproportionately from the average, which is another great result.

Overall, the 9800X3D easily takes our Best Gaming CPU of the Year award for 2024, following up the 7800X3D before it.

Best Upgrade Value CPU: AMD R7 5700X3D

Original review | Newegg | Amazon

Runner-Up: Intel i7-14700K

The next award is for the Best Upgrade CPU, a category we added 2 years ago. This is the best CPU that can drop into a prior generation of motherboards. In-socket upgrades are great for consumers, reduce wasteful spending on more boards and RAM but also reduce e-waste, and keep builds more affordable. 

This year, the AMD R7 5700X3D easily secures the award. The preceding 5800X3D (watch our review) got Best Upgrade last year, but has since largely disappeared as a new-in-box purchase and its price has gone up. The 5700X3D has taken its place, commonly available between a shockingly low $180 or a more typical $230.

The 5700X3D (read our review) is able to achieve most of the performance of the 5800X3D CPU when looking at our gaming charts. Typically, it’s within 10 percentage points, often fewer. The 5800X3D benefits from a higher frequency though, so there are some situations where you’ll see the 5700X3D dip down the charts. 

One example is the Final Fantasy XIV: Dawntrail benchmark, where our 5700X3D ran at a framerate that gave the 5800X3D an advantage of 11%. This is one of the highest deltas we saw in gaming between the 2 parts. The 5600X3D (read our review) even outperforms the 5700X3D in Final Fantasy, which sometimes confuses people. The 5600X3D may have fewer cores, but its frequency is 300MHz higher. Both are 105W TDP parts, but reducing core count means more power budget available for the clock. Generally speaking, the 5700X3D is the better performer between the 2.

In Baldur’s Gate 3, we saw the 5700X3D just behind the 7600X3D and about 10 FPS behind the 5800X3D. That’s an incredible rank for a part around $230, especially considering it’s outperforming a $630 285K (read our review) or $440 14900K (depending on when you read this, as that price keeps changing).

Broadly speaking, the 5700X3D’s price is wild, and the fact that AM4 boards exist in high quantities on the used market also makes it a good consideration for a used build.

The CPU can socket into AM4 motherboards, keeping cost down. If you already have an AM4 board from an old Ryzen build, it’s likely that a BIOS update will allow it to support the 5700X3D. Some older boards, especially from the first gen of Ryzen, are spotty on their support, so check the motherboard CPU support list to be sure.

We’re going to give a runner-up mention to the i7-14700K (read our review), which is about $350 these days. If you’re on an Intel Alder Lake 12600K (watch our review) or something, this would possibly make sense to upgrade to.

Most Efficient CPU: AMD R5 7600X3D

Original review

This is a short one: This award is for the Most Efficient CPU we’ve tested this year, and it goes to AMD’s somewhat obscure Ryzen 5 7600X3D (read our review). This part is sold in Micro Centers and, from what our European viewers have told us, sparingly in some European retailers. It’s not widely available, but when it is, it seems to be around $300.

In gaming workloads, the 7600X3D had impressively efficient results. Its power consumption was just 43W in our Baldur’s Gate 3 test, which allowed it to run at 2.7 FPS/W and outrank even the 9800X3D and 7800X3D in efficiency (both of which were impressive already).

In Dawntrail, we saw chart-breaking results that required us to adjust our chart axis. The 7600X3D outdid everyone for efficiency.

This CPU is somewhat rare, but its combination of lower TDP and X3D on Zen 4 is what enables this performance. Definitely an impressively efficient part and we want to encourage more CPUs like this. It’s too bad about the limited supply: If it were more widely available, we might have selected it for our best mid-range CPU as well. Speaking of, that category is next.

Best Mid-Range CPU: i7-12700KF

Original review | Newegg | Amazon

The next award is for the Best Mid-Range CPU. This lineup is lacking in $150 to $200 parts so far.

• 

• 

• 

Considerations for this category would include: AMD’s R5 7600 non-X (watch our review) at $198 and benefitting from a modern AM5 platform; Intel’s i5-14400F or 13400F (neither of which make sense at their prices); and Intel’s i7-12700KF at $190. We’re giving this award to the Intel i7-12700KF. It feels weird since it’s from 2021 -- but the modern market means prior generation parts are persisting longer than typical.

Above $200 and with a gaming-only build, the 5700X3D might make more sense. But the 12700KF (watch our review) gives a great balance of workstation performance and gaming performance -- and it has a big benefit of not being on Intel’s plagued architectures.

• 

• 

In gaming, the 12700KF typically lands alongside the Intel 245K (currently $320) or AMD 5700X3D, 5600X, or 7700X and 9700X (read our review). The 5700X3D is more gaming-focused and tends to be a little more expensive, the 7700X (watch our review) is frequently more expensive, and the 9700X is $320 to $350.

In production, the 12700KF is similar to an R7 7700X or 9700X in many cases, outperforms most of the X3D parts thanks to its core count, and equates Intel’s 14600K (watch our review) or 245K. The 245K (read our review) does well in our production tests when compared to the 12700KF, despite its shortcomings overall.

It’s weird to put a CPU this old on the list, but AMD and Intel really haven’t had a lot of activity below $200 for a few years.

Best High-End Desktop CPU: Threadripper 7970X

Original review | Amazon

We’ll keep this one short. This is for Best High-End Desktop CPU. For the second year in a row, mostly because there aren’t any replacements, this is going to AMD’s Threadripper 7970X.

We like the 7980X, but its super high core count is more limited in use case. If you need 64 cores, you probably know you need 64 cores.

We’ve found the 7970X, which is the 32-core part, to be an excellent balance in production workloads that we use in our office. The 7970X does great with Adobe Premiere, has a ton of I/O capabilities with extra PCIe lanes that open up more storage options for true workstation-class computers, and along with other Threadripper parts, has some of the best efficiency results we’ve presented. 

• 

• 

It also performed exceptionally well in some of our SpecWorkstation benchmarks previously, to the point that a 7950X wasn’t even close in some comparisons. 

But for now, we’d advise waiting on Threadripper. It sounds like the beginning of next year will be busy for everyone, and current rumors about the 9950X3D may be reason to pay attention there as well.

Best Gaming CPU Under $100: R5 5600, i3-13100F, and i3-12100F

Original review | Newegg | Amazon

Next up is the best gaming CPU Under $100. This one is self-explanatory.

This goes to a split because the price is variable: The R5 5600, and the i3-13100F or i3-12100F all get attention.

Once again, multiple years running, the Intel i3-12100F ends up on this list - but unlike last year, the i3-13100F has also fallen below $100. The R5 5600 non-X also ends up on this list, with all 3 CPUs taking this category. Out of these three CPUs, the AMD R5 5600 (watch our review) is the best overall balance and performer in most categories. The only reason to exclude it in favor of the i3s is if it’s over $100 when you check this -- but right now, there are OEM CPU-only listings for $99.99, which technically qualifies for the list. Value goes down at its other common mark of $120 and it falls off the list at that point.

The 12100F (watch our review) is $77, which would be crazy if it weren’t 2 years old. The i3-13100F (watch our review) is $80. Neither of these F-SKUs have IGPs, which is why they’re cheap.

When we looked at the i3-13100F last year, we found that the 13100F frequently outperformed the R5 5500 by a significant percentage. The 5500, in our opinions, is not good enough to consider. It’s slightly more expensive than the 13100F but worse, and both are on dead platforms.

For gaming, looking at our data from last year when we had all 3 of these CPUs on the charts, the R5 5600 regularly outperformed the 13100F and 12100F. In fact, it was even over the 12400 in some tests. If you can find it under $100, the 5600 makes a lot of sense for this category. Even a few bucks over that is not bad.

• 

• 

But where the i3-12100F and 13100F are $20+ cheaper and where the strictest possible budgets exist, they are still capable performers. We recently reran the 12100F and found that it was capable in every game we tested, with one exception of extremely spotty frametimes in our Baldur’s Gate 3 test.

It’s impressive how much these 3 CPUs are capable of. They’re still playing games, and mostly without major frame pacing issues. It’s just unfortunate that Intel and AMD haven’t really launched anything new into this market for such a long time.

Biggest Disappointment: Intel Instability

The last category is the Biggest Disappointment. Intel gets this one for the second year running -- 2022 saw it going to AMD for the 4500. But this year’s disappointment puts all other ones into perspective.

The 13th and 14th gen stability issues get this year’s award, reminding us that where other disappointments simply suck, there’s always room to suck more. The 13th and 14th Gen CPUs experienced a mix of problems, including fab-level oxidation defects confirmed by Intel all the way up to wider-scale stability issues relating to voltage and microcode. Intel fumbled the handling of this and lost consumer confidence, dragging its feet on a response and then trying to craft a careful narrative when it finally did respond.

This was followed-up by the Ultra 200 series, which went back to simply being a disappointing set of products rather than a truly catastrophic failure. But this industry needs Intel and AMD to compete in a healthy environment: Intel of the past showed us what happens when the market has only one dominating vendor, and NVIDIA in GPUs shows us similar. It’s just not the job of consumers to prop-up a company that sells sub-par products -- or in the case of the disappointment this year, ones which have potential time-bomb failures baked-in. In the very least, Intel at least eventually responded with microcode patches that the company says has fixed the issue.

Best CPUs of 2024 Conclusion

Pricing today is a giant variable. It moves all the time, and especially as we approach Black Friday. All of this was based on pricing at the time we wrote this story, which is leading into Black Friday. Remember that we almost always consider price in our choices, so some of those may move around as prices fluctuate.

It’s always weird when some of the best parts are prior generation components, but that’s becoming more common in both the CPU and GPU space lately.

The biggest area that’s lacking in CPUs right now is the sub-$100 market, which is aging with the 12100F and 13100F. The R5 5600 is only rarely sub-$100 right now, but would probably become the go-to choice if it ever solidifies in that range. Sub-$200 is better, but still mostly older components.

That’s it for this round-up though. We love working on these and they’re a ton of fun to work back through all the components and refresh ourselves. It’s also nice to have some more positive lookbacks at components to remind us all that there have been some pretty good launches.

We have a lot more of this coming up, including a Best Cases of 2024 story that we’re excited to finish and post.

---

Table of Contents

• AutoTOC

Intro

AMD somewhat secretly launched its R5 7600X3D. It’s a 6-core, 12-thread X3D part that we bought for $300 from Micro Center in Charlotte. AMD didn’t email us about this CPU, didn’t do review sampling, and just sort of quietly launched it to the market through Micro Center. It’s similar to the 5600X3D launch (read our review here). 
This runs on Zen 4 and is advertised as having a maximum 4.7 GHz boost clock and 4.1 GHz base clock. TDP is listed at 65W. AMD’s 7600X has a higher 105W TDP and runs an advertised maximum clock of 5.3 GHz, or 4.7 GHz base. That’ll make the 7600X better in some specific frequency-constrained scenarios that can’t utilize the extra cache, like in our non-gaming tests, but it’ll also make it more power hungry and having that extra cache normally allows better performance in gaming, even if the frequency is lower, and it tends to be more efficient.

Editor's note: This was originally published on November 7, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Testing, Video Editing

Mike Gaglione

Video Editing, Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

That’d make the 7600 more of a direct comparison for efficiency testing, as its TDP is also 65W.

The 7600X3D was a surprise to us. A few weeks ago, we had a viewer ask why we hadn’t tested it yet -- the answer was that we didn’t know it existed, and we’re excited to work on it now. The CPU ends up being one of the most efficient we’ve yet tested thanks to its low TDP and high performance.

Let’s get into the review.

AMD R5 7600X3D Overview & Specs

Technically, the CPU is currently a Micro Center-exclusive like the 5600X3D, so we’re not sure if that’ll change or how many they’ll have.

Getting into pricing first:

CPU Price Comparison | GamersNexus (Early November, 2024)

	Newegg Price	Amazon Price
AMD 9800X3D (MSRP $480)
AMD 9950X	$585	$600
AMD 9900X	$430	$383
AMD 9700X	$327	$325
AMD 7950X3D	$598 (OOS)	$598
AMD 7950X	$493	$487
AMD 7900X	$396	$319
AMD 7900	$358	$368
AMD 7800X3D	$449 (OOS)	$476
AMD 5700X3D	$229	$187
Intel 285K	$630 (OOS)	NFS / OOS
Intel 265K	$400	$400
Intel 245K	$320	$320
Intel 14900K	$440	$440
Intel 14700K	$347	$347
Intel 13900K	NFS / OOS	$445
Intel 13700K	NFS / OOS	$343
Intel 12900K	$310	$277

Currently, the 7600X non-3D (watch our review) is $224 on Newegg and $208 on Amazon at the time of reporting. It’s $190 from Micro Center, which is the more like-for-like comparison to the $300 7600X3D.

Other relevant parts include the 7800X3D at 8 cores and 16 threads, priced at $480 or so on Amazon and $450 on Micro Center’s site. Between the two, it’s $150 to $180 more expensive to get the R7 7800X3D (watch our review) or the new 9800X3D, which is $480.

Intel’s most relevant, modern competition might include the $320 245K, $350 14700K, $344 13700K, or, oddly, the 12900K at $277.

This review is going to be kept as simple as possible. We just launched a massive deep-dive into the 9800X3D and included a ton of charts, so coming off of all that hard work, we’re trying to go with a straight-forward review today. If you want more depth and more discussion with even more charts, make sure to check out the 9800X3D review. 

For now, let’s get into a simple review of the 7600X3D.

AMD R5 7600X3D Frequency

AMD R5 7600X3D All-Core Frequency

We’ll start with a frequency inspection to understand the performance we see later.

Here’s an all-core workload. The 9800X3D we just reviewed ran at about 5225 MHz average, which is an impressive clock to hold for a 3D V-Cache part. It’s also a flat hold of the frequency thanks to the extra power budget. The 7800X3D fluctuates more due to the power constraint, holding around 4850 MHz on average.

The 7600X3D comes in below that, at 4700 to 4730 MHz, also moving up and down through the test.

For perspective, the 7600X non-3D part ran at about the same clock as the 9800X3D, up around 5225 MHz. If you see the 7600X outperform the 7600X3D in some tests, that won’t be a surprise: Anything which relies less on the extra cache than on higher clocks will see a regression in performance. This is less prevalent on the Zen 5 9800X3D due to its higher power budget and because the 9700X had a lower power budget.

AMD R5 7600X3D Single-Core Frequency

This chart is for the maximum single-core boost during a Cinebench single-threaded workload. This allows us to see how fast the cores go when unlikely to breach power limits.

The 9800X3D we just reviewed held the same 5225 MHz clock we saw in all-core testing and did so throughout the benchmark.

The 7800X3D ran faster than its all-core workload, which is more typical. That had it at 5050 MHz during this test on the fastest core per interval.

The 7600X3D is far slower than that, with a comparatively huge drop to 4750 MHz for the fastest single core per interval. That’s a large drop that will likely contribute to losses in certain titles, maybe like Final Fantasy, more than the core count difference between the two would.

The 7600X ran at 5450 MHz for the fastest single core, which will also benefit it in lightly threaded scenarios where the extra cache doesn’t contribute much.

AMD R5 7600X3D Gaming Benchmarks

Efficiency: Baldur’s Gate 3

For gaming efficiency first, we were impressed with the 7600X3D’s efficient performance in Baldur’s Gate 3. It’s the new most-efficient CPU in our testing.

Our efficiency benchmarks are a combination of both power drawn during the test and the performance output in the test, which is FPS in this situation. Tweaking either dial can improve efficiency since it’s a calculation of the two.

The 7600X3D ended up at 42.8W in this test, which has it reduced from the 5700X3D, 5800X3D (watch our review), and 7800X3D. It’s among the lowest power draw parts in this test, alongside the older R7 2700. With the 12W reduction from the 7800X3D and with performance similar to the 5700X3D, the 7600X3D ends up at 2.7 FPS/W, which is an awesome result that we didn’t expect to see. The 9800X3D is alongside it at the top, roughly tied with the 7800X3D at 2.3 to 2.4 FPS/W. This is a strong start. And for the 7600, that landed at 1.6 FPS per watt, which puts it above the 5600X, below the 5800X3D, and predictably above the 7600X, but it’s still below the 7600X3D.

Efficiency: Final Fantasy XIV

Final Fantasy 14: Dawntrail also has insanely efficient performance. The CPU is, once again, not the best performer on the charts, but it encroaches on the top and manages to do so while at 34W during this benchmark. Because the power is so much lower overall, the efficiency climbs despite a reduction in max FPS versus better AMD parts.

The end result is a 10.3 FPS/W rating, the highest we’ve seen, against 8.3 FPS/W on the 7800X3D. The 7800X3D is reaching the upper bound of what it can achieve for performance: It’s possible that this number increases with the launch of the RTX 5090, but for now, it’s stuck at 8.3 FPS/W. The 7600X original part pulls a little more power and performs far worse, landing it down at 5.6 FPS/W. That’s still ahead of anything Intel has to offer, including the 245K at 4.3 FPS/W, but nowhere close to the 7600X3D.

Stellaris

In Stellaris, the 7600X3D required 33.1 seconds to complete the simulation. This reduces the time required from the 7600X’s 36-second entry by 8%. The 9600X, from Zen 5, outperforms the 7600X3D, benefiting from the overall Zen 5 uplift we’ve seen in this specific game. Aside from that, everything else above the 7600X3D is significantly more expensive, including the 285K at $630 but providing only a 1.8% reduction in time required to simulate.

Against the 5700X3D (read our review), the 7600X3D sees a 14% reduction in time required. The same is true compared to the 5600X3D.

The new 9800X3D is aided hugely by its architecture and frequency, requiring 22% less time than the 7600X3D. The 7800X3D benefits from a 5.4% reduction in time required. Overall, the 7600X3D encroaches on the 7800X3D while being significantly lower cost, but can’t come close to the 9800X3D.

Dragon’s Dogma 2

Dragon’s Dogma 2 is up now. This is another title where the 9800X3D saw strong performance.

For this game, the 9800X3D holds a 21% lead over the 7600X3D in average FPS, at about 129 FPS to 106. The lows are proportional to the increase in average FPS. There is no particularly abnormal advantage to those found on the 9800X3D, although they are good.

Compared to the 5700X3D’s 103 FPS AVG, the 7600X3D holds a small 3.1% lead. It’s 3.5% ahead of the prior generation 5600X3D, which was also Microcenter-exclusive.

As for Intel, key comparisons include the 245K (read our review) at 95 FPS AVG and priced similarly to the 7600X3D. This gives the 7600X3D a 12% advantage in average FPS over the new 245K, with lows again proportionally higher. The i7 and i9 CPUs from the 13th and 14th series are all roughly tied with the 7600X3D, but slightly ahead. The 5800X3D maintains its position at the top of this grouping, beaten only by its newer alternatives.

Starfield

Starfield is up now. This was another strong one for the 9800X3D when we reviewed it.

In this one, the 7600X3D ran at 122 FPS AVG. That means the 9800X3D’s 169 FPS result has it about 38% ahead of the 7600X3D, a significant jaunt. The 7800X3D ran at 145 FPS AVG, offering its own improvement of 19% over the 7600X3D. The 7600X3D is definitely a lower-end part than the two flagships. Compared to the 5700X3D though, the 7600X3D is technically improved -- though basically within error at 3.2% uplift. The 5600X3D gives the 7600X3D a lead of 8.2%, widening the gap in favor of the newer Micro Center-exclusive part.

Intel’s showing includes the 245K at about the same framerate and a similar price, the 12900K at a higher framerate, and the 13600K also ahead. The 7600X3D is stunted by a combination of its lower core count and its frequency, not able to leverage the cache to the same extent as the higher-end AMD X3D parts. The 7600X3D at least holds about a 20% uplift over the 7600X, which was down at 102 FPS AVG.

Baldur’s Gate 3

In Baldur’s Gate 3, the 9800X3D experienced a large performance boost that we spent several minutes digging into in its review. It’s an outlier.

The 7600X3D is more typical: It ran the game at 116 FPS AVG, which allows the 9800X3D another 38% lead. The lows remain impressive on the 9800X3D, followed by the prior 13 series and 14 series CPUs as the next most impressive for lows. The 7600X3D is overall “fine” for lows, but its position is predictable within this stack.

The 7800X3D posted a 126 FPS AVG in Baldur’s Gate, giving it a boost of 9% for the extra $180 or so -- that doesn’t seem worth it if you have a Micro Center near you. The 5700X3D remains pretty close by as well, with a 111 FPS AVG that would be indistinguishable from the results of the 7600X3D. The 5700X3D also benefits from a fiercely competitive price.

The 245K slips down this chart, especially with Windows 24H2. It’s at 93 FPS AVG, giving the 7600X3D a 25% lead in average FPS. Lows are about the same.

FFXIV Dawntrail

Final Fantasy 14: Dawntrail is next. This one has the 7600X3D at 349 FPS AVG with lows at 164 and 79 FPS. That positions it roughly tied with the 7800X3D, both of which appear to be encountering similar limits. The 9800X3D managed to break free of this bind and post a 373 FPS result. That’s good, but it’s not as impressive as we saw in other tests.

But this isn’t meant to be another review of the 9800X3D. The 7600X3D boosts over the 7600X’s 259 FPS AVG by 35%, leading the 259 FPS entry by a significant margin. X3D in general completely divides this chart, forming a rift below the 5600X3D and everything else.

Intel’s closest competitor is the 14900K, although it’s more expensive, up at 310 FPS AVG. The 7600X3D leads even the former flagship i9 by 13%. Its lead over the 5700X3D is noteworthy, at 16%. This is because of the 5700X3D’s drop in clocks, which we know to impact its Final Fantasy performance. That’s also why the 5600X3D runs higher performance: Its clocks are typically at least 300 MHz higher.

Rainbow Six Siege

Rainbow Six Siege is up now. In this one, the 7600X3D ran at 613 FPS AVG -- sorry, 613.3… That 0.3 is important, because elite FPS gamers can see even the difference between 0.000798 ms.

The 613.3 FPS result has it up alongside the 9600X and below the 9700X (read our review). The 7800X3D ran at 622 FPS AVG -- again, sorry, 622.1 -- which means it holds a 1.4% lead. That’s basically error. These CPUs are functionally the same, with lows differentiated only by how wildly variable this particular game is.

As for the 5700X3D, the lead from the 7600X3D is a more noteworthy 17%. There’s also a big jump over the 245K, way down at 476 FPS AVG.

Generally though, the top CPUs are mostly the same. The 9800X3D is boosted, but not in a huge way like in some other games. We’re up against limits.

F1 24

In F1 24, the 7600X3D ran at about 409 FPS AVG and was flanked by the 5800X3D and 7800X3D. The 7800X3D holds a 7.2% lead in average FPS, with a big jump in the 1% and 0.1% lows. Both are ultimately acceptable. The 5800X3D’s 391 FPS average gave up a 4.4% lead to the 7600X3D. The gap against the 7600X is massive, at 409 FPS to 312 FPS, or a 31% uplift for the 7600X3D. That’s a big climb when considering the frequency and cache changes discussed earlier.

Intel’s 245K is far down the chart, more similar to the 7600X than the newer X3D variant. The 14900K is closer, but still behind at 385 FPS AVG.

Total War: Warhammer 3

In Total War: Warhammer 3, the 7600X3D ran at 484 FPS AVG. That has it in the top grouping again, but technically behind the 14700K. Realistically, in average FPS, they’re the same; however, the X3D CPUs all have large advantages in frametime consistency and pacing, as illustrated by the 1% and 0.1% lows.

The 9800X3D and 7800X3D are both encountering external limits. 

The real takeaway though is that Intel continues to get bodied by X3D: the 265K and 285K both suffer from the same impact we saw to the 14900K (read our review) and 13900K, where frametime consistency plummets and large frametime spikes pop-up more frequently. This affects the averaging as well, since it drags everything down, which leads to the 285K and 265K seeming to perform worse than a lower thread-count 245K. This is a repeatable result. Even the DDR5-8600 285K entry has worse lows than the 245K, and that’s all because of scheduling. There are community mods to try and help with this, but we test games as the developer ships and maintains them.

AMD R5 7600X3D Production Benchmarks

Production workloads aren’t a big focus for R5 CPUs, so we’ll fly through these to hit the basics.

Blender

In Blender 3D rendering on the CPU, the 7600X3D required 21.3 minutes to complete a frame of the GN intro animation. This is one of the slower results, putting it in-line with the 5700X3D, 3700X (read our revisit), and 5800X. The 7600X is faster than the 7600X3D, which isn’t new behavior: While the 9800X3D actually outperformed the 9700X, generally speaking, X3D CPUs are worse in production-heavy, all-core tasks than the CPUs they stack the cache on top of. That’s because the frequency comes down on prior generations, mostly for thermal budget reasons, and this is also why Zen 5 X3D is so interesting.

So for the 7600X3D, we’re back to a situation where it’d typically make more sense to buy any other CPU above it on the chart for heavy workstation use cases that we test. If you’re building a machine that’s gaming-focused first and foremost, then the 7600X3D should still be in consideration. It’s capable of handling production-style work on the side, but there are better options for a build more purely focused on that kind of work.

The 7600X3D is beaten by the 12600K, 5800X, and many other CPUs. Blender scales cleanly with thread count, and that shows here. The 285K is also relatively competitive in this particular test, as are the 265K and 245K.

Chromium

In Chromium code compile, the 7600X3D required 208 minutes to complete the compile. That has it way down the chart and near the 5800X and 5800X3D. The 7600X3D is struggling with its thread deficit and just wasn’t built to excel at the production applications we test with. Even the 3700X isn’t distant.

The 7600X benefits from a time reduction to 198 minutes in this test, with the 9600X and 12600K ahead of that.

7-Zip Compression

In 7-Zip file compression, the 7600X3D ended up completing about 87K MIPS, putting it ahead of the 3700X and behind the 5700X3D. The 7600X is again better, at 91K MIPS, or improved by 5.2%.

As with other pre-Zen 5 X3D parts, if doing this kind of work heavily, it’d make more sense to buy something else. The part is acceptable for its price if gaming is the primary focus.

7-Zip Decompression

Decompression is similar: The 3700X outmatches the 7600X3D marginally, the 7600X3D is just ahead of the 12600K (watch our review) now, and the 7600X non-3D runs 8.8% faster.

The 9800X3D stood out here as improving on the 9700X, benefitted by its higher power budget that enables higher boosting. We covered this in the review. The 16-core parts, like the 9950X and 7950X, have a major advantage in this workload that the 7600X3D can’t unlock with its 6-core configuration.

Adobe Premiere

Adobe Premiere is tested with the Puget Suite, using a mix of RAW performance, intraframe, effects, and processing. The result is in aggregate points.

The 7600X3D scored 7832 points, which has the 7600X ahead of it by 4.2% at 8159. Next above that is the 12600K. The 245K posts a significant gain over the 7600X3D here as well. Against prior parts, the 7600X3D at least improves upon the 5700X3D and 5600X3D components in a more noticeable way. It’s just not impressive on its own.

Adobe Photoshop

In Adobe Photoshop and tested with Puget Suite again, the 7600X3D fell down the charts to 14600K levels of performance and functionally tied with the 245K. That’s an OK spot to be, but AMD dominates the top half of this chart with other components -- including the 7600X. The reason for this drop comes down to the loss in frequency, which is important to Photoshop right now.

AMD R5 7600X3D Conclusion

The efficiency on the 7600X3D is crazy. That’s great, but most people -- especially in this part of the market -- don’t buy on efficiency. From a pure performance standpoint, we get the expected Zen 4 behaviors: The 7600X3D is a little worse in production than its non-X3D counterpart in most of our non-gaming tests, but is far superior in gaming tests. Because an R5, like an “Ultra” 5, is targeted at gaming, that’s the most important aspect.

Performance achieves most of what the 7800X3D would have to offer, but at $300. For gaming, it’s a better option than Intel’s $300 options, especially over the new 245K at $320, but worse in non-gaming tasks. If you’re mostly playing games and have a budget limited to about $300 on the CPU, we think the 7600X3D makes good sense. The 9800X3D offers a meaningful uplift at $480, but is also a huge jump in price and that extra money would possibly be better spent on a better GPU to unlock more performance. 

The 7600X3D’s biggest strength is its impressive efficiency. Its biggest weakness is its core count, where non-gaming tasks will be restrained. For users who tend to mix heavier “work” applications than gaming, we’d encourage considering other options to better suit those needs. Our 9800X3D review goes into some of that. But for gaming-first users near a Micro Center, the 7600X3D performs overall well.

---

Table of Contents

• AutoTOC

Intro

This is the most excited we’ve been about a case in a long time. HAVN's case focuses on trying to innovate on cooling performance. In the vertical GPU orientation, the bottom-mounted fans install at an angle and direct the air towards the glass and use an included glass flow guide to direct the air away from the GPU. That might seem counter-intuitive, but this will be easier to explain using our 3D animation.

HAVN -- that’s the name of the brand for this case -- is separating the hot exhaust air out the bottom of the GPU from the cooler intake, pushing intake at the glass side panel of the case and almost sort of “bouncing” it into the front of the video card. There are trade-offs, but it’s a novel approach for a case.

Editor's note: This was originally published on October 17, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing, Writing

Patrick Lathan

Testing

Mike Gaglione

Camera

Vitalii Makhnovets

Camera, Video Editing

Tim Phetdara

3D Animation, Camera

Andrew Coleman

Writing, Web Editing

Jimmy Thang

---

• 

• 

• 

• 

• 

The case also comes in a standard horizontal GPU configuration, which effectively disassembles the entire bottom fan tray solution into a more traditional, elevated flat bottom case. It’s quick to pull the VGPU option apart entirely and reconstruct it. We’ll come back to these animations when we explain the airflow patterns.

There are two spots for rear fans, enabling a lot of combinations of intake or exhaust -- the rear-bottom has a dust filter, indicating the intent. The top is more usual, and the side can be either fully filled with fans or filled with drive cages in a sort of spine-like setup.

The case also has large, well-made panels that are heavily ventilated, including an interesting rails system that glides the panels into place. 

We noticed some similar manufacturing techniques and processes to Hyte, and checking with Hyte, they use the same factory. The two companies aren’t related, but the panel quality shines through from the shared source.

This case is called the HAVN HS420. HAVN is a new brand that’s part of ProGamersware, which is part of the CaseKing company group. CaseKing is one of the largest computer hardware retailers in Europe. 

The HAVN HS420 is $200, with the HAVN HS420 VGPU at $270. That’s a big jump and accounts for the PCIe riser, bottom fan tray, and additional glass.

We’re excited to test this one, so let’s get into it.

The first HAVN case is pretty cool.

We actually found the case interesting enough that we’re doing our first fan optimization guide in years. We used to do these a lot for major case launches, but we’re basically testing a ton of alternative configurations in addition to the 6 or 7 we did for this review. That’ll be a separate story later since it requires a ton of additional testing. But for now, we still went heavy on configuration testing.

HAVN says its team members include people with experience from Fractal, NZXT, Cooler Master, Sapphire, and others. In addition to having previously worked for some of its now-competition, HAVN also has a hit list of cases it’s targeting:

• 

• 

HAVN has called-out a list of competitors for its HS420, including the Phanteks NV7, Lian Li O11D XL (not the EVO), Corsair 5000D, NZXT H9 (2023, assuming they'll introduce a new one), and the Fractal Torrent. For the HS 420 VGPU, it's the HYTE Y60 (although the Y70 may be more fair). These are the cases that HAVN intends to outcompete at launch.

As for our own list, we’d also bring attention to price competition from the HYTE Y70 non-touch, the Fractal North XL, the Antec Flux Pro with 6 fans, and the Montech KING 95 Pro. We’re highlighting these primarily for proximity in price to the $200 case while still being good cases.

HAVN HS420 Specs

	HS 420	HS 420 VGPU
Form Factor	Mid Tower	Mid Tower
Material	SGCC sheet metal, tempered glass, ABS, Nylon, zinc alloy, NdFeB magnets	SGCC sheet metal, tempered glass, ABS, Nylon, zinc alloy, NdFeB magnets
Case Dimensions (L x W x H)	541 x 259.5 x 547 mm	541 x 259.5 x 547 mm
Net Weight	17.84 kg	19 kg
Warranty	3 years	3 years
USB 3.2 Gen 2 Type-C	1	1
USB 3.2 Gen 1 Type-A	2	2
Audio Jack	1	1
Power & Reset Button	1	1
Motherboard Form Factor	E-ATX (Max Width 277 mm) / ATX / M-ATX / Mini-ITX	E-ATX (Max Width 277 mm) / ATX / M-ATX / Mini-ITX
Expansion Slots	6	6 horizontal or 4 vertical
Fan Support	Maximum Amount: 11 Top: 3 x 120 / 3 x 140 mm Front: - Left: - Right: 3 x 120 / 3 x 140 mm Bottom: 3 x 120 / 3 x 140 mm Rear: 2 x 120 / 2 x 140 mm (Max thickness: 30mm)	Maximum Amount: 11 Top: 3 x 120 / 3 x 140 mm Front: - Left: - Right: 3 x 120 / 3 x 140 mm Bottom: 3 x 140 mm Rear: 2 x 120 / 2 x 140 mm (Max thickness: 30mm)
CPU Tower Cooler Clearance	Height: Up to 185 mm	Height: Up to 185 mm
RAM Clearance	Height: If the radiator thickness is greater than 65 mm: 420 mm - 32 mm / 360 mm - 43 mm	Height: If the radiator thickness is greater than 65 mm: 420 mm - 32 mm / 360 mm - 43 mm
Liquid Cooling Support	Top: 280 / 360 / 420 mm (Thickness: 65mm, not overlapping with RAM) Front: - Left: - Right: 280 / 360 / 420 mm (Thickness: 60mm w/ GPU >317mm, 80mm w/ top AIO) Bottom: 280 / 360 / 420 mm (Thickness: 120mm) Rear: -	Top: 280 / 360 / 420 mm (Thickness: 65mm w/ taller RAM) [Ed. note: this is the same] Front: - Left: - Right: 280 / 360 / 420 mm (Thickness: 60mm w/ GPU >317mm, 80mm w/ top AIO) Bottom: - Rear: -
GPU Support	Length: Up to 480 mm Width: 195 mm Height: 2 - 5 Slots	Length: Up to 470 mmWidth: 195 mmHeight: 2 - 5 Slots
Power Supply Support	ATX / SFX / SFX-L PSUs Length: Up to 220mm (Clearance - 290mm)	ATX / SFX / SFX-L PSUs Length: Up to 220mm (Clearance - 290mm)
Storage	Right Fan Mount: 4 Storage Bays: 4 x 3.5'' / 8 x 2.5'' Cable Management Tray: 1 Storage Bay: 1 x 3.5'' / 2 x 2.5'' Storage Bay Included: 4x	Right Fan Mount: 4 Storage Bays: 4 x 3.5'' / 8 x 2.5'' Cable Management Tray: 1 Storage Bay: 1 x 3.5'' / 2 x 2.5'' Storage Bay Included: 4x
Internal Display Support	VESA Mounting: 75 x 75 mm / 100 x 100 mm Display Size: 14 in. (Max) Location: Front Right Pillar	VESA Mounting: 75 x 75 mm / 100 x 100 mm Display Size: 14 in. (Max) Location: Front Right Pillar
Dust Filter	Front: - Bottom: 1 (Removable) Top: 1 (Built-in) Right: 1 (Built-in) Rear Bottom: 1 (Removable)	Front: - Bottom: 1 (Removable) Top: 1 (Built-in) Right: 1 (Built-in) Rear Bottom: 1 (Removable)

Specs copied from manufacturer materials, please read review for our own measurements and opinions

The Build

• 

• 

The HS 420 is an unusual-looking case. The headlining feature is the panoramic glass panel, which is bent at a 90-degree angle with a tight radius. There’s a jutting-out cable management area that sticks out under the motherboard, a design consistency with rounded edges even under the case, and rubber string-like cable management suspension in the grommets.

The panel rolls into the case with four plastic wheels, then slots into place with support along the whole bottom edge. It takes some getting used to, but HAVN includes stickers detailing removal. The manufacturing precision required to get this working, and working well, is beyond what we're used to seeing and is executed expertly. 

All of the exterior panels are firmly magnetically attached and all include optional screw holes for extra security while shipping. You hear in most of our reviews that we favor at least optional screws for panel security to prevent issues.

• 

• 

• 

The visual theming is consistent throughout the whole case. It even has lots of obrounds -- vocabulary that we haven’t had to bust out since the Odyssey X.

The case’s sharp angles are rounded off, like the edges and corners on the PSU enclosure and the bottom fan bracket in the VGPU SKU. 

• 

• 

The HS 420 was impressively packed with a level of care that we usually only see with pre-builts: It has instructions on the box about how to correctly unload the case, a big "START HERE" label for the manual and accessories, and even a custom-shaped piece of foam under the fabric loop on the top panel to hold its shape. The manual itself is well done and reminds us of how Fractal does its own manuals, which we’ve praised several times.

• 

• 

The accessory kit includes a VESA mount, ESD-safe gloves, and extra combs for the rubber cable organizers inside the case. The only really important things that are missing from the accessory kit are spares for the variety of unusual screws used throughout the case. We're used to seeing screw organizers in $200+ cases, but HAVN instead includes labeled reusable bags… with unlabeled disposable bags inside. 

The only difference between the HS420 and HS420 VGPU is the vertical GPU hardware. The cheaper non-VGPU model comes with a normal fan tray in the bottom of the case and a GPU support bracket, neither of which are included with the VGPU (since they're replaced by other parts). 

• 

• 

The bracket is solid cast zinc alloy, braced against a completely rigid section of the PSU shroud to support the GPU at its furthest corner. It works better than most GPU supports and we liked it. Our only complaint is that the support can only work as shown in the manual for GPUs up to ~300mm long: any longer and the bracket can't reach all the way to the end.

As for the VGPU model, it comes with a preinstalled four-slot vertical GPU bracket and what HAVN claims is a PCIe 5.0 riser cable. Currently, there are no PCIe 5.0 consumer GPUs; however, it’s possible the 50-series may go in this direction. There is not a great non-enterprise way to validate the 5.0 claim from HAVN and performance impact is yet unknown, but likely minimal.

The VGPU kit isn’t sold separately at launch. 

The VGPU model’s bottom fan bracket is replaced with an angled bracket for 3x 140mm fans and a curved glass pane intended to direct the air. The GPU bracket can be adjusted back and forth, which is necessary in order to get the corner of the GPU flush with the curved airflow shield as shown in the manual. Our 4070 Ti Super isn’t fat enough to get as close to the shield as HAVN suggests. Using it precisely as intended requires a GPU with a cooler that's four slots thick.

Even without the VGPU hardware, HAVN recommends an unusual airflow pattern for the HS 420. Top exhaust, side intake, and bottom intake are all typical, but all of HAVN's recommended configurations show the lower rear slot as filtered intake and the upper rear slot as exhaust. We had tested this even prior to checking HAVN’s recommendations, as it does make sense to flow the air in and around the GPU.

There are 11 total fan mounts regardless of SKU, all of which can accept either 140 or 120mm fans except the VGPU's bottom mounts (which are 140mm only). No fans are included with the case, but two PWM splitters with 6 outputs each are included, so up to twelve PWM fans can be controlled from two inputs.

To help explore airflow in this unique enclosure, we made a 3D animation. 

HAVN HS420 Airflow Animation

• 

• 

• 

There are a ton of potential airflow paths with the way this case is built. Our custom 3D animation we made will help explain some of the more unique ones.

First, let’s look at how the VGPU kit will direct air. 

This will cover the theory, but we’ll talk actual performance and impact to thermals and acoustics in our charts soon.

• 

• 

• 

• 

Mounting the fans to the VGPU fan bracket at the bottom of the case angles them toward the glass side panel, resulting in an indirect flow path to the GPU fans. The bracket itself is at about a 30-degree angle. The shortest and most direct path to the GPU would be flat fans on the bottom pointed straight at the video card, but that path would sort of “collide” with the hot exhaust coming out of the bottom of the card. This designed and angled path tries to almost “bounce” the air off of the glass or flow it out and nearby, at which point the video card fans will pull that air in. There are two interesting challenges here:

• 

• 

• 

First is the absence of a wall: Vertically-mounted video cards, which is the intended use with this kit, with vertically oriented fins will eject the hot air down out the bottom of the card and up. In this orientation, that means hot air shoots straight down into the intake fans, theoretically getting mixed-in with the cool intake as it is shot at the glass. But since there’s no wall, the air will also enter the GPU more directly and at a wider spread at the front of it, including some potential exhaust, and likely at higher speed. Thanks to Bernoulli’s Principle, the general idea is that air increasing in velocity will decrease in pressure and vice versa, although there are exceptions. 

The upside is that the air will be directly fed to the GPU. 

• 

• 

• 

• 

Adding a glass shroud to the fan tray acts as a flow guide, but will also require the air to now make multiple turns to get to the GPU. The glass attaches to the back of the angled fan tray. This flow guide will push cool air at the glass side panel, at which point it’ll either physically deflect or will be overpowered and pulled-in by the GPU’s own fans. Likewise, hot air exiting the GPU bottom out of vertically oriented fins will now get shoved behind the glass and behind the fan tray. The back of the fan tray is fully enclosed by steel and glass, so there won’t be any hot air ingress. The glass shroud walls the hot air off. Instead, it ends up in a small chamber between the cable management shroud and the fan tray. This air will need to find a way out, and the most likely pathway is the bottom-rear fan. We’ll come back to that.

This results in the intake air now making multiple turns, which will affect both air velocity and air pressure: Velocity will decrease at the point it enters the GPU fans as a result of the turns and indirect path. Because the fans themselves are angled, the air ends up making close to a 90-degree turn -- shallower than if they were flat in the bottom with a flow guide, but still a big change in direction. 

The glass shroud itself has two points where it bends and a straight section: The first is a 40-degree angle, then about a 50-degree angle. These bends will help reduce resistance as compared to harder angles.

Back to the hot air from the card that was stuck behind the tray, glass, and between the cable management indent: In theory, as exhaust, a rear-bottom fan should help pull hot air trapped under the video card and behind the tray out of the case. It’ll also steal some of the cool intake from the back-most bottom-mounted fan in the tray, but the angle of the tray should reduce this impact. 

If the back fan is flipped to intake, it might help push that hot air through and toward the front of the case, where it’ll likely slowly be pulled up and out by the CPU cooler fans or top-mounted fans.

Flow-through video cards like most modern cards, including ours, will exhaust their hot air through the right-side of the card and effectively straight at the chipset and RAM area of the motherboard. A CPU tower cooler will pull this air out. In absence of that, a top-mounted liquid cooler with exhaust fans could remove it.

As for video cards with horizontally oriented fins, there’s no real downside with this setup since they can typically neither breathe nor exhaust through the bottom of the card. It would mostly exit the right and left sides, typically out the rear PCIe slots. These fin orientations are less common these days.

HAVN clearly thought about the configuration with the fan tray and the glass. There are a lot of intentional angles and flow direction is clearly intention. This configuration was thought about. We’ll have to test how much the glass and angled intake actually do by running thermals against just standard flat-bottom intake fans.

Let’s visualize some other setups in 3D: This configuration uses a vertical GPU, removes the bottom fans and the shield, and installs 3x side intake fans, 2x rear exhaust fans, and 1x top-front intake positioned about 30mm away from the front of the tower cooler. 

• 

• 

• 

In this orientation, flow follows this path: It mostly enters the side, at which point that air is pulled toward the tower cooler and vertical GPU by the fans on each. The GPU hot exhaust will mostly get pulled out of the way of intake through the bottom-rear exhaust fan, which also helps provide the currents needed to pull intake toward the GPU from both the bottom of the case -- passively through the floor -- and the lower side intake fan. The rear exhaust fan at the top helps remove exhaust from both the CPU tower cooler and the vertical GPU fin stack. 

• 

• 

• 

Finally, front-top intake is interesting: We always see people asking why you’d “fight” physics, but the rate of passive molecular through heat rising is incredibly slow. The truth of the matter is that a 1000+ RPM fan will overpower just about any natural air movement, and so what matters more is its access to external air, not whether it’s at the top or bottom of the case. Top intake would provide direct, external air straight into the CPU cooler. 

• 

• 

Putting a top exhaust fan here will mostly serve to steal cool side intake and exhaust it before it ever hits the front of the CPU tower cooler, reducing cool air available to the CPU tower. Now, you could instead move that fan back to be rear exhaust after the CPU tower cooler, and that’s a test we actually did run.

There are a ton of other configuration options in this case. We’re just showing what you can do with 6 fans in our animation. We have about 5 other configurations we ran, but we’ll go through all of those in our testing section.

There are tons more combinations and testing this will be complicated, but this should at least get you in the mindset for thinking about it as we work through the benchmarking and optimized layouts.

Cable Management, Drives, and IO

The HS 420 is a big enthusiast case that exists in an era where big enthusiast components are dwindling (GPUs aside). Despite HAVN repeatedly referring to the HS 420 as a mid tower, it's imposingly large at 18-19kg (empty).

• 

• 

If you want a big case, you have to find something to put in it. Multi-GPU is dead, which removes some motivation for ridiculous open-loop setups, and 3.5" and 2.5" storage drives are becoming less common. Despite its size and claimed support for motherboards up to 277mm, we don't recommend installing E-ATX boards in the HS 420, as they'll overlap the cable cutouts and look messy. That brings us, finally, to the VESA mount.

The mount can optionally replace the side fan bracket with a 14-inch screen inside the case, which could pull some consumers away from the Hyte Y70 Touch at the cost of some drive support.

• 

• 

Even if you don't need the drive trays, though, they look cool enough to keep installed. Each of the four boxes can fit either one 3.5" drive or two 2.5" drives, with one mount recessed for cable clearance. 

• 

• 

• 

Cable management is laid-out logically and has plenty of velcro straps, although cables need to be managed tightly to keep the flexible center of the side panel from bulging out. The ATX 24-pin, EPS12V, and potentially GPU and SATA cables all share a single channel, which requires more effort to manage. The screws and magnets are more than enough to keep the panel from popping off, though. The rubber strips in place of normal rubber grommets are also different.

At the rear of the case, the motherboard and GPU I/O is hidden within the rear panel, making it difficult to see or reach the ports. This is an unfortunate combination with the minimal front I/O, which consists of 2x USB Type-A ports and 1x USB Type-C, along with a single audio jack. On a positive note, the internal cable for the USB Type-C port is a flat ribbon cable that's easier to manage than the round kind.

HAVN HS420 Thermals

Thermal testing is complicated this time.

The case doesn’t come with fans. Because it endeavors to compete with the best cases on our charts, we needed to see if it could actually beat the best. We decided to standardize against a set of fans from one of the best cases on the charts: The Antec Flux Pro, which comes with 6 fans and has held the #1 rank in some tests recently. We pulled all 6 of its fans (4x 140, which is perfect for the bottom intake configuration and 2x 120 reverse blade), which allows us to make a direct head-to-head comparison. We’ve noted for a long time the limitations of our standardized fan test methods in cases, so for this review, we decided to experiment. We might start instead standardizing against a case that competes in a similar class, which would give fewer individual case reference points, but allow us to test against head-to-head competition in a controlled way with the same fans in a more optimized setup.

HAVN Test Configurations

So this marks sort of a new trial in our case reviews. It also enabled us to do more configurations for this case. Let us know if you like it and we’ll apply it to some future cases as well.

Here are the 8 configurations we ran. We only had a few days working on this case with Arrow Lake coming up, so we might do more later.

The tests labeled “A” are with some form of bottom intake. A1 has the angled tray and glass, A2 ditches the glass, and A3 goes flat-mount ditching both. The rest of their setup is the same.

Tests labeled “B” are with side intake. Remember, the budget is 6 fans to match the leading Flux Pro, so we stopped there for now. For this, we experimented with variations on the single 140mm fan in the top, moving it either in front of the cooler as intake or behind it as exhaust. We also experimented with the rear-bottom fan, swapping to either intake straight into the GPU or exhaust.

HAVN Only: Optimal Fan Placement by CPU & GPU Rank

This quick table shows the rank of each configuration in GPU thermals and CPU thermals. We’ll look at actual results in a moment, but this is a summary of the 6-fan budget. 

The best combined performer here was configuration B1, which had a vertical GPU without bottom fans, 3x side intake fans, 1x mid-front intake up top, and 2x rear exhaust fans. The best performer for GPU thermals was B3, which has a horizontal GPU, side intake, no bottom fans, and intake in the rear-bottom of the case, but it was at the cost of CPU thermals ranking 7th of 8 configurations. The best CPU performer was B2, which has a horizontal GPU, side intake fans, rear exhaust, and the top fan arranged as CPU tower intake.

Let’s move to individual data for inspection with just the HAVN case first.

HAVN Only: GPU Thermals at 100%

Here’s the chart for GPU thermals at full speed.

The total spread is 4.6 degrees, which is actually pretty wide for GPU thermal differences in our testing.

Starting with the glass flow guide: Our direct comparison of A1 vs. A2 has identical setup aside from the guide. On the chart, A1 with the angled fans and glass shroud performed the worst, with A2 performing 1.8 degrees better for GPU average temperature. That’s a real difference and outside of expected variance. A2 is identical but removes the glass shroud.

The best performer for the GPU was a horizontal GPU and side intake with B3. This isn’t too surprising. Horizontal GPUs often breathe better, but if you truly wanted a vertical GPU for display, our best-performing vertical setup was B1, which uses side intake, a top intake fan in front of the CPU cooler, and two rear exhaust fans. A3 is next, this time with 3x bottom intake fans but mounted flat rather than in the angled tray. This is a direct comparison to A1 and A2 without any changes. In our testing, it appears that the special bracket for the bottom fans and all the complicated curves and turns and glass are hindering performance. This makes sense. Air shooting straight in and without reroutes is just performing better with A3. 

HAVN Only: CPU Thermals at 100%

Looking at CPU thermals in detail: Configurations B2, B1, and A3 are functionally identical. A3 with the flush-mounted bottom intake fans did better for the CPU than those where the fans were angled at the glass. Config A2 was about 2 degrees worse than B2, so the total spread here is largely inconsequential. It’d make more sense to choose based on GPU performance than CPU tower performance between these configurations. Side intake is favored.

HAVN Only: VRM & RAM Thermals at 100%

For VRM and memory thermals, there’s another relatively wide spread of about 6 degrees on system memory. B1 runs at 13 degrees over ambient for DDR5, which is impressive. B3 runs at 19 degrees over ambient -- totally fine, but 6 degrees warmer shows an inefficiency. The cooler result benefits from an intake fan immediately above the RAM and VRM.

Time to get into the competitive benchmarks.

GPU Thermals: Full Speed

This chart is for GPU thermals with the fans at full speed. We’ve removed all configs for the HAVN case except the best and worst. You can find the full HAVN data in the prior charts. 

Again, remember that the HAVN case does not include fans. Because of this, its performance is basically whatever we want it to be, based on fans we choose. We normalized to the Antec Flux Pro, so that’s our direct comparison.

And the Flux Pro wins. Antec’s thermals are superior for the GPU while maintaining roughly equivalent noise levels, functionally noise-normalizing HAVN for the Flux Pro. The HAVN case still does well and ranks right alongside the Torrent, which has remained an impressive show of force for Fractal. It’s just that HAVN isn’t dominating the charts as much as we’d have liked for a case so thoughtfully focused on new approaches to fan placement. It’s a good start, though.

A1 has the bottom intake bracket and VGPU and landed lower down the chart. We’d favor the horizontal GPU solution. The HS420 HGPU B3 is outperforming the Antec C8, but only slightly outperforming the Hyte Y70 with half the fan count.

CPU Thermals: Full Speed

For CPU thermals, we’ve added B2 to this chart as it was the best for CPU performance.

At full fan speed, the HAVN with the best configuration we’ve tested so far landed functionally tied with the Flux Pro using its alternate PSU mount. Noise levels are about the same. This has the HAVN HS420 led by the Flux Pro with standard PSU orientation, the 216, and 207. The North XL is also about tied with the HS420, but is much louder at 45 dBA, where these HS420 configs land closer to 39-40dBA. The Antec comparison is the most relevant since the fans are identical.

HAVN HS420 Conclusion

The case does fine. It’s sad to see it not do better with all the special tuning, but looking back at our emails, even the HAVN team itself thought that the horizontal GPU configuration would do better in our testing. They were right. The vertical GPU setup, as fancy as it is, is still about balancing looks. Vertical GPUs are often disadvantaged, and HAVN hasn’t defeated that here.

Thermals are our focus. From a pure thermal performance standpoint, HAVN has not become the new best on the charts. The ideas are there, and for a first case, they got close -- but it is beaten by cheaper cases on this front.

Cases aren’t all about thermals, though. Despite that being our top priority, all 3 of us (Patrick, Mike, and Steve) who were hands-on with the case would consider buying one.

We all spent a lot of time personally working on this case. Each of us has gone through full assembly at least once on the system. All 3 of us were in total agreement that the quality is overall phenomenal and that we can’t overstate the attention to detail put into the HS420. For length, we cut a lot of commentary from this review. As a rapid-fire list, some small details we liked: 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

The top cover of the case has structural ribs that are specifically spaced to align with the hubs and edges of top-mounted fans, minimizing overlap with the fan blades. The right side panel does the same, intentionally aligning the blockages from structural elements with the dead zones on the fans. This is awesome attention to detail that we don’t commonly see. There are protective rubber dampers on everything: on the screws and the tabs that attach the fan trays to the case, on the 120mm adapters for the rear fans, on the posts that the bottom fan mounts rest on, on the airflow shield, and under a fragile corner in the middle of the rear panel. There are color-coded stickers with recommendations for routing cables through certain channels, and these stickers are placed over molded outlines. The case has optional screws everywhere, magnets to help click panels into place, and guides that are sturdy and well-built. The rear slots were thought-through with an optional filter on the bottom and none in the top, since blocking exhaust is pointless. The vertical GPU, despite its weaker performance, is positioned to maximize viability for hot cards.

• 

• 

It's easier to point out areas where this attention to detail is lacking, since we only really found two: first, there's no vibration damping for HDDs despite all the rubber washers elsewhere in the case, and second, the bottom filter is flimsy and doesn't eject smoothly.

The other main downside was just doing reconfigurations with the vertical GPU mount, which can make accessing cables along the motherboard challenging. Most people won’t do that more than once, though.

Case testing is complicated: It’s possible that a configuration with a different GPU, such as a 4090 Founders card, would behave differently. But based on our data, simpler might be better -- and the case would be $70 cheaper. Patrick and Steve both feel the same way, which is that we’d buy the $200 version of this case. It is much more competitive with direct peers at $200. At $270, it’s really only something you’d do for looks.

While the $200 case is worse than the Flux Pro thermally, the build quality and ease-of-installation features are far superior and the case is more future-looking with its concepts. The Flux Pro does well on tradition, and that’s just fine too. We like both cases. If you want something less traditional, HAVN manages to at least hit the basics while advancing the industry.

We don't want to beat a dead horse here, but this feels like the case that TRYX wanted to make with the LUCA L70 (read our review). This is a weird new design from an unknown company with some big money behind it, it's roughly the same size case as the L70, and the LUCA's price falls right between the two SKUs of the HS 420. The HS 420 is more solidly built, it's more practical, it has better cooling (once fans are added), and in our opinion, it looks better too.

Patrick said this about his concluding thoughts: “I would personally buy this case retail if I didn't need to fit a ton of drives; I haven't felt that way about a case we've covered in several years.”

Mike and Patrick both are hopeful that HAVN branches into smaller and cheaper cases. Patrick said, “I hope they don't just make a CLC. It feels inevitable that a company like this is going to make a CLC.”

The VGPU version is not really our thing. We think the cheaper one looks better and works better and is more widely compatible.

Wrapping things up, this is a great first showing from HAVN. The case represents a fantastic starting point for a platform to try and do some cool things thermally.

---

Table of Contents

• AutoTOC

Intro

Now we’re reviewing the Intel Core Ultra 7 265K - Core Ultra 7 (Series 2) Arrow Lake 20-core 8P+12E LGA1851 125W Desktop Processor.

We initially found the “Series 2” on the product box to be confusing, but then we realized that this is the Core 200 Series, and since 2 is sort of like 200, it all makes sense. These names won’t cause any problems and are very clear…
This is going to be the simplest and shortest of our 3 reviews for this CPU lineup. If you want the full depth and technical detail, check out the 285K review for the deep-dive into efficiency, gaming, and production. This one is going to focus on just the charts and conveying information quickly for you all as we wrap this series (for now).

Editor's note: This was originally published on October 26, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Mike Gaglione

Camera, Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

The 265K is currently available for $404 on Newegg and Amazon right now. That puts it predictably between the $630 285K and $320-$330 245K. As a recap, thus far, the 245K has made the least sense since it’s more gaming-oriented and less focused on potential workstation applications, whereas the 285K could maybe make an argument in some workstation or production use cases. The 265K theoretically balances between.

Intel 265K Price Comparison

We’ll start with a quick pricing recap of what’s available around the time we’re writing this review.

CPU Price Comparison | GamersNexusLate October, 2024

	Newegg Price	Amazon Price
Intel 285K (MSRP $590)	$630	N/A
Intel 245K (MSRP $310)	$330	N/A
Intel 265K	$404	$404
Intel 14900K	$470	$470
Intel 14700K	$375	$350
Intel 13900K	N/A	$415
Intel 13700K	$350	$290
Intel 12900K	$300	$280
Intel 12600KF	$160	$160
AMD 9950X	$600	$710
AMD 9900X	$430	$430
AMD 9700X	$330	$330
AMD 7950X3D	$600	$600
AMD 7950X	$510	$510
AMD 7900X	$400	$400
AMD 7900	$370	$370
AMD 7800X3D	$480	$480

The 265K is about $404 right now, which is fitting, because…Error Lake value can’t be found.

The price has it similar to the 7900X, which is currently $400. The 7900 non-X is $370 or so, with the 9700X at $330. The 9900X is $430. AMD has this territory encircled with similarly priced options. The 7800X3D comes in 19% more expensive at $480 and would also be less viable in most production use cases we test, but far more viable in gaming scenarios. If you’re only doing one of those, that allows you to either ignore or focus on the 7800X3D (watch our review). If you’re doing both, that’d be where the 265K might make sense -- if it’s not beaten by Intel’s own 14700K at $350 and 13700K at $290. 

And AMD has announced that its 9800X3D will be arriving on November 7th. We’re not sure what it’ll cost yet, so it’s possible it’s not a direct comparison here.

As a reminder, other considerations for Arrow Lake are that it requires an entirely new platform and Z890 boards are some of the most expensive we’ve seen in a given class. It also benefits from faster, more expensive RAM in ways that other CPUs can cut costs and corners. As a positive though, the reduced power consumption means reduced cooler requirements as compared to the prior Intel generations, so that can reduce some cost.

That’s it for the basics and positioning. If you want more, check out the prior two reviews or the efficiency bench setup video. Let’s just get right into it.

Intel 265K Efficiency Testing

Our efficiency testing has been explained in-depth in two pieces now and briefly introduced in a third. To learn more about what we’re doing here, check out our 285K review and our preceding video where we set up our monstrous test bench.

Efficiency: 7-Zip Compression

We’ll start with 7-Zip compression efficiency. We’re just going with the ATX12V and EPS12V rails for now, as we still think these are the closest to accurate on our setup.

The 265K ended up at 163W in this test, which has the efficiency at 968 MIPS/W. That has the CPU as less efficient than the 285K, which pulled 162W on the same ATX12V and EPS12V rails, but produced a higher throughput, yielding a 1051 MIPS/W result.

AMD’s 7800X3D, 7700, 9700X, and everything else, including the 3700X (read our revisit) from 2019, ranks above the 265K as a result. The 265K really is only better than the 9950X, 2700 (read our revisit), 5800X, and Intel’s own lineup here.

Efficiency: 7-Zip Decompression

In decompression efficiency, the 265K repeats its rank and falls below the 1194.9 MIPS/W result of the 285K. The power drawn is the same since this test suite runs both consecutively and at about the same power level. The end result is that the 265K is marginally less efficient than the 245K, more meaningfully behind the 285K, and is otherwise mostly just ahead of the prior Intel CPUs. Generationally, the improvement on the 14700K is actually good. It’s 47%. It’s a real uplift, and Intel does deserve credit for that; however, again, the entire top half of the chart is AMD-dominated. It’s just not even close. The 7950X in Eco Mode leads the 265K in efficiency with an advantage of 87%. The 9700X (read our review) is cheaper and also holds a substantial lead, up at 1624 MIPS/W. The 9700X won’t produce as high of a result in this test as the 265K for raw performance, but in efficiency, it is more efficient.

Even the 9950X, which is AMD’s highest power draw non-HEDT part we’re testing right now, is more efficient than the 265K. This is partly because its performance is so much higher in this test.

Efficiency: Baldur’s Gate 3

Here’s gaming with Baldur’s Gate 3. Here, the 265K pulled 89W when measuring both the ATX12V and EPS12V rails. That has it similar to the 285K. We noticed during this testing an initially higher reading on the 265K by almost 10W exactly, but after troubleshooting Vcore and CPU Package Power, we were able to isolate that 10W as being from a measurement tool difference on the 265K platform. We were able to make adjustments to effectively calibrate it for the data we had from all other tests on this chart. As we’ve stated, this is all brand new methodology and is more complicated because of ASUS’ decisions to split the rails for the CPU, but we’re staying on top of hunting down these deviations and accounting for them. This also means we’re going to continue studying the measurements and tools to refine the readings.

With the 89W measurement, the 265K ends up just below the 285K for efficiency (tied when rounded). Its performance is lower, but power is similar, so the two are effectively equal. The TDP is the same on these CPUs, so if they’re drawing up to the power limit, then this is expected.

Neither of these is particularly impressive when considering the 7800X3D or 5700X3D, both of which have higher framerate and lower power consumption, resulting in higher efficiency. The 245K (read our review) is more efficient than the 285K and 265K, up at 1.3 FPS/W and tied with the 7700 and 9700X.

Efficiency: FFXIV Dawntrail

In Final Fantasy 14, the 265K calibrated to prior tests pulled 65.5W during our testing. We didn’t add a line item for ATX5V in this one since we’ve been saying that we don’t think its impact on the CPU is significant in any way.

This lands the 265K at around the same power consumption as the 285K for the same two rails. The efficiency is lower because the framerate is lower. The 245K is more efficient, up at 4.3 FPS/W. 

The 265K is the least efficient of these three parts so far on this chart. It’s still improved on the 14700K (read our review), but because the performance is regressive in this particular title -- and by a lot -- the efficiency struggles to take off despite the reduction in power consumption from what was 98.2W.

The 7800X3D has an efficiency lead at 8.3 FPS/W, or a 131% improvement. This is actual insanity. The large framerate boost combined with the steep power reduction gives AMD a big lead here. Again, whether you’re talking outright efficiency or environmental stewardship and absolute power consumption, the answer to both of these would be AMD as the most efficient. Intel cannot fight on these grounds, despite improvements over its own prior architecture.

Efficiency: Stellaris

Stellaris is up.

In this one, we have the 265K at 1.5 simulations per watt-hour, about the same as the 5800X, especially considering rounding. 

If we look for a performance-normalized comparison, the closest would be Intel’s own 14700K. The uplift over the predecessor is 50%. AMD’s 9600X is somewhat close in performance as well, yet holds an advantage at 1.7 simulations per watt-hour.

Efficiency: Starfield

We didn’t capture power data for the 245K in Starfield, but we have the 265K and 285K.

The 265K here pulled 144W, putting it around the same as the 285K. Because the performance declines and the power is the same, the efficiency is worse than the 285K once again. The 285K appears to be a better bin in combination with higher efficiency.

The 265K ends up about the same as the 14600K. Fortunately, it improves on the 14700K’s 0.6 FPS/W result. The 14900K has a lower power consumption than the 14700K as a result of an external bottleneck limiting CPU utilization and also the binning. It can’t be fully leveraged. The 285K and 265K are both fully engaged though.

Intel 265K Gaming Benchmarks

Dragon’s Dogma 2

Dragon’s Dogma 2 is up. In this one, the 265K ran at 99 FPS AVG, which has it 4.6% ahead of the 245K and means the 285K is about 4-5 FPS ahead of the 265K. The 14700K leads the 265K by 8.6%, with the 13700K about the same. The AMD 7800X3D is ahead by 11%, with the cheaper 5700X3D also ahead of the 265K while running on an older, cheaper platform.

The 265K is worse value than even the 245K in this particular game, and we wouldn’t recommend that one, either. Its low performance here is fine, with no particularly meaningful deviation from the expectation, but overall, the 265K is just not impressive in this test even against its already unimpressive brethren of the same 200 Series. Or Series 2. Or whatever Intel is calling it.

F1 24 - 1080p

F1 24 is up now. In the very least, the gap is wider against the 245K, now with a 7.5% uplift in framerate. Lows increased proportionally with the average.

This positions the 265K as equivalent to the AMD 7700X and a bit ahead of the 7700 (watch our review) and 7900 non-X parts. The 7900X is functionally tied with the 265K, making it both performance and price matched in this test -- at least, at the prices right around launch.

The 285K’s 343.5 FPS AVG has it just 4% ahead of the 265K. It wouldn’t be worth buying the $630 285K regardless, and especially not for gaming, but it’s especially not worth it here. Even the 265K makes more sense, and it still doesn’t make a ton of sense.

The 5700X3D (read our review) further establishes that storyline with its 355 FPS result, led further by the once-upon-a-time cheaper 5600X3D (when it was briefly available at Micro Center). That one has a lead because its frequency is higher.

F1 24 - 1440p

At 1440p, we see some drop in performance from the resolution change, but a similar hierarchy overall. The 265K ends up sandwiched between the 14600K and 13600K as we bounce off of occasional framerate limits. The entire top of the chart is brushing against external bottlenecks at least occasionally, so let’s move on.

FFXIV Dawntrail - 1080p

Final Fantasy 14: Dawntrail is up now, first at 1080p.

AMD holds a domineering lead over this chart and keeps a stranglehold on the top 3 entries, with 4th held closely against the 14900K with APO on. 

The 265K landed at 236 FPS AVG here, which has the 12900K ahead of it by 4%. The R9 7900 non-X also leads, up at 253FPS AVG and holding an advantage of 7.4%. Even the 7600X leads. The 14700K’s 287 FPS AVG result has it 21.7% ahead of the 265K, establishing, without a doubt, that the 265K has gotten thoroughly “wrekt” in this test against its predecessor. The 285K also did poorly in this game; in fact, this is one of the games Intel openly stated it has regressive performance in, and we can definitely confirm that. Almost anything else makes more sense than Arrow Lake in this benchmark.

Baldur’s Gate 3

Baldur’s Gate 3 had the 265K at 96 FPS AVG, which ties it almost perfectly with the 12900K. That includes tying the 1% lows, with 0.1% close to the usual wider error range. The 265K only leads the 245K by a few FPS on average, rendering it relatively uninteresting. The 245K was already terrible value against alternatives here, and that includes not just the clearly dominant X3D CPUs -- where we have 4, and soon to be 5, topping the chart -- but also Intel’s own prior CPUs. The 14700K leads the 265K by 6%, and for perspective, the 7800X3D leads it by 32%. The 5800X3D (watch our review) isn’t far below that, at 120 FPS AVG. The 5700X3D, which is around $200 to $230 lately (but has been cheaper), has an advantage over the 265K of 16.4%.

For the 265K, we’re just not seeing it in this one.

Stellaris Simulation Time

Stellaris is up now. This one uses a late game save file and tests for simulation time rather than framerate. Players of 4X games are likely aware of how bogged-down the CPU can get in late game stages. 

The 265K required an average of 33.9 seconds for its simulation. The 13900K averaged 33.5 seconds, with the 7700 non-X at 34.2. The 14700K predecessor is indistinguishable from the 14900K (read our review) here, with both exhibiting a 2% simulation time reduction from the 265K.

The 9600X is also improved, with a simulation time reduction of 1.2 seconds. Zen 5 generally does well in this test, shown also with the 9700X at the top.

The 285K with our default settings did better in this one relative to the 14 series than other tests, but was still beaten by AMD’s 7800X3D and 9700X.

Rainbow Six Siege

Rainbow Six Siege is up now. In this test, the 265K ran at 519 FPS AVG. We observed that the 285K had frametime pacing issues in this game as compared to the 14 series, which persist here. Our 285K review remarked that we think this may be related to specific optimizations made by the Rainbow Six team, as APO’s benefit has been reduced to effectively nothing.

By average FPS, the 265K is behind the 5700X3D, and further behind the higher clocked 5600X3D. Both of these AMD CPUs have better 0.1% lows; however, the 14600K has far superior 0.1% lows than all 3 of these and is functionally tied in average FPS with the 265K. It’s within error for average, but better in low performance. The same goes for the 13600K, embarrassingly for the 265K.

Here’s how it stacks up: The 7800X3D leads the 265K by 20%, the 9600X leads it by 19%, the 14900K with APO off by 13%, and the 285K with APO off by 12%. Even the 7900X (read our review) leads here, and that’s not explicitly been branded as a gaming part, yet competes in both production workloads and price.

Starfield

Starfield is up now. This one has the 265K at 134 FPS AVG, just ahead of the 5800X3D and behind the 13700K (watch our review) and 14700K. These two CPUs (and the 14900K) are all bouncing off of another limit, and the 285K with DDR5-8600 makes it appear as if that limit is memory. This is also reinforced by the cache-boosted 7800X3D propelling to the top here.

The 265K has an improvement on the 245K’s average FPS of 120.5 of 11%. It also leads the 5700X3D here by 13%.

AMD has historically had issues with Starfield, despite being the GPU sponsor for the game. The 9600X is less competitive in this gaming test than some of the others, down at 101 FPS AVG.

265K Production Benchmarks

Time to move on to production benchmarks. This testing will look at a shortened list of workstation applications.

Blender

Blender is up now for a 3D rendering benchmark.

In this one, the 265K required 8.7 minutes to complete the render. That has it near the 14900K. Against the 14700K, the 265K also benefits from a render time reduction of 8%, meaning 8% less time required to complete the work. The reduction in time required against the 13700K is 20%, down from almost 11 minutes. AMD’s 7950X in ECO mode leads the 265K with an 8% reduction in render time needed, with the non-ECO result at 7.4 minutes and in line behind the 285K. 

The 265K beats the 7900X, which at least helps its positioning against similarly priced AMD competition. That isn’t always the case.

The 265K also benefits from a 31% reduction in render time requirement against the 245K.

7-Zip Compression

7-Zip Compression is up now. In this test, the 265K ran at 158K MIPS, which puts it roughly tied with AMD’s R9 7900X and behind the 9900X. The 265K improves on the 245K by 29% in throughput, with the 285K leading the 265K by 8%. This is a test where we saw a large improvement from the memory upgrade in our 285K test with DDR5-8600. We might revisit that topic if we can find some time.

The 9900X (read our review) is about 6% more expensive than the 265K right now and performs about 3.5% better. The 14700K is cheaper than the 265K, but uses more power to score 8% better. The 7950X (watch our review) is one of the more interesting CPUs still, but it depends heavily on price. It’s listed at $500 as we write this, or about 24% more than the $405 listing for the 265K on Newegg. The 7950X performs 18% better without Eco Mode on and similarly with it enabled.

7-Zip Decompression

In Decompression, the 265K ran at 168K MIPS and landed just behind the two-generation-old 13700K. The refreshed version of that, the 14700K, is up at 195K MIPS, landing the 14700K ahead of both the 285K and the 265K. The 285K only topples it when upgraded with faster memory, but of course, giving the same treatment to the 14700K would also leapfrog it ahead.

The 14700K leads the 265K by 15.7%. The 265K leads AMD’s 7700X by 19%. X3D doesn’t really help here, so the 7800X3D and 5700X3D fall down the stack comparatively, despite strong gaming performance and really good efficiency.

AMD’s production-oriented 7900X leads the 265K by a staggering 24%, with the 9900X leading by 27%. The 7900X is currently the same price as the 265K, making it a better value in this comparison.

Adobe Premiere

Adobe Premiere is up next, tested with the Puget suite.

In this one, the 265K scored 10718 points in aggregate. Puget Systems takes the intraframe score, RAW score, GPU effective, and other filter and editing testing into consideration for this score.

The score has it roughly tied with the eco mode 7950X and 9900X. Its advantage over the 7900X is at least a little more, at 8%. The 7900 (watch our review) scored similarly. The 265K also leads the 13700K by 1%, so basically error. The 14700K leads the 265K by 2.4%.

The 285K did well in this particular test and managed to outrank the 14900K, leading the 265K by 5.8%.

Intel 265K Conclusion

The 265K is, in several cases, less efficient than our 285K. That’s not unheard of: The 285K is a higher performer, which benefits the efficiency, and the power budget is the same. Still, it’s improved over the 14700K. Intel has retained that much, but as we said before, in all comparisons we run, AMD is more efficient still.

So that wraps-up that side of things.

It’s clear once again that the 265K is surrounded by good options on all sides. The 285K is exceptionally bad value for gaming users, with worse performance than you’d get on a $230 5700X3D or $480 7800X3D and also a far higher cost. In non-gaming uses, there are some limited scenarios where you could make an argument for the 285K, but they are relatively rare among our test suite.

The 245K made even less sense: Production use cases effectively vanish as an argument, as the part is majority targeted at gaming users. In gaming, it gets absolutely crushed by not only AMD’s $100 cheaper 5700X3D (which itself would give you $100 more budget to either keep or throw at a GPU), but also by Intel’s own predecessors that are now cheaper or the same price.

The 265K suffers from a bit of both: The cost goes up by $70-$90, and yet the performance doesn’t scale anywhere close to linearly in gaming. The production performance is better, and in that situation almost solely, the 265K can make some stronger arguments for itself. It can outmatch its closest price competitors from AMD at times, so that’s at least good for the 265K. It’s just not a sweeping victory and Intel’s total platform cost is also questionable, especially with the potentially short lifespan of this one.

That’ll wrap our reviews of these for now. We’ll have more Arrow Lake content, but as far as the core part reviews, they’re done until Intel launches more. We wanted to keep this one concise.

---

Table of Contents

• AutoTOC

Intro

AMD’s 9800X3D review embargo lifts today. The 9800X3D is a Zen 5 CPU with extra cache, but with critical changes to the location of the cache within the die stack. 
As we talked about in a separate technical video, the 9800X3D has shifted the extra cache under the core complex, leading to more direct contact from the CPU cores to the IHS lid underside. AMD has also eliminated some of its bonding layers within the CPU, which reduces insulation further, and it has removed the structural silicon that leveled the top of the cache die to better mate the IHS in prior X3D designs. That’s because the extra cache die and the core complex die are now the same physical dimensions.

Editor's note: This was originally published on November 6, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Mike Gaglione

Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

The 9800X3D is a $480 MSRP part for the AM5 platform. Let’s get into the review.

AMD 9800X3D Overview & Pricing

The only way AMD could screw up this launch would be with the price, so let’s start with a price comparison and update. This was all collected in the days leading up to review publication.

CPU Price Comparison | GamersNexus | Early November, 2024

	Newegg Price	Amazon Price
AMD 9800X3D (MSRP $480)
AMD 9950X	$585	$600
AMD 9900X	$430	$383
AMD 9700X	$327	$325
AMD 7950X3D	$598 (OOS)	$598
AMD 7950X	$493	$487
AMD 7900X	$396	$319
AMD 7900	$358	$368
AMD 7800X3D	$449 (OOS)	$476
AMD 5700X3D	$229	$187
Intel 285K	$630 (OOS)	NFS / OOS
Intel 265K	$400	$400
Intel 245K	$320	$320
Intel 14900K	$440	$440
Intel 14700K	$347	$347
Intel 13900K	NFS / OOS	$445
Intel 13700K	NFS / OOS	$343
Intel 12900K	$310	$277

The 9800X3D’s MSRP is $480. It’s not listed as we write this.

Assuming that price is accurate, some of its direct neighbors and competitors include the others on this list. The 7800X3D is about $476 now and is out of stock with a listed former price of $450 on Newegg. The 7800X3D (watch our review) has previously been as low as $300 to $330, with drops in August to around $360. Lately, its price has stabilized higher. We have a policy of making comparisons against pricing at the time of the review, so we’ll use the $476-$480 figure today.

The 5700X3D is priced at $187 to $230. The 7950X (watch our review) is a similar price to the 9800X3D, landing at about $500. That’d be one to pay attention to for non-gaming workloads.

Intel’s competition includes the 285K, to the extent you can even call it “competition,” seeing as it barely competes with Intel’s own parts on value. The 285K(read our review) is $630 where it can be found, and you shouldn’t buy it, with the 265K(watch our review) at $400. The prior generation parts, like the 14900K, have fallen in price lately. That’s at $440, with the 14700K at $347.

We have a lot to get through, so let’s get into the benchmarks. We’ll have frequency, gaming, efficiency, and production tests.

AMD 9800X3D Frequency Comparison

AMD 9800X3D All-Core Frequency Comparison

We’ll start with a comparison of frequency against the 7800X3D. This will help explain the performance differences we’re going to see today.

In an all-core workload with Blender, we observed the 9800X3D maintaining an average all-core frequency at a fixed 5225 MHz without any dropping. The stability and flatness of this frequency is a result of the higher power limit, which is allowing the core all the power it needs to maintain this boost.

The 7800X3D averaged between 4800 and 4890 MHz during the original review cycle for this same test. It’s significantly lower and less fixed, shown by the line’s variability.

AMD 9800X3D Single-Core Frequency Comparison

In a single-thread comparison with Cinebench, the 9800X3D also held a maximum single-core boost of 5225 MHz throughout the duration of the test. The 7800X3D held at 5050 MHz in the same test as a maximum single-core frequency.

Between these two tests, the 9800X3D holds a significant frequency advantage over its predecessor in both all-core and single-core loads.

VID Comparison

We also ran a voltage ID comparison just with software. In Blender, the 9800X3D held a VID of 1.18V on average, with the 7800X3D at about 1.07V. The 7800X3D is power limited and also running at lower clocks, which combine to yield a lower VID.

AMD 9800X3D Thermal Benchmark

Using the same Blender all-core workload, we measured a Tdie CPU temperature of approximately 77 degrees Celsius in a 21-degree Celsius environment when using a 360mm Liquid Freezer II at 100% pump and fan speeds. CPU die average was similar.

The L3 cache plotted just over 50 degrees Celsius, with the CPU IOD temperatures around 41 degrees with hotspot readings at about 50 degrees.

The 9800X3D cannot be directly compared to the 7800X3D thermally by using internal sensors picked up by software. It would be erroneous without accounting for the changes AMD has made in its sensors. The AMD temperature sensors have been relocated within the silicon and moved to cooler locations to prevent clock dropping too early. We talked about this in our 9700X review and provided a controlled comparison so you can better understand why the sensor readout isn't directly comparable between the architectures.

9700X vs. 9800X3D Thermals

It can be directly compared to the 9700X, though. The 9700X pulls less power in this test and so will be cooler, but it’ll still help to compare. The 9700X (read our review) ran a Tdie of about 52 degrees Celsius under the same test conditions and with the same cooler.

With PBO maxed-out on the 9700X, we saw about a 13W higher power draw than the 9800X3D stock. This landed the 9700X at about 86 degrees Celsius. It isn’t perfectly comparable because the power isn’t exactly matched and we ultimately don’t have an easy way to directly compare the 7800X3D’s thermals from the cache change without solutions that require more time than AMD gives in a review cycle. Even still, the key takeaway is that the 9800X3D is not burning the cores at excessive temperatures even with the stacked solution.

AMD 9800X3D Gaming Benchmarks

Let’s get into gaming benchmarks.

Stellaris Simulation Time

Stellaris is a 4X game that we test for simulation time rather than framerate. Lower is better, with the result measured in seconds.

Zen 5 has already shown advantages in this test, with the 9700X previously scoring the new top rank in the benchmark and outmatching all prior CPUs. Core count has limited benefit in this beyond 8 cores for AMD, but IPC and clocks seem to matter.

The 9800X3D absolutely crushes the prior top result, with a 25.8-second result against the 9700X’s already barrier-breaking 30.3-second result. This previously appeared to be a limitation of the memory, which we can see reinforced by the 31.1-second 285K result at DDR5-8600 vs. the stock result at 32.5. 

The original X3D CPU with the 5800X3D (watch our review) posted a 35.4-second result against the 40.7 result of the 5800X, so we’re seeing a repeat of that era, which was also impressive. 

The 9800X3D also reduces the 7800X3D’s simulation time by 17.6%.

Intel’s best showing is the 285K with faster memory, but that’s not a like-for-like comparison as we could boost memory on AMD also. The 285K stock required 32.5 seconds to complete the simulation, meaning the 9800X3D reduces time required by 21%. The 7950X3D shows why we haven’t retested it, underperforming overall for reasons explained in our review of the part. Even with the new data, it underperforms vs. the 7800X3D.

The 14900K and 14700K are effectively tied and within error of each other, and removing any limits wouldn’t change that much.

This is a great showing for the 9800X3D.

Dragon’s Dogma 2

Dragon’s Dogma 2 is up now, another of our 2024 titles. This game, even with all its updates to improve CPU performance, is still heavy on CPUs. The 9800X3D distances itself from the 7800X3D with a 16.2% uplift, gaining on the original 111 FPS AVG to an impressive 129 FPS AVG. We reran this one multiple times to ensure the result was right, but it ended up not even being our biggest gain in the suite.

The gain on the 14900K (read our review)was about 17%, up from 109.6 FPS AVG. The same is true of the prior 5800X3D. Over the 285K, we saw a 23% uplift. 

Frametimes were improved in this game, but not as much as we saw in some other games. The improvement is more or less proportional to the average framerate, so there is no particularly impressive divergence in lows. In other words, it tracks with what we’d expect of a 129 FPS AVG.

FFXIV Dawntrail - 1080p

Final Fantasy 14: Dawntrail is up now, released this year. The 9800X3D breaks the previous upper bound of the chart. 

Let’s run through the flagships: The 9800X3D CPU ran at 373 FPS AVG, which has it at 5.6% higher framerate than the 7800X3D while costing the same as the current 7800X3D pricing at time of writing. The 14900K ran at 310.3 FPS AVG, giving the 9800X3D a massive lead of 20.2%. This is one of the games where the new Arrow Lake CPUs are hugely regressive and impressively bad value, so the 285K’s 270 FPS AVG gives the 9800X3D an overwhelming lead of 38.3% higher average framerate. As for the 5800X3D, the lead is 11.8% over the prior AMD gaming flagship’s 334 FPS AVG.

For some anchors to older parts: The R7 2700 (read our revisit) and 2600 (read our revisit) were around 145 to 150 FPS AVG, with the 2600 benefiting from a frequency advantage. The 3600 (read our review) was at 170 FPS AVG. The Intel 12100F is down at 186 FPS.

FFXIV Dawntrail Frametimes

For frametimes, we observed overall similar frame-to-frame intervals for the 9800X3D and 7800X3D. Typically, they oscillate between 2ms and 4ms intervals, where 16.667ms would be 60 FPS. The spikes are what matter: The 9800X3D in this title has slightly shorter excursions from the mean than the 7800X3D. The difference is hardly perceptible, but on a technicality, the 9800X3D does deliver more consistent frametimes in this title.

Baldur’s Gate 3

Baldur’s Gate 3 from 2023 is up next. We want to caveat this one with a disclaimer: We saw a larger than typical gain in the 9800X3D versus the 7800X3D for this. After we spent a day troubleshooting it, we have come to the conclusion that there is no evidence to suggest any testing errors or uncharacteristic issues caused by the test or game itself, so we’re going to run the data because it’s interesting. We’d pull it if we had any substantiated concerns. We checked with two other reviewers in peer review, ran HWINFO logging, reran the tests multiple times on both the 9800X3D and 7800X3D, and came to the same conclusions.

The game has the 9800X3D way up at 160 FPS AVG, leading the 7800X3D by an obscene 26.9%. The 7800X3D in both our shown dataset and our unshown additional tests have it in the 124-126 FPS range. This was one of the games we double-checked and got the same results each time.

The 14900K ran at 105 FPS AVG, giving the 9800X3D a crushing advantage of 53.2%; the 7800X3D already had a lead over the 14900K of around 21%, depending on which data set we looked at. 

The 9800X3D leads the 285K’s 100 FPS AVG by 60%. Against the 5800X3D, we’re seeing a boost of 33.3% from the prior 120 FPS AVG.

Baldur’s Gate 3 Frametimes

Baldur’s Gate 3 frametimes for the 9800X3D also improved over the 7800X3D, and this time, it wasn’t just proportional to the average uplift. Here, you can see the 9800X3D maintained a tighter frame-to-frame interval with reduced amplitude of excursions from any given prior frame.

Baldur’s Gate 3 Monitoring

Because this was such an outlier, we ran HWINFO logging against the test as a separate run. 

In this test, the 9800X3D average all-core frequency was about 5225 MHz, which is impressive for all-core. We think this is why the performance is so disproportionately good: There’s a much higher power budget, so we’re not trimming the frequency peaks, and the power consumption itself overall isn’t excessive in this game. The two combine for full boosting.

Plotting the 7800X3D, we see a much spikier, less predictable average all-core frequency that bounces between 4500 MHz and about 5050 MHz under load. The highest single-core frequency is also variable. This is contributing to the behaviors discovered in this test. 

Rainbow Six Siege

In Rainbow Six Siege, the 9800X3D landed at 643 FPS AVG and led the 7800X3D’s 622 FPS result by 3.4%. There are limited gains to be found in a game already over 600 FPS. We’d have to re-evaluate this with whatever flagship NVIDIA puts out next to try and find the true CPU ceiling. It at least gives perspective that you may not see big differences in heavily bound scenarios.

The 14900K ran at 586 FPS AVG here, so the 9800X3D has a 10% improvement. The new X3D part is also about 10% over the 285K. Compared to the 5800X3D, the 9800X3D gains 12% in AVG FPS.

We removed some other CPUs from this one to make space for APO on results. As a reminder, APO is now doing little for performance in any of our tests. This includes Rainbow Six, which previously saw larger impact. The 14900K is about 10 FPS different with APO on. The 285K saw change only within error and run-to-run variance so it’s irrelevant.

Starfield

Starfield is up now, a 2023 title.

In this game, the 9800X3D leads the 7800X3D by 16%, impressing once again generationally. The new result is 169 FPS AVG to the 145 FPS of the 7800X3D.

The 14900K is led by 24.6% with its 135 FPS result. The new 285K did a little better than the 14900K in this game, at 143 FPS AVG. That’s still an 18% higher average framerate for the 9800X3D. Even with the DDR5-8600 memory in Gear 2 with the 285K, it still caps-out at 152 FPS AVG. That was a great result, but the 9800X3D puts things into a new perspective.

Against the prior AMD 5800X3D flagship and its 128 FPS result, the new 9800X3D has a 31.4% uplift.

Some quick reference older parts include the 2600 at 59 FPS AVG, the 2700 at 66, the 3600 at 70, and the 12100F at 71 FPS.

F1 24 - 1080p

F1 24 is another of the games with a lower boost over the 7800X3D. The 9800X3D ran at 464 FPS AVG here, a 5.8% improvement on the 438 FPS result of the 7800X3D. The 14900K ran at 385 FPS AVG, so that’s about a 21% uplift for the 9800X3D. The 285K was massively regressive in this game and was down at 344 FPS, giving the 9800X3D a 35% advantage. 

Against AMD’s flagship from the 5000 era, the lead is 18.6% over the 391 FPS result of the 5800X3D.

Total Warhammer 3

Total War: Warhammer 3 is one of the older games we test, but is important for perspective on GPU and memory bottlenecks.

The 9800X3D ran at 490 FPS AVG, which is within error of the 7800X3D. These two are bound by the same bottleneck, which is the GPU in this situation.

There’s only a 5% lead over the 14900K due to the same limitation, with an 8% advantage over the regressive 285K. The 5800X3D ran at 457 FPS AVG, so a 7% gain for the 9800X3D. The 14600K with its recent rerun we ran landed it up alongside the 14700K (read our review), with both encountering the same bottleneck as the 13700K and the 7800X3D. Sometimes people ask why a 13700K could be better than a 14700K. It isn’t: It’s just that we can’t actually see unbound scaling, limiting the usefulness of this test. 

This game has 0.1% low issues with the i9 CPUs due to their thread count. We don’t test community FPS mods, so it’s up to the devs to fix this.

Efficiency Testing

We’ll now get into efficiency testing, then production benchmarks. This testing looks at CPU performance per Watt, typically presented as FPS/W for games (which would be frames per joule). We’ll start with MIPS/W for 7-Zip, though.

These tests look at a simple formula of the power in Watts drawn versus the performance of the task. Adjusting either of the two parameters has an impact on results.

Efficiency: 7-Zip Compression

In 7-Zip compression, the 9800X3D computed to 1298 MIPS/W, or millions of instructions per joule, putting it barely ahead of the 5700X3D. The increased power consumption of the 9800X3D reduces its efficiency as compared to the lower power 9700X, at 81.6W to 98.4W. The 9700X scored 1389 MIPS/W.

The 9800X3D ends up relatively efficient in compression, but less efficient than the prior 7800X3D. The 7800X3D was not a higher performer, but because this is a calculation of both power and performance, either one can help to counterbalance a deficit in the other.

Meanwhile, Intel’s 285K measured at 1051 MIPS/W, which is improved upon its prior efficiency performance in the 14900K, which was way down at 672 MIPS/W. That’s with the EPS12V and ATX12V line measured on the 285K. The AMD CPUs do not pull from ATX12V in any meaningful way in our testing. The PCIe slot is isolated, as are fans and RGB LEDs. 5V from I/O is also isolated. This means there is some overhead in ATX12V on Intel from the RAM and other small devices, but nowhere near enough to meaningfully move that needle closer to AMD.

Efficiency: 7-Zip Decompression

Decompression efficiency is up now.

In this one, the 9800X3D ends up at 1482 MIPS/W, just under the 5600X (which is benefitted by its 60W power reading) and above the 7950X at 183W. The 7950X in ECO mode is more efficient for its trade outperformance, down to 133W and now at 1936 MIPS/W. The 7950X3D boosts higher, mostly because its power draw drops even more -- now at 124W for 16 cores. The 7950X could also be limited to 124W and would achieve a similar score.

The 7800X3D is more efficient than the 9800X3D due to its lower power consumption, but its performance is also lower. The 9800X3D trades efficiency for more boosting headroom. This benefits it in the gaming tests we saw earlier, despite costing more power.

Intel’s closest CPU is the 285K toward the bottom of the chart, at 162W in the same test and 1194 MIPS/W.

Efficiency: Baldur’s Gate 3

Moving on to games, we’ll start with Baldur’s Gate 3.

Baldur’s Gate 3 positions the 9800X3D as the new king of efficiency in the test, with a 2.4 FPS/W result that has it just above the 2.3 FPS/W result of the 7800X3D. The 7800X3D utilizes about 13W lower power as averaged, resulting in a smaller gap than we might otherwise see. The 9800X3D at least maintains AMD’s trend of X3D parts becoming new efficiency leaders in gaming.

Intel’s CPUs first appear with the 245K at 1.3 FPS/W. The 285K is below that, at 1.1 FPS/W, followed by Intel’s last generation parts.

Efficiency: Starfield

Starfield efficiency is up now.

The 9800X3D is almost at the top, but not quite. Its performance gains were relatively high in this game, but the power consumption increased significantly. The 9800X3D averaged at 98.7W, with the 7800X3D at 69W. This is what leads the 7800X3D to a victory in this test, despite the performance uplift. AMD has lost efficiency here.

Intel’s CPUs first appear at the 285K, down at 1 FPS/W. The 14900K is at about 0.7 FPS/W.

Efficiency: Stellaris

Stellaris is something of a repeat of that: The 9800X3D pulled 54W on average, leading it to a 2.6 simulations per Watt-hour score. The 7800X3D ended up at 2.7, or about a 4% advantage in efficiency despite a reduction in performance compared to the 9800X3D. The 12W lower power leads to this discrepancy.

Intel’s closest part is the 245K at 1.9 simulations per Watt-hour, with the 285K at 1.5.

Efficiency: Final Fantasy XIV

Final Fantasy 14 is last for efficiency comparisons in gaming.

For this one, the 9800X3D ranked at 7 FPS/W, which has it below the 5700X3D (read our review) and 7800X3D. The CPU pulls just under 54W on average, while the 7800X3D ran at about 43W and the 5700X3D ran at an impressive average of just 39W. Despite higher performance, the higher power tips the scale away from the extreme efficiency we’ve seen in previous X3D parts.

It’s still relatively high in the ranks, above everything Intel and above AMD’s non-X3D parts, but it’s clear that AMD favored an increase in power for an increase in performance as it tries to balance between.

AMD 9800X3D Production Benchmarks

Production benchmarks are next. These tests look at a suite of applications outside of gaming. The 9800X3D is ultimately a gaming CPU. Our non-gaming tests do not regularly show advantages for extra cache.

Blender

Blender is up first for a 3D rendering pass of our intro animation.

The 9800X3D completed the frame render in 12.5 minutes, about tied with the Intel 245K (read our review). Although we don’t recommend the part, the 245K is cheaper at $320. AMD’s 9800X3D manages to at least outperform the 9700X, with an atypical render time reduction of 16%. It’s atypical because prior X3D parts do not necessarily see such gains, such as the 7800X3D at 15.9 minutes to the better results from both the 7700 (watch our review)and 7700X. The reason for the atypical gain is largely the power budget, where the 9800X3D has more power available out-of-box to clock up.

The 13700K also requires less time. The 9800X3D is improved, which is better than we’ve seen in past X3D parts in this test, but there are still far better performers and value options from Intel and AMD alike if applications similar to this test are your daily use case.

7-Zip Compression

7-Zip can be one of the more sensitive to cache, but it depends on whether it’s compression or decompression.

In compression testing, the 9800X3D completed 128K MIPS, which puts it within error of the 14600K and 13600K (watch our review). It at least posts a 13% jump over the more power-constrained 9700X. The lead over the 7800X3D is similar. 7950X3D (watch our review) performance is about the same as the 7950X and within about 1%. 

Intel’s 285K has a large lead here, up at 170K MIPS with DDR5-6000. The gain in our DDR5-8600 test was massive in this benchmark, putting it at nearly 202K MIPS. The 9950X approaches that result, meaning an upgrade to its memory as well would leapfrog the 285K.

7-Zip Decompression

Decompression has the 9800X3D at 146K MIPS. It’s between the 12900K and 7700X (watch our review). The 9700X was regressive in this test against both the 7700X and the more like-for-like power 7700.

The 9800X3D leads the 7800X3D by 10.2% and the 9700X by similar.

Again though, if you’re heavily represented by this test, you’d be better off with a different CPU. The 16-core AMD CPUs lead this chart and set that example. The 7950X has been below $500 lately, so it’d be a price comparable alternative to the 9800X3D that’s more suited to this.

Chromium Compile

Chromium compile is up now. 

The 9800X3D requires 130 minutes to complete the compile with our settings, putting it at the same level as the 14600K (read our review) and Intel 245K. The 7900 non-X (read our review)leads the 9800X3D here and benefits from the extra threads.

The 9800X3D also benefits from an 18% reduction in time requirement against the 7800X3D’s 160-minute result. Likewise, it leads the 149-minute result of the 9700X. Still, the 12900K, 13700K (watch our review), and 285K are all advantaged over the 9800X3D, as are AMD’s own 9950X and 9900X (read our review)parts.

Adobe Premiere

Adobe Premiere testing is done with the Puget Suite.

This testing lands the 9800X3D at 10,050 points in aggregate for the extended test, which includes RAW, intraframe, and effects performance. The result is between the 13700K and 9900X above it and 12900K (watch our review) and 7900X below it. 

The 9800X3D does actually improve on both the 9700X and 7800X3D, though, both of which are around 9100 points in aggregate.

Intel’s 265K outperforms the 9800X3D by 6.6%, with the 7950X (watch our review) a bit above that. The 285K does impressively well in this specific test, with the 14900K alongside it.

Adobe Photoshop

Photoshop also uses the Puget Suite.

The 9800X3D does impressively well in this benchmark, with its averaged score landing ahead of the 9950X (read our review) and 9700X alike. The 7800X3D’s 10,162 point result gives the 9800X3D a lead of 17%.

AMD does well in this specific Photoshop test right now. Intel’s first CPUs show up way down the chart, at around the 7600X (watch our review)and 7900 levels of performance with the 14900K and 14700K. The 285K does about the same as the 13700K here.

This is one where the 9800X3D manages to boost itself even in a non-gaming scenario, benefitted by the extra power available for clock boosting as compared to the 9700X.

Value Comparisons (USD/FPS)

This chart is an experiment. We don’t want you to place too much usefulness in this particular chart because it is experimental, but we also can’t try new things if we always try to perfect them first.

This chart shows the delta in USD per FPS at various simulated prices for the 7800X3D. The 9800X3D is fixed at $480 in all of these simulations, with the 7800X3D variable based on the number you see in the chart legend on the right. A negative value means that the 9800X3D is that amount cheaper per FPS than the 7800X3D, while a positive value means the 9800X3D is that amount more expensive per FPS than the 7800X3D.

At near price parity, the 9800X3D has significant value advantages in Dragon’s Dogma 2, Baldur’s Gate 3, Starfield, and even Phantom Liberty. Value is functionally the same in F1, Rainbow Six, and Final Fantasy. 

It isn’t until a $420 price for the 7800X3D that the Starfield value gains are mostly eliminated. At $400 7800X3D versus $480 9800X3D, the 7800X3D starts to make a ton of sense as a better value, assuming that’s how you shop. 

Efficiency vs. Performance Change

This is also an experimental chart. 

In this one, we’re showing the percent increase in power consumption for the application in the left axis vs. the percent increase in performance. The best result would be a larger performance increase than the power increase by percent, although linearity is also good.

In Baldur’s Gate and Stellaris, with the latter converted to simulations per hour so that higher is better, the percent performance increase is either improved or nearly linear with the percent power increase. In Starfield, the performance increase costs a substantial power increase. We saw the same in Phantom Liberty. These are our two most power-hungry games in the suite and fully leverage the power budget. 7-Zip also saw this swing. Dawntrail is bottlenecked, so can’t reliably be used. Even limited though, power is higher.

AMD might be past the most efficient point in the curve, but results like Stellaris support that it is still relatively balanced. 

AMD 9800X3D Conclusion

Intel’s 285K was already lowered into its grave, but the 9800X3D just contributed a significant aqueous mixture of hydroxide, chloride, potassium, and about 95% water onto it. After this impromptu hydrolysis reaction, there aren’t many compelling reasons to pick one up.

To recap some key points quickly:

The 9800X3D is definitely the new gaming king, and this time, it’s priced similarly to the 7800X3D. The original launch price of the 7800X3D was $450. Even at that price, the 9800X3D is a worthwhile improvement and doesn’t feel like stagnation. 

There are some big boosts to performance over the 7800X3D: Stellaris saw the 9800X3D break through a glass ceiling and propel to new heights against the already-dominant 7800X3D; Starfield saw about a 16% improvement; Dragon’s Dogma 2 saw a 16% uplift, also breaking through what seemed like a ceiling; Baldur’s Gate 3 had a 26-27% improvement, which was such a big swing that we re-ran the tests on both CPUs, collected HWINFO logs during execution, and closely inspected the graphics to ensure equal renders. We also checked with another reviewer who saw a 17% uplift in the game while testing in a different area and under different settings, which is close enough when considering those changes.

Even without that, the CPU is just overall competitive in gaming.

For efficiency, the 9800X3D doesn’t manage to hold onto AMD’s multi-generation history of its new X3D CPU becoming the most efficient in our charts. The CPU is still good, and we’d still classify it as “efficient” overall when considering everything else on the charts.

Production isn’t as competitive as other parts, so if you’re not building a gaming-first system, you should just buy a different CPU. But it’s good enough for a mix of work and high-end gaming, and likewise, it uncharacteristically improves upon the 9700X. We haven’t always seen this with X3D, as frequency is typically sacrificed for power as a result of the thermal limiters imposed by the heat sandwich AMD previously had with the stack.

Intel isn’t anywhere close on the gaming charts. It isn’t only beaten, its new 285K flagship -- which is, for some completely insane reason, $630 -- is sometimes beaten by 40-60%. Typically, we’re seeing 25-35% when unbound by the GPU. 

Intel does better in the production benchmarks, but we still wouldn’t recommend the Ultra 200 series broadly. The 9950X and 7950X remain competitive, with the 14900K and 14700K also somewhat competitive here. There are some very limited use cases, such as the buzzworded “creation” scenarios, where the 285K makes some sense. But Intel is marketing the 285K on efficiency, and AMD is more efficient in every test we’ve run.

The entire last few months have been wild: We watched AMD open it up with Zen 5, where it drunkenly fumbled the football back 5 yards on the field. Intel then picked up the ball and proceeded to run towards its own goal, at which point it tripped over its shoe laces that came unglued. AMD then picked the ball up and walked it the rest of the way with the 9800X3D.

And somehow, it worked. AMD was set up for a slam dunk on this launch with both its own and Intel’s fumbles, and by under-marketing the performance, it was in a position where only the price could screw the launch.

This launch was also the most organized out of the last several months, including AMD’s own Zen 5 launch. An organized launch is important: With about 2 weeks of testing time rather than the few days we’ve had, plus a mature platform without mid-testing reworks, it is clear that AMD really sat down and planned the launch rather than shoving it out the door in a panic to respond to something else. This gives us some additional confidence that there won’t be any unexpected, major problems with issues like BSODs or BIOS.

That’s it for the 9800X3D review. You have all the numbers to make decisions. The 7800X3D may make more sense in the event its price falls significantly, but overall and all things totaled, we’re positive on this one.

---

Table of Contents

• AutoTOC

Intro

Now we’re reviewing the Intel Core Ultra 245K CPU, which you can think of as sort of a 15600K, just with a new name. This is a 6 P-core, 8 E-core part with 14 total threads. It’s priced at $310 MSRP and is meant to be a new mainstream gaming part.

Intel’s big Arrow Lake claim is efficiency and power consumption reductions, so we’ll spend a lot of time looking at that in addition to gaming and production performance.
Key comparisons against competing modern CPUs would be the 5700X3D, such as if you’re on AM4 and can upgrade one more time, the 9600X, and Intel’s prior generation parts. We’ll get into all of that.

Editor's note: This was originally published on October 25, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Mike Gaglione

Quality Control

Jeremy Clayton

Camera, Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

Intel Core Ultra 5 245K Overview

Again, this review is hyper condensed compared to our 285K review. We really are leaving a lot of detail, but that’s because it’d be complete repeat information and is the same. We’ll focus mostly on charts today.

Intel Core Ultra 5 245K Price Comparison

Let’s start with the price comparison. These prices were taken in the week before reviews are going live, so they may change by the time it publishes.

CPU Price Comparison | GamersNexusLate October, 2024

	Newegg Price	Amazon Price
Intel 245K (MSRP $310)	$330	N/A
Intel 14700K	$375	$350
Intel 14600K	$250	$225
Intel 13700K	$350	$290
Intel 13600K	N/A	$225
Intel 12900K	$300	$280
Intel 12600KF	$160	$160
AMD 9700X	$330	$330
AMD 9600X	$250	$250
AMD 7900	$370	$370
AMD 7700X	$275	$275
AMD 7600	$200	$200
AMD 5950X	$350	$350
AMD 5700X3D	$230	$210
AMD 7800X3D	$480	$480

Here’s the list. Prices for both Intel’s 13th and 14th Series CPUs have recently dropped, as have some of AMD’s Zen 5 CPUs.

The 14700K is as cheap as $350 right now, which makes it a direct alternative to the 245K, despite buying into an older platform. It has at least gotten microcode updates that should resolve the major concerns. The 14600K is $225-$250 now, which is about the same price as the AMD 9600X. The 13600K is dwindling in supply, but $225 lately. The 12900K is also cheaper than the 245K, despite being on a pretty old platform, and is $280-$300. AMD’s 5700X3D (read our review) is around $210-$230 now, making it a key upgrade pathway for AM4 users. The 7800X3D is priced out of this territory, up at $480 typically. Keep in mind that a 9800X3D -- or whatever they end up calling it -- is due out on November 7th.

Here’s a slide showing the basic specs. The 285K is a 24-thread part and has 8 P-cores and 16 E-cores. Intel ditched hyper-threading, so it’s just the two core counts combined for thread-count. The 245K is a 14-thread part and has an advertised boost of up to 5.2GHz. Overall, Intel is focusing heavily on power consumption reductions and efficiency improvement.

But in the 285K review, we saw several areas where the CPU regressed in gaming performance. This complicates efficiency, because if the CPU is doing less work but also lower power, it’s not a clean improvement.

Let’s start with all of that.

Intel Core Ultra 5 245K Efficiency Testing

This section gets into efficiency testing. There are some really important details in our 285K review that you need to see to fully understand and appreciate this section. We’re not going to go through it all again here. The most important part is that ASUS is directing some power through the ATX12V cables on the 24-pin connectors now, so we can’t rely on purely EPS12V. We’re going to leave a lot of detail out here on the assumption you got it in the longer review.

Efficiency: 7-Zip Compression

In 7-Zip compression efficiency testing, we measured the 245K with EPS12V and ATX12V at 108.1W on average, which produced a MIPS/W rating of 1130.6, or in other words, 1130.6 million instructions per Joule.

The 245K outranked the 285K’s efficiency in the same test, meaning that the 285K is beyond the optimum point in the curve for efficiency. It’s still improved over the preceding 14900K.

The 14600K scored 768.3 MIPS/W here at 166.5W, so the 245K is improved in efficiency. The 245K’s performance in Compression MIPS is actually lower than the 14600K’s (read our review), but its power consumption dropped so drastically (even with ATX12V measured) that it has measuredly improved in efficiency, at 47% more efficient than the 14600K. Adding the 5V line includes I/O and so isn’t perfect, but even that is more efficient. We talk about why we add both in the 285K review.

AMD remains the leader, with the 9600X at 1152.8 MIPS/W, the 5700X3D at 1281, and the 7700 non-X at 1436.2.

Efficiency: 7-Zip Decompression

7-Zip decompression uses power from the same test since we combine them, but has different scores. In this one, the 245K held a score of 1090 MIP/W. Its actual MIPS throughput, which we’ll show later, is significantly behind the 14600K, so the efficiency percentage uplift is dragged down by regressive performance, despite the total power reduction. The improvement over the 14600K’s 821.9 result is 32.6%.

The closest score in MIPS throughput to this is the 5800X3D (watch our review), which is just outside of error. The 5800X3D and 245K are therefore completing about the same output and are normalized in this way, allowing the 5800X3D a staggering lead of 28% more efficiency than the 245K.

Efficiency: Baldur’s Gate 3

In Baldur’s Gate 3 efficiency, the 245K ran at 71.5W and scored 1.3 FPS/W. That has it tied in efficiency with the R7 7700 and it tied with the 9700X which wasn’t distant. These three CPUs can be looked at as nearly work-normalized, or FPS-normalized, in addition to scoring the same efficiency. 

The 5800X3D has a substantial framerate lead and runs at only 65.7W, giving it a 1.8 FPS/W result at 38.5% more efficient than the 245K. The 5700X3D has a larger lead with its 51.1W result and is also typically cheaper, though on an old platform now with old I/O support. Still, it’s impressive.

Comparing the 245K to the 14600K, the lead is 44% uplift in efficiency compared to the 14600K’s 0.9 FPS/W result. The 14600K also had a slightly lower framerate in this one.

Efficiency: FFXIV Dawntrail

In Final Fantasy 14: Dawntrail, the 245K ran at 4.3 FPS/W and 51.8W, putting it behind the 9600X (which has a significantly higher framerate but about 10-11W higher power consumption) and ahead of the 14600K. The improvement over the 14600K’s 3.6 FPS/W result is 19%. The 5700X3D crushes here, up at 7.7 FPS/W and leading the 245K by 80% thanks to its lower power and its higher framerate.

Efficiency: Stellaris

Stellaris shows us simulations per Watt-hour, so higher is better despite the test being in seconds (with lower being better). 

In this test, the 245K ran at 53.5W and completes 1.9 simulations per Watt-hour. That has it about tied with the 5800X3D, which likewise is roughly tied in power consumption. These are good direct comparisons.

The 5700X3D’s power consumption drop benefits it greatly here, although it slows down in simulation time. Its score is 2.4 simulations per Watt-hour, or an improvement over the 245K of 26%.

Against the 14600K, the 245K improves by 46%. The 14600K is slower in simulation time here.

Frequency Comparison

Single-Core Frequency Comparison (14900K vs. 285K)

We’ll look at maximum single-core frequency under a Cinebench workload first, just to verify that the CPU does what Intel advertises.

The 285K plotted at about 5700MHz during this workload, with the 14900K at 6000MHz. The 14600K ran a maximum of about 5300MHz at any given period, with the new 245K at 5200MHz max. This is what Intel claims as the maximum, so that’s the bare minimum bar to clear and they have at least done that.

P-Core Frequency Comparison

This chart checks the P-core boosting behavior in an all-core workload. We measured the 285K at 5400MHz on average here. The 14900K’s (read our review) most recent result fluctuated around 5100-5200MHz. The 14600K has a flat 5300MHz P-core average from the original review. The 245K plots at 5000MHz for P-core boosting.

Intel Core Ultra 5 245K Gaming Benchmarks

Dragon’s Dogma 2

Dragon’s Dogma 2 is up. In this one, the 245K ran a baseline of 95 FPS AVG. This allowed it a slight lead over the similarly priced 9700X, at 91 FPS. The 245K also technically outperforms the 14600K, although realistically, they’re about equal. 

The cheaper and low-power 5700X3D outperforms the 245K by 8.4% with 103 FPS AVG, a considerable lead for the older platform with the X3D-refreshed part. Moving to the 285K yields a 9.8% uplift, but you should watch our 285K review for all the nuance on that. The price jump is so significant that, especially for gaming-first builds, the 7800X3D likely makes more sense for most people.

APO did nothing here.

F1 24 - 1080p

F1 24 is up next.

The 245K ran at 307 FPS AVG here. That has the 9700X (read our review) 23% ahead at 378 FPS, the 5700X3D 16% ahead at 355 FPS, and the 14600K 4% ahead with a 319 FPS AVG. Intra-architecture, the 285K offers a 12% uplift at 344 FPS AVG. Notables that the 245K outperforms include the 12600K from 2021, at 13.7% ahead.

APO actually does something in this game sometimes: We talked about the "exciting" 1.2% gain in the 14900K. The 245K gained 1.8 FPS AVG from APO. Some of that might even not be margin of error. Groundbreaking stuff. That’s almost 0.6%!...

F1 24 - 1440p

Because APO was so impressive, you can understand why we decided to not bother dedicating time to running it for the 1440p test of F1. We didn’t want it to embarrass everything else…

At 1440p, the chart becomes truncated by the resolution increase. The end result is largely the same lower down the stack, though: The 245K remains outflanked by the 13600K and 14600K above it and encroached upon by the 12600K below it. These CPUs were already mostly CPU-bound and remain so here.

FFXIV Dawntrail - 1080p

Final Fantasy 14: Dawntrail is up now. At 1080p, the 245K ran at 220 FPS AVG. This allowed the 9700X a staggering lead of 41% for a similar price. The 7700X is cheaper than the 245K, at $280 at the time of writing, and produces a 265 FPS AVG result. 

Against other CPUs: The 5700X3D outperforms the 245K by 37%, up at 301 FPS, with the 14600K leading it by 12% with its 248 FPS, and the 285K leading by 22% at 270 FPS. In our 285K review, we noted that Final Fantasy is one of Intel’s worst generational titles between the 14 Series and the Ultra 200 CPUs. As for APO, we were not able to pick up a difference outside of usual margins. The APO on result is 0.77% ahead of APO off, which is functionally noise.

FFXIV Dawntrail - 1440p

At 1440p, the 9700X’s lead is cut down to “only” -- and that “only” is doing some heavy lifting -- 22%. This is because of other limitations from the resolution change, but it shows that even at 1440p, the 245K is disadvantaged.

Baldur’s Gate 3

Baldur’s Gate 3 is up now, tested in the densely populated city center in Act III.

The 245K ran at 92-93 FPS AVG here. As a result, the 9700X improves upon its framerate by 7%, the 5700X3D’s 111 FPS result improves by 20%, and the 14600K is behind the 245K and alongside the 9600X. The 245K leads the 14600K by 6.2% here. This is at least better than what we saw with the 14900K and 285K. The 14900K was also encroaching on memory limitations, as evidenced by the 13700K and 13900K bouncing off of the same limit and the 285K with DDR5-8600 pushing ahead.

As for the 285K’s like-for-like lead over the 245K, that’s 8% at 100 FPS AVG.

Stellaris Simulation Time

Stellaris is up next for something different. This one gives us simulation time in seconds rather than framerate per second.

The 245K required 34.9 seconds to complete the simulation, with APO doing nothing for simulation time. The result has the 245K just ahead of the 5800X3D, behind the 7700 non-X, and behind the new 9600X. Zen 5 does well in this particular benchmark.

Against the 14600K, the 245K benefits from a simulation time requirement reduction of 4.6%. The 245K also leads the 5700X3D with a simulation time reduction of 8.9%, with the 5700X3D falling back from its reduced frequency. The 285K reduces the time requirement from the 245K to 32.5 seconds, or reducing the time by about 7%.

Rainbow Six Siege

Rainbow Six Siege is next. In the 285K review, we showed the above frametime plot -- which we’ll show again here -- to illustrate some behavior where it appears as if frametime pacing is a little more tuned on the 12-14 Series CPUs than most the others we test.

Here’s the chart for the 245K. The 245K ran at about 476 FPS AVG, allowing the 9700X a staggering 31% lead at 621 FPS. The 5700X3D ran at 526 FPS, or a 10% lead. Next is the 14600K at 517 FPS for a 9% advantage over the 245K. Intra-architecture, the 285K leads the 245K by a noteworthy 22.6% -- though we still don’t recommend it against alternatives mentioned in the 285K review. That’s at least a real gap for the price gap.

Toggling APO didn’t change the 285K or 245K, though that’s not shown on this chart.

Lows remain reduced on some architectures, including Arrow Lake.

Starfield

In Starfield, the 245K ran at 121FPS AVG. The 9700X held a 117 FPS result, within about 3% of the 245K but technically behind. The 5700X3D is also slightly behind, at 118 FPS. Intel’s own 14600K breaks away at 127 FPS, a 5% uplift over the 245K. Finally, the new 285K is about 18% ahead of the 245K -- similar to the gains we saw in Rainbow Six.

Toggling APO had no appreciable impact.

Intel Core Ultra 5 245K Production Benchmarks

Time to move into production tests. Like with the 285K, we’ll keep this short and focused but dense with charts.

Blender

In Blender with CPU rendering, the 245K required 12.6 minutes to complete a render of a single frame from the GN video intro animation. That has it about a minute faster than the 14600K, reducing the time requirement by 7.4%. The 245K is beaten by the 14700K at 9.5 minutes, or a 24% reduction in time required. The 14700K (read our review) is $350 as we write this, which is a 13% increase in price over the 245K. The 285K was a relatively impressive performer in this test as well, with only the 9950X ahead of it. This is one of the few where Arrow Lake does OK in competitive scenarios.

7-Zip Compression

File compression with 7-Zip has the 245K at 122K MIPS, landing it behind the 13600K (watch our review) and 14600K. The 13600K improves upon the 245K by 4.1%. The 245K is just ahead of the 7700X and 9700X. The 7800X3D (watch our review) isn’t far back, but the extra cache doesn’t benefit it in our production suite.

The 285K with DDR5-8600 illustrated that this particular test benefits from better memory, providing a large uplift over the baseline 170K MIPS result that tied the stock setup with the 14700K.

7-Zip Decompression

Decompression doesn’t benefit nearly as much from the memory improvement, but does benefit from the extra threads found on some CPUs. The 14900K-to-285K comparison is an easy example of this, as is the presence of both of AMD’s 16-core SKUs at the top of the chart.

The 245K lands far down the chart, just ahead of the 5800X3D and behind the 5800X and 7800X3D. The 5800X illustrates that the higher frequency is more beneficial to this test than the extra cache on the 5800X3D. The 245K ends up beaten by the 14600K at 137K MIPS, an increase of 16%. The 13600K is similarly advantaged.

The 285K’s 193K MIPS result has it 64% ahead of the 245K, similar to the lead we see from the 14900K over the 14600K.

Ultimately, the 245K is beaten by several CPUs that are direct competitors, including the 7700 non-X (watch our review), the preceding 13700K and 14600K, the 7800X3D, and the 9700X.

Chromium Compile

Chromium code compile is up now. We explained the memory capacity change for high-end parts and the frequency limit on the 9950X in our 285K review.

In this one, the 245K required 134 minutes to complete the compile. That’s tied with the 14600K. The time requirement reduction for the 245K from the 9700X is 10% here, with the latter at 149 minutes. It’s also predictably better than the 7800X3D.

The 13700K (watch our review) reduces the compile time to 105 minutes, or a reduction in time of 21%.

Adobe Premiere

Using the Puget Suite, Adobe Premiere lands at 9824 points for the 245K. That has the new CPU sandwiched by the 7900 and 7900X above and the 13600K and 14600K below, with a more meaningful gap over the 9700X and 7800X3D. The $250 i7-13700K gives the 245K a serious run for its higher cost though, at 10616 points. 

The 285K posts a large uplift over the 245K, at 15% higher in score. This was one of the more favorable tests for Arrow Lake with the 285K. That’s true too of the 245K, but there are better value alternatives (like the 13700K - especially now that Intel claims the microcode issue is fixed).

Adobe Photoshop

Adobe Photoshop is dominated by AMD in the current iteration of the software and Puget test. The 245K and 285K are far down the list, both behind the 13700K, 14700K, and almost everything modern from AMD. It’s not much of a contest here.

Intel Core Ultra 5 245K Conclusion

Here’s how it all breaks down: First off, we think you should probably just wait a week and see what the 9000X3D CPU does to the pricing and market. 

That out of the way: It’s clear that Intel has, in fact, reduced its power consumption. That doesn’t mean, however, that it makes sense to purchase an Arrow Lake CPU. 

Even catching the ATX12V power changes on our ASUS board, the CPU is lower power consumption and higher efficiency, even when it sometimes regresses in performance. Some games saw improved performance and lower power consumption, like Stellaris, which resulted in larger uplift than areas where we saw heavy performance regression, like Final Fantasy XIV: Dawntrail as compared to CPUs like the 14600K.

In production workloads, we saw regression from the 14600K in 7-Zip compression (though it still beats the 9700X there, so that’s impressive) and a few other areas. In Photoshop, the 245K improved on the 14600K technically, but AMD still holds a massive lead in this particular test. Premiere was a cleaner victory for the 245K, improving on the 14600K and reducing power draw.

Gaming had the 245K down around the 7700 levels in Baldur’s Gate 3 and ahead of the 9600X, but getting crushed by the 5700X3D, 5600X3D, and obviously the 7800X3D. Although the 245K’s relative positioning against the 9600X and 14600K flip-flopped, like in F1 where it was remarkably non-competitive, the global constant was that X3D outperformed the 245K. Fortunately for Intel, AMD doesn’t have any modern platform X3D CPUs that are in this price range. Until it replaces the 5700X3D in price with an AM5 part, Intel will at least have an advantage in fully modernized I/O despite disadvantages elsewhere like in raw performance as compared to something like an AM4 platform build, which is also just building into a close-to-EOL ecosystem.
Intel’s strongest argument right now is in the non-gaming workloads, but like the 285K, we just think the 245K has too much fierce competition to make sense at the moment. The efficiency improvement is important, and we talk at length about the power cost savings in the 285K review to calculate it across several years of ownership. It’s good that they improved there. It’s just that this platform (1) isn’t really ready, having already half a dozen major bugs leading into launch, and (2) is embattled on all sides by similarly priced or cheaper options.

---

Table of Contents

• AutoTOC

Intro

We’re reviewing the Intel Core Ultra 285K CPU.
Take a look at the giant spiky ATX12V line for the 285K on the chart below compared to the 14900K ATX12V line. This is going to be a problem for testing power today and efficiency. It’s not as simple as just measuring the EPS12V cables anymore, at least not on ASUS.

Editor's note: This was originally published on October 24, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan

Testing, Editing

Mike Gaglione

Quality Control

Jeremy Clayton

Camera, Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

We had to build a monstrosity to isolate the power consumption to the CPU for Arrow Lake, because ASUS is pulling significant power down the ATX 24-pin connector and not just the EPS12V cables. 

• 

• 

• 

If we didn’t do all of this, we would have measured the 285K as the most efficient CPU possibly ever made, but that’s because the power has simply moved to split across 4 phases for the 24-pin and 18 for the rest. 

If it’s been a while since you checked in, you can think of the 285K as the “i9-15900K,” except Intel stopped that naming. 

Intel’s big claim is that power consumption is halved, so we’ll spend a lot of time validating that today. Let’s get started.

Intel Core Ultra 9 285K Pricing

We’ll start with a price round-up of the current landscape.

CPU Price Comparison | GamersNexusLate October, 2024

	Newegg Price	Amazon Price
Intel 285K (MSRP $590)	$630	N/A
Intel 14900K	$470	$470
Intel 14700K	$375	$350
Intel 13900K	N/A	$415
Intel 13700K	$350	$290
Intel 12900K	$300	$280
Intel 12600KF	$160	$160
AMD 9950X	$600	$710
AMD 9900X	$430	$430
AMD 9700X	$330	$330
AMD 7950X3D	$600	$600
AMD 7950X	$510	$510
AMD 7900X	$400	$400
AMD 7900	$370	$370
AMD 7800X3D	$480	$480

We put this table together a few days before launch, so the exact prices may be different when this story goes live; however, as Chronomancers haven’t been a core D&D class since AD&D 2e in 1995 with TSR’s publication of the Chronomancer handbook, we haven’t had a way to time travel for a couple decades and regretfully can’t review future prices anymore.

The 285K is $630 with the pre-order pricing, or $160 more than the 14900K (read our review). We don’t care about the launch MSRP of prior parts, only what they’re available at today since that’s when people buy things.

The 14700K is $350-$375, the 13700K is $290 in some places, the 12900K is down to $280, and the much weaker 12600KF is now $160. Buying into these dead platforms isn’t a great feeling, though.

AMD’s direct alternatives include the $480 7800X3D with a rumored 9800X3D in a week, so if that’s true, we’ll have more to say soon. The 285K is 31% more money than the 7800X3D. The 9950X is also cheaper than the 285K’s current pre-order price.

The new 285K is flanked by the best gaming CPU with the 7800X3D (watch our review) -- which is also the most efficient CPU we’ve tested recently -- and the 9950X and 7950X for production.

Intel Core Ultra 9 285K Specs & Basics

For the absolute barebones basics, the main thing you need to know is that Arrow Lake and the Core Ultra 200 lineup (so far) move away from monolithic silicon and toward a tile-based approach, including manufacturing from Intel competitor TSMC.

Intel has gotten rid of hyper-threading and is instead moving to just P-cores and E-cores with these CPUs, so the 285K has 8-P cores, 16 E-cores, and 24 threads against the 14900K’s 32 total threads. Frequency is also down, now at 5.7GHz advertised but because the architectures are different, you can’t directly compare the frequency numbers. We’ll review the 245K next, but that one’s at 6 P-cores and 8 E-cores. 

The new platform is LGA 1851 and requires new motherboards. These will not work in LGA 1700 motherboards, and LGA 1700 CPUs will not work in LGA 1851 motherboards. There are also multiple independent loading mechanism options chosen by motherboard manufacturers -- we’ll have a separate video shortly with laser scans and pressure maps of those. More specs can be found in our Arrow Lake announcement coverage.

Not Ready for Launch

First off: Arrow Lake really doesn’t seem ready for launch. Intel told us this about Arrow Lake in a briefing:

“The big change is that every ODM now, without fail, every ODM ships with this [APO] enabled. In the BIOS, the Camarillo device is turned on, there’ll be a yellow bang, the driver will install, and APO will be enabled by default. That’s the out-of-box-experience everybody is going to have. You’d want to mimic what folks are going to see.”

“This is going to be something that is going to be probably just as important as the hardware improvements. [...] All your ODMs, the board vendors, it’s all enabled by default and should auto-install, 13th Gen onward.”

This is important because we don’t test with APO so we had to consider it for this review. 

Intel presented first-party benchmark slides with APO enabled and emphasized with us that not only would it be on by default with Z890, but that it would also be on by default with Z790. Neither was true. We discovered this issue and brought it to Intel only because we weren’t sure if it was our fault, and it wasn’t. The company seemed totally unaware of this failing and, after some back-and-forth, eventually realized it had simply screwed up. As a result of our findings, Intel is moving to publish a list of motherboards that support APO out of the box on by default in BIOS going forward. 

While that’s great for consumers so that it clears up the misleading nature of the original claims, this product is clearly not ready if we’re the ones beta testing such strong statements as “without fail” for a feature that wasn’t present, as pointless as it often is. 

Dynamic Tuning also no longer exists on our board, despite Intel telling us that it should be present and enabled. The option has been renamed to Innovation Platform Framework and was off by default, which meant APO was off.

With the latest microcode and BIOS on Z790 for these reviews, we also found that APO was off by default for those platforms as well.

We consider APO off the default user experience for the product as Intel shipped it to us; however, we tested with both. APO was ultimately irrelevant in our test suite.

This was one of many unforced errors from Intel, with others including BSODs on Windows install due to a driver conflict with NVIDIA devices when the IGP is enabled. This issue did not exist on prior architectures and seems to be a combination of the new IGP claiming the PCIE resources and of NVIDIA’s driver code not having an error handler for 0 PCIE resource availability. You can bypass it by disabling the IGP, but for a lot of people, this will be frustrating.

The platform also has some issues with Easy Anti-Cheat, which Intel will soon be publishing a statement about, but we’ll leave that to Wendell’s coverage, but basically, if you disable security features, you’re able to work around the issue. We just wouldn’t recommend it.

Additionally, Intel sent a statement about power profiles where balanced power on 24H2 results in uncharacteristically bad performance. We test with high performance except where we’re specifically required to use balance in some X3D situations so this didn’t affect us but does represent one of Intel’s launch problems. 

Intel even got its own specs wrong on the slide above where it said the chipset can support up to 32 USB 3.2 devices but actually it’s supposed to be 10. This is all to point out that Arrow Lake is really just not ready for launch.  

Efficiency Testing

Power at 24-Pin & Raw Power Draw (Blender)

Let’s get into the efficiency testing.

• 

• 

This is really important, but we’ll try to keep it short: On at least the ASUS Z890 motherboard we have, the 24-pin is doing disproportionate work. This is likely a board-to-board thing, not Arrow Lake as a whole.

24-Pin & EPS12V Power: 14900K (Baldur’s Gate)

Proving our work: This plot shows the 14900K’s power draw in a known workload spread across multiple rails. You can see that EPS1 and EPS2 power is similar, both at around 70-80W in this lighter weight gaming workload. The 24-pin has several voltages, including 3.3V, 5V, 5VSB, and 12V shown here. 

The ATX12V rail includes things like fans, the CPU, and if we hadn’t isolated it, slot power, among other miscellaneous controllers on motherboards. 

Slot power is around 30-35W, which is why it’s important to isolate it if capturing 24-pin power. 

Slot Power Problems

Here’s why we isolate slot power: This benchmark was for 7-Zip on the 14900K. The GPU does literally nothing in this test except spit out the display. It does not engage in the test. However, occasionally in any test, we measure seemingly random spikes to slot power. This could be for a number of unpredictable reasons, including innocuous ones like Windows background tasks or NVIDIA’s drivers doing something in the background, such as telemetry.

24-Pin & EPS12V Power: 14900K vs. 285K

Here’s the comparison we’ve wanted:

In 7-Zip, ignoring EPS12V, we noticed that the ATX12V power was exceptionally high on the 285K and Z890 Hero combination compared to the 14900K and Z790 Hero. It’s at about 50W here, whereas the 14900K had ATX12V down around 30W.

ATX3V power is comparable on both and within 2W. ATX5V power isn’t easy to call a simple average since it fluctuates so much: The range is 14-30W on the 285K+ASUS combination and similar on the 14900K, with peaks about 3-5W lower than the 285K. We don’t know if any of that power is getting used for regulators that might feed the CPU, but if any of it is, it’s not much. A lot of it is driving other components, like I/O.

The ATX12V line is clearly important though, as not factoring-in that 20-30W difference in this test would mean representing the 285K as artificially efficient because the power is sort of “hidden” in a cable where it’s not typically meaningfully high. ASUS’ reasoning for this is theoretically better power regulation. We don’t know if anyone else is doing this.

24-Pin Power: 7800X3D vs. 285K ATX12V

Finally, as an example of AMD, here’s the 7800X3D and 285K.

The 285K and 14900K ATX12V remains from last time. The 7800X3D was pulling about 12W ATX12V fixed during this test and didn’t seem to fluctuate based on load at all. It appears fairly isolated. Typically, this gap of 10W comes out in the wash when you’re talking a 280W 14900K versus an 80W 7800X3D, so it’s just part of the usual margins. In this scenario, however, the difference against the 285K really starts to show.

24-Pin Power: 7800X3D vs. 285K ATX3V

Here’s ATX3V. The 7800X3D platform is marginally higher here -- we’re using an ASUS board for this also. This is part of why 5V normally comes out in the wash: Some rails are a few Watts higher, some a few Watts lower. There is always going to be error.

24-Pin Power: 7800X3D vs. 285K ATX5V

ATX5V is about the same between them all. There is possibly spikier behavior on the 285K from the I/O, but that’s about it. 

Software Readings

All this is done with hardware. We also can’t trust various software readings because they can be manipulated heavily by software or motherboards. Years ago, there was an issue where AMD motherboards were underreporting the power in a way that allowed boards to pull more or less power while hiding it in software, thus manipulating the report. We also spoke with Der8auer about his experiences, and he noticed that CPU Package Power is calculated based on VID, so if anything is wrong with voltage, it will manifest in erroneous software readings.

• 

• 

• 

And that leads us back to our monster we’ve built. We are fairly confident in the results, but we have to caveat that this is a very opaque platform currently and it’s not fully clear how all the power is routed. For the most part, it should be proportional to the VRM split of 18:4 of the 22 phases total, so we were able to use that as a guidepost.

Finally: You might be asking “who measures the measurers,” and the answer is... us. Working with Elmor, we hooked up other current monitoring devices in series to monitor the current monitoring devices, and we also clamped the cables to monitor the monitoring of the monitoring devices while monitoring the software of the monitoring-monitoring devices and the CPU-monitoring software monitor.

Once that was all done and we had tested against calibrated Fluke meters with known current loads, we proceeded with the most accurate combination of monitoring.

Let’s get into the efficiency testing.

Efficiency: 7-Zip Decompression

7-Zip decompression efficiency is measured in MIPS/W, or millions of instructions per joule.

In this test, the 285K ended up in the lower-third of the chart. Intel’s 14900K pulled 273W in this test, with the 285K now at 162W for ATX12V and EPS12V without slot power, or 175W with ATX12V, ATX5V, and EPS12V. We think the 162W number is more accurate, but want to transparently present both since the power split is still not fully clear.

Because performance is lower and the formula relies on both, the efficiency increase isn’t as impressive as the power drop. The 285K runs with an efficiency of 1101 MIPS/W decompression with the reading that includes 5V, or an improvement of 29% over the 14900K. The reading without 5V is 40% improved over the 14900K, which we think is the more accurate one. Whether 30% or 40% is closer, the point is that it’s a big uplift. AMD still dominates here, holding the entire top-third of the chart, though. The 7950X in ECO mode sets a seemingly untouchable score at 1936 MIPS/W. The 7800X3D, which is much slower than the 285K in 7-Zip decompression, is more efficient thanks to its 70W power draw.

Efficiency: 7-Zip Compression

Here’s the 7-Zip compression result. Power consumption is measured across compression and decompression in the same test suite, so those figures are unchanged.

The 285K scored 1051 MIPS/W with ATX12V and EPS12V, an improvement over the 14900K’s 672 MIPS/W of 56%. This uplift versus decompression comes from the higher performance in 7-Zip compression relative to its scaling in decompression.

AMD’s 7800X3D is at the top here. Again, it’s not the best performer, but its 70W power is impressive and allows it to remain the true most-efficient CPU we’ve tested lately. 

Efficiency: Baldur’s Gate 3

Baldur’s Gate 3 could get more complicated, but we’ve isolated out the PCIe slot power by using an interposer to power the video card instead.

The entire X3D lineup continues its domination in this game, and most others. The high framerate and low power land the 7800X3D at 2.3 FPS/W, as it’s pulling only 55W during this gaming workload. The 5700X3D is about the same, with the 5800X3D (watch our review) slightly less efficient. Intel’s new 285K isn’t remotely close to these numbers.

The 285K with ATX12V and EPS12V ran at 1.1 FPS/W, so the 7800X3D is creating over 2x as many frames per Watt of power. The 14900K ran at 0.7 FPS/W here. Looking at just Intel, the 1.1 result of the 285K has at least improved substantially over the 14900K, with an uplift of 57%; if we include the 5V power, since we’re not certain if any of it is being used for the CPU, that would be a 29% uplift. 

Efficiency: Starfield

Starfield is a heavier workload from gaming.

In this one, the 7800X3D is again the most efficient. Its result was 2.1 FPS/W, holding a large lead over the next closest CPUs. The next closest CPU is the 5700X3D, also at about 70W. The 5800X3D, 9700X, and 7700 non-X follow this. We eventually hit the 3700X (read our revisit), then the 285K.

With the ATX12V and EPS12V measurements and without PCIe slot, we end up at about 147W for a 1.0 FPS/W rating with the usual rounding.

The 14900K ran at 0.7 FPS/W here, so Intel has improved by 43%. Its Ultra 9 is now about where the 14600K was previously. Measuring 5V as well had it at 172W, or 0.8 FPS/W. That would instead be an improvement of 14%. As before, we believe the 1.0 figure is closer to reality, but it likely falls in between.

Efficiency: Stellaris

The next game is Stellaris, where we measured simulation time in seconds and power. In order to make this as easy to follow as possible, we’ve done some simple arithmetic to convert this into “simulations per watt-hour,” meaning that a bigger number is better despite the base metric result being lower-is-better.

The 7800X3D has an impressive score here, at 2.7 simulations per watt-hour. The next closest is down at 2.4 with the 5700X3D (read our review).

The 285K ran 1.5 simulations per Watt-hour. That means for each watt-hour, the 7800X3D is able to complete 1.2 more simulations, or about an 80% increase. 

The 14900K completed 0.9 simulations per Watt-hour. That means the 285K’s best entry is an improvement of 67% in efficiency. Remember: Most of Intel’s presented numbers were only for power reductions, not necessarily for efficiency. The 285K’s simulation performance is better than the 14900K’s in this benchmark, so its combination of higher performance and lower power (which was reduced by 46.6W from the 14900K) yields this huge increase in efficiency. 

Efficiency: Final Fantasy XIV

Finally, Final Fantasy 14 for efficiency. In this one, the 7800X3D led with an impressive 42.5W power draw while spewing hundreds of frames per second, allowing it a result of 8.3 FPS/W. That’s great. The 5700X3D keeps its second-place spot at 7.7. The 14900K ran at 3.1 FPS/W, so the 285K is 32% improved by these numbers. This is one where the 285K was regressive in performance against the 14900K -- and by sort of a lot, but we’ll come back to that. The result is reduced efficiency gains.

Frequency Testing

Single-Core Frequency Comparison (14900K vs. 285K)

Next, we’ll validate the frequency behavior to establish how the CPUs behave out of the box.

This chart plots the highest single core frequency per interval during a single-threaded Cinebench workload.

With the 285K, the max single-core frequency is 5700MHz during testing, with frequent drops to 5400MHz. These drops are abnormal on Intel during this test.

The 14900K with the latest microcode held 6000MHz. These are different architectures, so it’s not as simple as stating that higher is better; however, it helps at least illustrate where some of the performance losses are coming from.

The original launch microcode had it at the same frequency.

P-Core Frequency Comparison

This chart looks at P-core frequency averages during a Blender workload. 

The 285K maintains an average P-core frequency of 5400MHz in this testing. It’s relatively flat. That has it higher than the 2024 ASUS test entry, which is what we used for our review comparison today, at around 5100-5220MHz. The 2023 launch entry was higher, at 5280-5400MHz on average. The microcode changes may have affected this since launch. The MSI 0x11D microcode entry had the CPU at around 5000MHz from the same window as the ASUS 2023 entry. The most up-to-date frequency entry sits between the two flanks of early 14900K tests. 

All-Core Frequency Comparison

Finally, this chart shows the all-core averages in the same test. The 285K averaged 4867MHz when factoring-in the E-cores, with the modern 14900K just below 4500MHz and flanked by the two older entries.

Intel Core Ultra 9 285K Gaming Benchmarks

Now we’re going to get into gaming benchmarks. We have a lot of 2023 and 2024 games in our test suite along with mainstays from the past.

Dragon’s Dogma 2

Dragon’s Dogma 2 is up now. This is one of the 2024 titles we added to our suite last time. It’s had updates that affect performance since our last round of benchmarks. APO doesn’t do anything in this benchmark and isn’t supported. We tested with it on and off for both the 14900K and 285K and its performance was the same, so we removed the redundant entries here since they have no impact.

The 285K landed at about 104 FPS AVG, with lows at 64 FPS and 51 FPS. 

This allows the AMD 7800X3D at $480 to lead the $590 285K by 6.1%, with the 14900K at 5% ahead. The lows between these 3 CPUs are functionally identical. The 14700K and 13700K also lead the 285K here, with the 14700K (read our review) far cheaper at $350. The 5700X3D and 5600X3D (read our review) are effectively tied, with the 5700X3D at $230, or sometimes down closer to $200.

F1 24 - 1080p

In F1 24, which, for hopefully obvious reasons, is a 2024 title, we end up with this set of results when fully retested in 24H2 and with new microcode and AGESA.

The 7800X3D sets a high ceiling and establishes a 28% lead over the more expensive and higher power-consuming 285K, at 438 FPS to 344 FPS AVG. The 14900K is also advantaged, this time by 12% with a 385 FPS result. The lows for the 285K were 186 FPS 1%, which is a worse entry than the 14900K’s 250 result. The frametime pacing appears better, in-step with the average, on both the 14900K and 7800X3D.

Looking now at the APO results, toggling APO does appear to do something in this game: The 285K gained 1.3% with APO on. How very exciting -- we can hardly contain ourselves. You’ll surely notice the extra 3-4 frames per second on top of the other 343 before them. We’re so glad we spent half a day troubleshooting Intel’s fumbled pre-launch settings to gain those frames back... Whatever would we have done without the extra 0.0369 ms reduction in frame time?...

The 14900K also gained about the same amount. It’s irrelevant on both.

Using DDR5-8600 on the 285K with Gear 2 and blasted VDIMM boosted it to 359 FPS AVG, a 4.5% like-for-like lead over the 344 result. This is almost enough to get the 285K tied with the two-generation-old 13700K (watch our review) from 2022, but not quite. It does, however, tie it with the AMD’s 2020 architecture found in the cache-bolstered 5700X3D with its lower power.

F1 24 - 1440p

1440p results are mostly uninteresting: The top is truncated heavily by the increase in resolution. Even still, the new Ultra 9 ends up in the middle. It’s fitting that the Core Ultra 9 is in the middle of a cluster.

FFXIV Dawntrail - 1080p

Final Fantasy 14: Dawntrail is up now. This is another 2024 addition to our testing.

The 7800X3D again establishes the ceiling. It leads the 285K by 31% in this game, at 353 FPS to 270 FPS AVG. The 14900K’s 310 FPS AVG establishes a 15% lead over the 285K, marking one of the largest declines of Intel’s performance in our gaming suite.

The 5800X3D and 5600X3D are also at the top here, showing that Dawntrail just generally really likes cache, at least on these CPUs. The 5700X3D is a little lower down since its frequency is 300MHz below that of the 5600X3D, so these results make sense.

The 285K ends up below the 14700K and 13700K, with APO doing nothing on the 285K. It’s within error. Final Fantasy 14 is officially supported by APO though and is on the games list, and we do see a slight uplift on the 14900K of 1%, basically error. Switching to DDR5-8600 on the 285K provided an uplift of 4%, with it still below the 13700K after that. Of course, the same memory treatment would lift others up also.

FFXIV Dawntrail - 1440p

At 1440p, the limited ceiling is shuffling the stack as a result of the framerate bouncing off of other limits. We saw a 1.9% uplift from APO on the 14900K. The 285K, again, did not benefit from APO, with both entries at 255 FPS AVG +/- 1FPS.

Baldur’s Gate 3

Baldur’s Gate 3 is up now, a 2023 title that we added to our permanent suite this year. We test in Act III in a densely populated city area for a heavy CPU load.

The 7800X3D establishes a 26% lead over the 285K, at 126 FPS to 100 FPS AVG. The 14900K leads the 285K by 4.5%, at about 105 FPS AVG.

The 5800X3D, 5700X3D, and 5600X3D also lead the 285K, as do the 13900K, 13700K and 14700K.

Toggling APO did nothing here and was within variance for both the 14900K and 285K. Swapping to DDR5-8600 memory boosted the 285K by 7.6%, allowing it to pass everything except the X3D CPUs (this also perfectly aligns with why X3D does so well here). Of course, shoving better memory into the 14900K or 9000-series AMD CPUs would also boost them fairly proportionally.

Stellaris Simulation Time

Stellaris is up next. 

APO has no impact on this game. 

The 285K required 32.5 seconds for the simulation, finally putting it ahead of the 14900K. The 285K requires about 1-2% less time than the 14700K and 14900K CPUs.

The 7800X3D outranks the 285K again and requires about 3.7% less time for the work. The 9700X remains the fastest here, which is consistent with our last round of data.

Boosting memory helps in this game, so the 285K benefited from a 4.3% reduction in time to complete the simulation.

Rainbow Six Siege

Rainbow Six Siege is up. In this one, the 7800X3D leads by a much smaller amount, at only 6.6% for 622 vs. 584 FPS AVG when APO is in its default off position that this board shipped in. The 14900K was tied. Enabling APO did nothing on the 285K. The result was within error when considering we’re almost at 600 FPS. Enabling it boosted the 14900K by 1.7%, and this is with Intel’s sh*tty Microsoft Store App confirming that it’s enabled for both CPUs. This boost is reduced for the 14900K versus APO’s launch because Rainbow Six has updated the code that was causing problems.

Going to DDR5-8600 increased performance over the 584 result by 2.6%.

The 285K is at least better than the 14700K in this one, with an uplift of 2%. It’s also about 1.8% ahead of the 7700 non-X, which has been $280-$290 lately when you can still find it. 

As a general note, this game has some serious 0.1% low consistency issues. We noticed that the 12-14 Series CPUs have significantly higher 0.1% lows than AMD’s options and than the 285K alike. It’s easier to look at a frametime plot.

Rainbow Six Siege Frametimes

Here it is. Because Intel noted that Rainbow Six had specifically tuned to reduce reliance on APO, and because of the low advantage of Alder Lake onward, we’re assuming that Rainbow Six Siege has specific and manual tuning for the Alder Lake architecture and its more recent derivatives.

This frametime plot shows the 14900K is overall highly consistent frame-to-frame, with most of the intervals deviating at most by 2ms. There are a few spikes, but only one notable spike to 6ms -- which is a 4ms deviation, and thus wouldn’t be noticed by the vast majority of users.

Adding the 285K, we see several spikes to 10-11ms. By themselves, these frametimes aren’t bad. 60 FPS would be 16.667ms, so this isn’t a huge stall. But it is a big change from Raptor Lake and is objectively worse.

Starfield

In Starfield, another 2023 game, the 7800X3D maintains a 2% advantage over the 285K, at 145.4 to 142.5 FPS AVG. We’re bound elsewhere -- and that elsewhere is obvious: The DDR5-8600 result jumps to the top of the chart, illustrating that at least some of our limit is memory. As a reminder, applying the memory tuning treatment to everything would boost all numbers, so looking at baseline will give you an idea for their headroom.

The 14900K actually falls behind for once, at a slight loss to the 285K. We think this corresponds with the cache changes, based on the memory performance we just saw.

APO, once again, does nothing between the two main 285K entries.

Total Warhammer 3

Total War: Warhammer 3 is up now. The 7800X3D led the 285K by 7.2%, with the 14900K leading by 2.8%. Total Warhammer 3 is on the APO list, so it’s supported. On the 14900K, we saw an uplift of about 1.2%. On the 285K, we measured an improvement of 0.6% -- basically error.

Lows in this game have been historically awful on Intel’s i9 CPUs, which we previously proved is due to scheduling challenges with the thread count. The 285K at least seems to somewhat improve here, possibly by the reduction in threads.

Intel Core Ultra 9 285K Production Benchmarks

We’re getting into production benchmarks now. 

Blender

Blender is up first. We’ve updated it, so it’s not comparable to prior results.

The 285K is our second-highest performer on the charts for desktop-class CPUs. The 9950X (read our review) benefits from a render time requirement reduction of 5.6%, while the 285K reduces render time from the 14900K by 16.5%. That’s finally some uplift.

These types of tests are also where AMD’s 7800X3D and 5800X3D show their limits as 8-core CPUs, both falling far down the charts. That’s a known quantity. If you’re only gaming, they make the most sense. If you mix in a good amount of heavily threaded software for work, it may make sense to buy something else on this list.

7-Zip Compression

In file compression with 7-Zip, the 285K completed 170K MIPS and roughly tied the i7-14700K. The 14900K holds an advantage of 8%. The 9950X and 7950X (watch our review) outperform the 285K significantly, with the 9900X just behind. Memory seemed to really help in this test, pushing the 285K up by a gargantuan 18.7%. That’s a huge uplift. 

7-Zip Decompression

Decompression doesn’t benefit as much, posting a 4% improvement. The stack also shuffles, with the 9900X and 7900X both surpassing the 285K here. The memory subsystem benefits don’t translate as much as they did in compression, both stock and with the better RAM.

The 14900K leads the 285K by a staggering 21% in this benchmark, benefitting from its extra threads. The 9950X establishes an impressive ceiling, at 42% ahead of the 285K.

Chromium Compile

Next is our Chromium code compile test. For this test, we double the RAM capacity on some CPUs when they page-out or otherwise fail to complete the compile. This only occurs with the highest-performing few CPUs. This test has been updated since last round, so load on CPUs has slightly changed.

In the situation of the 9950X, there is one disclaimer here, we were not able to get it stable at DDR5-6000 with 64GB when running our usual tighter timings. We experimented briefly with reducing the timings, higher voltages, and altered infinity fabric, but ultimately had to call it for now and run it at DDR5-5600 to accommodate 64GB. That makes the 9950X an imperfect comparison. With that limitation disclosed due to stability issues, the 9950X and 285K are roughly equal. That may change if we can rerun the 9950X and find stability.

For direct comparisons, the 285K leads the 14900K by only about a 3-minute time reduction or a total compile time requirement reduction of 3.8%. As usual, the 8-core X3D parts are lower down the stack, like the 7800X3D and 5800X3D. 

Adobe Premiere

Now for the Adobe software, tested with the Puget Suite. 

Premiere has the 285K as the new chart-topper, at 11,336 points. That’s how this whole review should have been, especially with a power reduction. Unfortunately, it wasn’t this way.

The lead over the 14900K is small, at 2.5% uplift in aggregate. The 9950X is AMD’s top CPU here, at 10,914 points. The 285K has a 3.9% lead.

Adobe Photoshop

In Adobe Photoshop, AMD has a clean sweep of the entire top half of the charts. Those of you who’ve been watching us for at least 5 years will remember an era where it was the opposite. The first Intel entry doesn’t appear until below the 7600X, and that’s the 14900K. 

The 9950X leads the 285K by 22% here and the 14900K leads it by 3.5%.

Intel Core Ultra 9 285K Conclusion

Simplifying all of this, our current conclusion is this:

The 285K does not have a particularly strong place in any one market. 

The 7800X3D is far superior in both gaming and efficiency. The 285K can hardly make sense against even the 5700X3D, and AMD’s 9800X3D is days away at this point. For non-gaming, the CPU is more interesting but we still don’t think it is broadly justifiable at the current CPU and platform costs. 

Even Intel’s prior-gen parts are far cheaper right now, like the 13700K and 14700K, which theoretically have been fixed with microcode. If you were to argue that the inefficiency of the prior gen parts counters some of the savings in the form of lower power, it’s just not a strong argument. The 14700K is $350 now, so it’s $280 cheaper than a 285K. 

Power Value - Cheap Power

If your power costs $0.10/kWh like ours, then you’d need to play the highest CPU load games for 8 hours a day for an entire year in order to gain $19 of value in power bill reductions for the 285K against the 14700K. At $0.15, the shut-in gaming levels would give the 285K a savings of $30 over one year of literally playing Starfield or Cyberpunk as a full-time job. At the peak of degeneracy that we all strive to achieve, you’d have to play games at this pace for 10 years to have spent more on power than the cost savings of buying the cheaper CPU (it’s the delta, not total cost). Maybe throw in a cheaper cooler as well since power is down and call it 8 years.

Power Value - Moderate Power

Even at $0.30/kWh, the $280 price savings from the 14700K would require 5 years to wipe-out with the power bill while playing Cyberpunk 8 hours per day for 5 years, or in other words, 14,560 hours of Cyberpunk.

Power Value - Expensive Power

Let’s just say you pay $0.50/kWh. In one standard 3-year upgrade cycle, you can save the amount of money in power you’d save buying the 14700K -- which we still don’t recommend for gaming -- rather than the 285K. That’d be a savings of $92.60 per year (at $307.7 - $215.1).

Or, you go with the 7800X3D and get higher framerates and still a power cost reduction, even at these high electricity prices, of $114 per year at these crazy gaming hours.

And ultimately, no matter what argument is made for power savings, like room temperature or something, you can still point to AMD as an answer.

The reduction in power consumption is important and does mean cheaper cooling is possible, so that’s good. Intel is objectively getting more efficient, but regressive performance in games is hard to deal with. The production performance mostly improved at lower thread count, and that actually is great. The problem is that at $630, there are very few cases where the 285K makes sense right now.

Our approach to reviews is very consumer-oriented and value-oriented. That means for us, we need to see a strong value argument to recommend a part. Even forgetting completely about power as a factor and looking only at performance for the price, it just doesn’t make sense to us despite being an important building block.

The best news out of the 285K is that it might be a fresh start for Intel and that it is genuinely a lot easier to hit higher memory clocks than before, and that benefits it in tests like our 7-Zip benchmark.

But we just think there are better CPUs out there at lower prices.

---

Table of Contents

• AutoTOC

Intro

The Lian Li Lancool 207 is the new best-performing case on our thermal charts -- and it’s $80. The case shoves the power supply to the front to accommodate 2 shroud-top fans that are closer to the floor of the case, providing intake directly into the GPU. The front of the case has a shaped front inlet for 2x 140mm fans cooling the rest, and the back is entirely ventilation to allow exhaust. 

From a usability perspective, it struggles in some key and basic areas: Cable management is challenging. Closing the right side panel isn’t always easy, especially if you have more cables, thicker gauge wire and sleeving, or accessories that add a ton of cabling. The side panel is tool-less and the back compartment is shallow, with the PSU’s orientation complicating power supply length and cable bends.

Editor's note: This was originally published on October 4, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing, Writing

Patrick Lathan

Testing

Mike Gaglione

Camera, Video Editing

Vitalii Makhnovets

Writing, Web Editing

Jimmy Thang

---

• 

• 

• 

• 

• 

• 

• 

The Lancool 207 is an exercise in compromise, but some of those compromises didn’t need to exist.

We’re reviewing the Lancool 207 with our new case testing methodology. Previous Lancool cases were similarly priced and also heavily competitive for thermals. Repeating that price in 2024, with the gradual climb of pricing for everything, shows a highly aggressive Lian Li.

The new Lancool 207’s $80 MSRP comes close to the (originally) $70 Lancool 215 that impressed us back in 2020.

• 

• 

• 

• 

We saw the 207 at Computex earlier this year. We've spotted a few changes to the case since that Computex showing: the weird removable plate at the bottom of the front panel is gone, the bottom intake fans are conventional rather than reverse-blade, and the side panel vent is smaller. None of these changes are meaningful from a consumer perspective. The removable plate never had an assigned function other than to possibly be “DLC” add-ons, the bottom intake fans look basically the same, and the functional area of the side vent hasn't changed.

Lian Li Lancool 207 Specs

Dimensions (DxWxH)	455.6mm x 219mm x 456mm
Color	Black / White
Motherboard Support	ATX (Width = 240mm) / Micro-ATX / Mini-ITX
Expansion Slot	7
Storage	2 x 3.5” HDD or 2.5″ SSD
GPU Length Clearance	410 mm
CPU Cooler Height Clearance	167 mm
PSU Support	ATX (Under 160mm)
Radiator Support	Top: 240mm / 280mm/ 360mm
Fan Support	Front: 140mm x 2 (pre-installed)Top: 120mm x 3 or 140mm x 2Bottom: 120mm x 2 (pre-installed)Rear: 120mm x 1
Dust Filter	Bottom x 1
I/O Ports	USB 3.0 x 2USB 3.1 Type C x 1Audio x 1Power Button x 1
MSRP	Black - US$79.99White - US$84.99

Specs copied from manufacturer materials, please read review for our own measurements and opinions

We like the Lancool 207, and we have a lot of good things to say about it, but we consider the power supply situation to be plagued with oversights. Overall, we’re positive -- so we’ll start with the basics and positives, then spend some serious time talking about the unfortunate cable management and power supply situation.

• 

• 

• 

• 

For the basics: The case pulls apart in 3 primary panels, including the typical left and right sides and a lower quarter-panel that is perforated for intake into what would traditionally be the PSU shroud. Rather than elevating the floor and pulling through that, Lian Li pulls through the rear and sides. This allows the floor to be available for hard drives and SSDs. Even with a 3.5” drive installed, there’s still plenty of room between it and the sunken fans to pull air through.

The panels are also exceptionally sturdy for a case of its price. The top panel snaps into place with heavy-duty studs and uses folded-over columns of steel to reinforce it. This avoids that cheap stamped steel wobble we see on some sub-$100 cases -- and on Corsair’s $200+ 6500 cases. 

This also appears to be a direct response to issues we found with the Lancool 216, where there was around a half-a-centimeter depression in the top panel. They’ve fixed that issue here.

• 

• 

• 

The front panel also uses a folded-over steel, which doubles as a catch for the magnets at the top of the easily removable filter. This should make it trivial to clean. We’ve always been advocates for ultra-fine mesh panels rather than dust filters, as these serve the same purpose as a dust filter, but are much better for airflow. 

The entire front panel can be removed, but there's no reason to do so. If you do it anyway, the I/O cables have been screwed down to the corner of the panel for strain relief. 

• 

• 

The porosity on the back of the case is also at insane levels for mostly good reasons: We typically see much smaller holes with thicker steel, resulting in a lot less breathability. Because this case is fully positive pressure, we don’t need to worry much about dust ingress from the rear of the case. It’ll get pushed out naturally. This shows an attention to detail that illustrates Lian Li actually understands why you’d make decisions like denser holes for intake and wider for exhaust. 

There is one missed point, though: The very bottom of the rear of the case will serve as intake, not exhaust, as it is separately chambered for the shroud-top fans. Lian Li should be filtering these or using ultra-fine mesh.

Overall, Lian Li has come a long way since we criticized some of its older cases for airflow and is really starting to show a mastery of thermal design with the small details.

The only fit-and-finish issue we noticed were some imperfections and scratches in the infinity mirror fan hubs, but these aren't visible while the fans are spinning.

• 

• 

The motherboard tray is offset from the standard, creating an ATX-sized indent. This means that the case is absolutely not compatible with any so-called “E-ATX” boards and that there's only about 1cm of clearance behind the motherboard tray, but it also means that compatibility with motherboard edge connectors is excellent. Lian Li also claims that the offset better aligns the GPU with the bottom intake fans, but cable routing is the more tangible benefit.

• 

• 

• 

Front I/O is extremely basic: 2x USB 3.0 Type-A ports, 1x USB 3.1 Type-C port, 1x audio jack, and the power button. The flat, flexible Type-C cable in particular is the least annoying one we've ever worked with. The case also has a convenient all-in-one front panel connector, but we'd actually rather have the old-fashioned two-pin connector here, since it's just for the power button. There is no reset button. Typically, these would combine power, reset, and power LED.

• 

• 

The case has two vibration-damped drive mounts in the floor for a single 3.5” or 2.5” drive each. The drive mounting holes are suboptimal: SATA cables have to be bent at an uncomfortable angle to fit and the 2.5" mounting holes could easily have been moved to alleviate this. 

Each pair of fans has 1x four-pin fan connector and 1x three-pin ARGB connector, and the two fans within each pair are linked by proprietary connectors. We'll begrudgingly accept this as a compromise since it cuts the number of fan cables in half. The downside is that it's impractical to replace or reconfigure individual fans, contributing to a larger downside of having to go through Lian Li for replacements if one breaks. The front and bottom mounts aren't compatible with any sizes other than the ones that are preinstalled (140mm front, 120mm bottom). On top of that, the front mount is built around Lian Li's 30mm-thick fans, and all four stock fans are attached with radiator screws. If you still want to replace any, Lian Li includes 16 additional screws sized for 25mm-thick fans. All of this is non-standard and will be unfamiliar to experienced builders. It’s workable, but it will impose atypical limitations on fan compatibility if you want to change them. For instance, you’d typically have 120mm options where 140mm fans install. Here, that’s not the case. The upside is that there’s one less set of rails obstructing intake at the cost of flexibility, so Lian Li is going for the focused airflow approach.

Lian Li describes the 207 as a "compact M-ATX-sized case with ATX compatibility," which is only true by Lian Li standards. The Lancool 207 is the size of a normal mid-tower; it has almost exactly the same dimensions as a Fractal North. In fact, it’s really not much different than most of the cases in the comparison image above. The SilverStone 515XR is an example of a truly compact ATX case, or the Easy-Bake Cooler Master Q500L.

Now to the power supply:

• 

• 

• 

• 

• 

The 207 has an unusual layout with the PSU rotated 90 degrees and placed at the front of the case rather than the rear. This means that there's a 160mm size limit for PSUs, but even our normal 150mm unit was cramped. Without dropping out of the ATX spec, 150mm is about the smallest you’ll typically get.

Because of the rotation, cables are pointed at the tool-less side panel and have to bend sharply to route. Using flat cables helps, but this is a situation where you’ll need to consider how the cables in your power supply are sleeved. We’d also strongly recommend sticking to 150mm rather than using the maximum 160mm under the spec. We’d also strongly recommend modular power supplies for this case.

The PSU shroud is effectively a fan mount, and even if you could find a way to stick extra cables under it, it'd interfere with the intake fans and the entire point of the design. 

All of this would only be an inconvenience, except that the steel side panel is tooless. If you're using bulkier cables with individually braided sleeves like in our example, it is possible to choose power supply and cable combinations that make it nearly impossible to snugly close the side panel. It’s possible, and we did it with our bulky sleeved cables to prove that, but this took a good amount of time to manage and flatten.

We’d strongly recommend flat cables like the ones shown in the 207 at Computex. The more accessories and PCIe power cables you run, the more difficulty you’ll have managing the cables in this particular case. 

Screws for the side panel would have helped this by allowing more tolerance for cable bulge. 

The manual illustrates installing a Lian Li EDGE PSU, which looks like it could help; however, the 207 is $80, and the cheapest EDGE PSUs cost at least $130 to $140. It’s a mismatch of the most likely buyer for the case if Lian Li expects most to use these power supplies.

The illustration also shows the EDGE PSU being installed fan-side-up, which isn't an officially supported option for other power supplies. The usual arrangement of mounting holes requires installing standard ATX PSUs fan-down in the 207, with only 1cm of clearance above the table.

In another major concern, in Lian Li’s quest to move the power supply, it has overlooked both cabling and cooling considerations. With the bottom filter installed, which we’ll come back to, the power supply is given mere millimeters to breathe with a bottom-oriented fan. We don’t currently run power supply thermals during our testing; however, this will cause the power supply to heat up, objectively, and it may result in the PSU fan running at a higher RPM (if it is a power supply that bases fan speed on any internal temperatures).

Removing the filter functionally doubles its breathing room, but ultimately, we think this is a major design shortcoming. Lian Li’s pursuit of minimizing the case size has limited its area for power supply intake. This is making us consider running a thermocouple wire to our power supply for future case reviews.

• 

• 

The PSU filter is another small design fault: It’s easiest to lift the case to remove the filter, which isn’t a dealbreaker, but could be fixed with a more accessible tab. 

• 

• 

On the cable management side on the back of the case, the attachment points for the built-in velcro straps are angled so that the straps curl over as they're inserted, which is a much bigger quality of life improvement than it sounds like. The manual describes intended cable paths, most of which are obvious, but there are specific routes for 24-pin STRIMER cables and for CPU 8-pin cables. The 8-pins are held in by plastic hooks, but the hooks might work better if they were flipped. There isn't a suggested path for GPU power cables in the manual. This is something we would have liked to see for new builders.

As for vertical GPUs, the expansion slots are bridgeless, so other vertical GPU add-in kits may work, but Lian Li specifically markets its universal 4-slot bracket. As demonstrated at Computex, using this vertical bracket will block the bottom intake fans, and we don't recommend it. 

It may technically be possible to get a 280mm radiator installed in the front of the case, but it'd need to have exactly the right hole placement. Lian Li doesn't claim any formal radiator support anywhere except the top of the case, and we agree. 120mm-wide rads fit more easily than 140mm-wide ones, which may make the CPU power cutouts difficult to reach.

Lian Li Lancool 207 Thermals

As we've covered, it's inconvenient and pointless to replace the 207's stock fans. Still, we had a couple of alternate configurations we wanted to test that didn't involve altering the fans. We also did a pass with the small mesh section on the steel side panel taped off. The bottom intake fans have a larger section of mesh available on the other side of the case, as well as a wide-open vent at the rear, so we suspected that taping off this little vent wouldn't matter much.

The 207 is one of the cheapest cases on our charts currently. Out of our recent reviews, it lands between the $100 SilverStone 514X and the $68 515XR. We also retested the older Lancool 216, which is still available for around $100. In terms of fan layout, the 207 is most similar to the more expensive Antec Flux Pro.

We only ran basic tests in "standard mode" on the 216. There's also an "air cooling mode," but in our original review, we found that this helped GPU thermals while simultaneously hurting CPU thermals with our bench hardware. Check our Lancool 216 review for more detail on alternate configurations, including an external fan bracket attached to the back of the case.

GPU Full Load Thermals - Noise-Normalized

The 207 is a loud case with all fans at full speed, but we'll start with noise normalized results. For this testing, all case fans are lowered to reach a target noise level of 27 dBA measured at 1 meter in our hemi-anechoic chamber, with CPU and GPU fan speeds controlled and constant. 

We’ll start with noise-normalized GPU thermals since those bottom fans should help there.

The GPU temperature averaged 42 degrees above ambient, 47 on the memory, and 55 hotspot. That's one degree better than the 216 in each category, but still close. Impressively, the Fractal Torrent remains the leader in this particular test. That’s not always true, but it is here. It has its own bottom intake fans that feed the GPU directly, which help. 

We haven't always seen a benefit from shroud-top intake fans, but cases like the 207 and the Flux Pro that commit to building the case around that concept do show good results. If anything, it's a testament to the 216's cooling that it comes so close to the 207 without any shroud fans.

The Antec Flux Pro's average GPU temperature was about one degree cooler than the 207's.

There's no contest versus the 514X and 515XR, which fall on the complete opposite end of the chart. The 207 is in a different class in many ways, but Lian Li has priced it so aggressively that we have to make the comparison; it undercuts the 514X and is only $12 over the 515XR.

Fractal’s Pop Air is another price-competitive case, but lands at the bottom of this chart. It’s still a good case, and when it’s one sale, it is impressively cheap, but the 207 is thermally superior.

CPU Full Load Thermals - Noise-Normalized

Here’s the CPU noise-normalized chart: Instantly, the Lancool 207 is tied for chart leader in CPU thermals. The average all-core CPU temperature was 41 degrees Celsius above ambient and 44 degrees on just the P-cores. The 216's numbers round differently, but its results were near-identical to the 207's and within error. The 216 was one of our all-time best performers for noise-normalized CPU cooling when we originally reviewed it, so that's good news for the 207. The 207 technically tops our updated chart within error of the 216, but the Antec Flux Pro is also within a one-degree margin. The Flux Pro is at least $180, though.

In comparison to the more closely price-matched 514X at 48 degrees all-core and 515XR at 53 degrees all-core, the 207 has a massive lead. So far, the 207 looks like another Lancool with an unbeatable price-to-performance ratio.

CPU Full Load Thermals - Full Speed

At full speed, the 207 sounds like a server rack. The 41.6dBA result ties the 216. The 216 is the best performer when allowed to brute force its performance by fan speed, roughly tying the 207 when we remove its front panel to test restrictions. With the panel, the 207 ran at 41.6 degrees for P-cores, or about 2 degrees warmer than without a front panel. As far as restrictions go, that’s pretty good: The panel is not substantially hurting performance, unlike some other cases we’ve tested. If you buy the optional filter, that would change. 

Using two layers of gaffer tape over the small mesh section on the steel panel would affect GPU thermals if anything, so it's no surprise that it had no effect on CPU thermals.

The 207 at 41.6 degrees ties the Flux Pro for performance. Again, though, the noise normalized results are better for case-to-case comparison.

GPU Full Load Thermals - Full Speed

Now for GPU temperatures at full speed. In the same test, the GPU was 38 degrees above ambient. Taping the side panel vent did nothing. We think this is simply because the bottom intake fans have plenty of other ventilation available through the mesh on the opposite side of the shroud and the wide-open unfiltered vent at the rear of the case. If these inlets had been minimized, it might have mattered more. It still doesn’t hurt to have more mesh in these areas, especially if you shove drives down there, but it just didn’t matter for our GPU.

GPU thermals effectively tied the Flux Pro, but with some back-and-forth over the memory and hotspot temperatures. No matter what, though, the 207 sits alongside the Torrent and Flux Pro at the top of this chart. The 216 fell behind here with an average GPU temperature of 43 degrees above ambient, which is still excellent, but not as good as the brute-force bottom intake of the 207.

GPU Full Load Thermals - Standardized Fans

Our standardized fan tests use three Noctua fans, always the same ones: two 140mm front intake, and one 120mm rear exhaust, all at 100% speed. We regularly explain the limitations of this testing. You can learn more in our methodology piece here. This allows us to remove stock fans as a variable and compare just the enclosures. With cases like the 207, the comparison is theoretical (since it doesn't make any sense to replace the stock fans), but people request it, so we wanted to provide the data.

Our standard fan configuration was obviously worse for GPU thermals than the stock configuration. With the standard fans, average GPU temperature was 43 degrees above ambient, 48 degrees for the memory, and 56 degrees for hotspot. 

CPU Full Load Thermals - Standardized Fans

On the CPU with standardized fans, we measured CPU temperatures at 38 degrees above ambient all-core and 42 degrees for the P-cores, within error of the top-scoring Torrent. Mesh fronted cases with straightforward front-to-back airflow patterns do well in this test, and the 207's CPU thermals are particularly good because the front fan slots are biased high, sending more air above the level of the GPU backplate. We usually try to position the intake fans for a better balance with GPU thermals, but that's not an option in the 207.

VRM & RAM Full Load Thermals - Noise-Normalized

The VRM and system memory are located close to the CPU in our test system, so it makes sense that a case with good CPU thermals would have good thermals in those two categories as well. The 207 slightly outperformed the Flux Pro for lowest memory temperature at 21 degrees above ambient, tied with the Torrent. VRM was tied with the Flux Pro at 27 degrees, but that's tied at the top of the chart. The 216 was warmer for both sensors, 23 degrees for the system memory and 28 for VRM.

Lian Li Lancool 207 Conclusion

Like the Lancool 215 and 216, the Lancool 207 is a well-built case with chart-topping performance while simultaneously managing to be one of the cheapest cases on our charts. The cases that perform equivalently are more expensive, like the Flux Pro, and the cases that are in the same general price category perform worse, like the 514X. Lian Li already has some strong competition in this area, including its own 216, but the 207 is even cheaper than the 216.

Phanteks has its new cases from Computex coming out soon and those should give the 207 some competition, but they’re not here yet -- and they might end up a little more expensive, but we’re not sure yet.

The Fractal Pop Air remains a good quality ultra-cheap case when it’s on sale, which has been frequently lately, but it’s just a totally different class of performance. The 207 holds the new crown for the best performance for the price -- nothing else that we’ve tested anytime recently is even close when you factor in the price. Lian Li’s 216 is the closest competitor, but is frequently around $100, making it $20 more expensive than the 207.

• 

• 

The biggest problem with the 207 is the cable management and power supply fit: The problem with getting the side panel on over the PSU is potentially a huge obstacle for some cable-heavy or thicker cable configurations and is not shared by any of the other cases we've mentioned. If you buy this case, we recommend choosing a 150mm deep PSU rather than 160 and opting for slimmer cables. Depending on your PSU and the cables you use, there's a serious risk that the side panel won't fit and you'll get annoyed enough to send the case back. Flat, slim cables have a better chance of working.

This is its only major issue that we encountered. Build quality of the panels is overall excellent, especially at the price, and thermal performance is among the best. The 207 gets a strong, but caveated recommendation, with the caveat again being related to cable management (especially for less experienced builders). If you plan ahead and have patience to thoroughly route cables, and if you can stick to a shorter PSU, we think this case is one of the most competitive on the market for its price. There aren’t a lot of well-built options in this price class.

---

Table of Contents

• AutoTOC

Intro

This article is an entry in our GN Mega Charts series. All Mega Charts are listed on the Features page, including these:

• CPU Power Consumption & Efficiency Mega Charts
• CPU Cooler Mega Charts

This section contains disclaimers, limitations of the process, and disclosures relating to data quality control. We think that this is all important for your understanding of how this page works and so that you can adjust your own expectations and potential reliance on the data to calibrate with the two groups (“Active” and “LTS”); however, if you’d like to just jump straight to the charts and ignore all of that, you may bypass the wall of text and auto-scroll down with this link.

This article contains our ‘Mega Charts’ for CPU performance benchmarks, including our production tests (commonly referred to as “creation” benchmarks) and gaming tests. Our power testing can be found on the above-linked page and is isolated, as it tends to be more static.

This page will be regularly updated with the latest of our CPU benchmark performance numbers. It will consist of two types of charts: Long-Term Support (“LTS”) and Active. The long-term support charts have several special caveats, but are intended to be available to help people better determine upgrade paths. The LTS charts are more likely to contain older CPU results.

The page also includes links to CPU reviews and comparisons, such as historical AMD vs. Intel benchmarks. It will be updated on a slower cadence from our latest reviews (so you should always defer to those for the most recent numbers), but will be updated a few times a year with larger charts than are found in our reviews. This is for a few reasons, but one is that we shorten review charts due to video height limitations (16:9 aspect ratio). The other is that it’s just too crowded for the regular updates.

This page is intended to be used long-term for our Mega Charts. You can bookmark this page, as our future updates for CPU Mega Charts will land at this same URL. The update log will be posted at the bottom of the page so that you always know the latest data set. It will be updated a couple times a year, with more frequency updates in the CPU reviews themselves.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing

Patrick Lathan
Mike Gaglione

---

How to Use This

Even as data ages, it is often still relevant for comparison -- particularly for older CPUs which mostly stop receiving performance-affecting changes, such as microcode updates or Windows patches. We are constantly re-running our CPU tests to keep data fresh, but unfortunately, this refresh cycle means that it is difficult to stack more CPUs on the charts before some sort of major change comes in. For example, major Windows updates necessitate full re-tests for reviews, but may not be as important for someone who just wants to see their older CPU represented for a “good enough” gauge of where things fall.

That’s why we split these into the LTS (Long-Term Support) and Active charts. It allows us to maintain one older dataset that has more CPUs represented, at the cost of reduced insights gained from our most modern test methods. Active gives you that for more of a modern head-to-head. It’s the difference between precision (Active) and quantity of CPUs (LTS). This is the best balance that a small team can produce, especially since we provide this website completely ad-free (you can support us on Patreon or by buying something useful for your PC builds on our store).

Current Best CPUs (Generalized Recommendations)

The below is a simple list of CPUs that, at the time of writing (dated in the columns below), we think make sense or would make sense with caveats noted.

Pay careful attention to the second column. We may only recommend some parts under certain pricing conditions. Generally speaking, we do not recommend buying CPUs above MSRP. They come down pretty regularly, especially with launch cycles. For instance, when we first posted this, the 7800X3D was about $250 overpriced. It's still on the list so that people are aware of it, but we advise waiting for it to approach that MSRP marker or to be replaced with the pending 9000X3D parts.

For a better and more thorough list of Best CPUs, please check our Best CPUs of 2023 article. We will update for 2024 also.

CPU	Reason for Recommendation	Release Date	Platform	Date of Recommendation	GN Original Review*
AMD R7 7800X3D	The original launch price was $450. We don't recommend buying much above that. Depending on price (fluctuates), this is the best gaming part at the time of writing. It is sometimes the best value gaming part. At the time of writing, value is terrible -- it lands on this list only with the note that you should wait and see. Also, 9000X3D may be around the corner.	2023	AM5	October 13, 2024	R7 7800X3D Review
AMD R7 5700X3D	This is often the best-value drop-in upgrade for AM4 platforms to give a major gaming performance boost without a totally new system. The 5800X3D would be best, but runs higher price.	2024	AM4	October 13, 2024	R7 5700X3D Review
Intel i9-12900K	At the time of writing this, the CPU has a $100 discount code that lands it at $260, which is very good value. The usual listing price is $360 at time of writing, but you'd be buying into an abandoned platform.	2021	LGA 1700	October 13, 2024	12900K Review
AMD Threadripper 7980X	For extremely core-intensive tasks that could benefit from the increased capabilities of an HEDT platform (more RAM, more cores), the 7980X remains overall unbeatable. You can check our review for more details.	2023	TRX50	October 13, 2024	TR 7980X Review
Intel i3-12100F	Alder Lake has been a time-tested architecture without major concerns. The 12100F is frequently one of the cheapest new CPUs (if not buying used) that can still play most games well. It has severe limits in some games. This is not particularly strong, but is affordable and often acceptable as a compromise.	2022	LGA 1700	October 13, 2024	12100F Review
Notes	We are posting this just ahead of Arrow Lake, so we may add some Intel recommendations to this list. Check back in a few weeks to a month to see if this has changed. The older Intel stuff is on here for its price benefit, but with the proximity of Arrow Lake, we'd generally advise waiting a few weeks rather than committing to a dead platform with the more expensive CPUs (such as 14th Series)

Is Your CPU Missing From This Data? Here’s What to Do

We frequently receive questions from people asking where their particular CPU would land on a chart. There are thousands of CPUs that could be tested, so we obviously don’t have all of those listed. The best bet is to approximate the positioning by just thinking through the data that is present and using deductive reasoning. Newer builders may not realize that it is often this simple, so we’ll outline some concepts so that you can at least get a rough idea of where your part might fall. You can apply this to any reviewer’s charts. And of course, if we’re missing a part, there are plenty of qualified reviewers out there who may have something we’re missing (and we likewise try to cover what they miss) -- that’s the value of multiple qualified reviewers.

CPU Performance Interpolation/Deductive Reasoning Examples

Example 1: The Ryzen 5 1600AF is not present on the charts.

Solution: The Ryzen 5 1600AF is functionally an R5 2600, just at slightly different clocks. Looking up original reviews would get you this information, which you can then apply to modern charts. Looking at the R5 2600 in a chart is close enough to the R5 1600AF that you could base your decisions off of that part.

Example 2: The Ryzen 5 2600 is also missing from the charts.

Solution: Pull up a few old/original reviews of the R5 2600 and identify parts that are nearby or adjacent in performance. Look at a few charts, as some games can differ. Once you have found the most commonly comparable part, you can use that as a rough gauge on charts.

Example 3: The i7-10700K is missing from the charts.

Solution: If the i5-10600K and i9-10900K are present, it’s reasonable to assume that the 10700K is between them. Although this can have a relatively wide range, the reality is that, especially upgrading from something older, it won’t matter enough to hurt decision making on new processor purchases (since anything will be a huge upgrade).

Example 4: The Sandy Bridge i7-2600K is present, but not the i7-2700K

Solution: In situations such as these, where the part is basically just a slight change (example: 2700X vs. 2700, 2600X vs. 2600, 7700X vs 7700), you can just look at the one that is present and assume close enough performance to compare. This is again where it’s important to keep perspective: If the goal is to upgrade, being 2-5% off on the estimate isn’t going to meaningfully impact decisions if the alternative is no good modern data as reviewers move on.

As a last-ditch solution: You can look at games or benchmarks which are least likely suspected to have had major patches, such as GTA V, and calculate the percent difference between the target part and mutually present part on both an older chart and the new one. Then apply this to the modern chart to approximate or interpolate the rank. For example, if you pull up a chart from two years ago with the 8700K and 12700K on it and calculate the difference (typically, (new - old) / old), you can then apply that on the same game chart from modern times. This is the least perfect method because newer games may have architecturally evolved and may not be linear and older games get patches. You’d frankly be better off finding it somewhere else on the internet, but if you really can’t, this is a method that helps get at least something to work with.

DISCLAIMER: Data Accuracy Standards & Reduced Vetting Practices

For standalone reviews that receive full video treatment, we run through a quality control process that is intensive and often takes several days to complete. These are time-intensive, cost-intensive, and critical to the accuracy of our mainline reviews. We would never skip steps for the fully produced reviews.

But we want to publish more of the data we collect outside of the reviews process because it’d help people with purchasing decisions. We have been resistant to publishing the extra data because the ‘extra’ data doesn’t go through the same validation processes. It’s still useful and rarely has issues; however, our QC standards are high and we are careful with what we release. Anything we’re uncertain of or haven’t vetted fully is withheld until it clears those bars. 

But a lot of people would benefit from knowing where an AMD R7 1700 lands today, and outside of full revisits, we don’t have a good avenue to release data that is useful, likely up to our usual standards, but just not quadruple-checked. That’s what the Long-Term Support (LTS) charts are for. 

We’ve decided to release the extra data to try and cast a wide net to help people upgrading from older or more obscure parts, but are doing so with the upfront disclosure that the difference between the Active charts and the LTS charts is the validation process. In other words, data which exists on LTS charts but not Active charts should be understood by the audience as more likely to have some sort of data confidence issue or possible deviation from expected results. It is still unlikely, but more likely than Active charts. With that transparency and understanding, here’s the difference in our processes:

Active Chart Vetting Process

One of our biggest hangups for publishing a full list of our “Mega Charts” for CPUs has been that we simply cannot sustain the intensive QC process for a secondary dataset that contains CPUs that haven’t been published on the channel recently. We often still collect data for older CPUs, but may not publish it for time reasons (meaning that it was collected for internal review, but not fully vetted for publication). 

Here’s the full process, with some specialty confidential steps removed:

• The Test Lead for the tests (typically Steve) establishes a strict SOP for the test suite to ensure data consistency. The Test Lead also determines which software is tested, performs benchmark analysis for standard deviation and consistency, and determines the testing methodology
• The Test Lead (typically Steve) establishes a Test Matrix containing all CPUs to be tested, which tests they get, and data exporting practices
• The Subject Matter Expert (Patrick) builds software in collaboration with Steve for the test suite and builds the operating system configurations, establishing a “pre-flight checklist” for testing
• The Technician who runs the tests (primarily Mike, with assistance from Patrick and Steve and occasionally Jeremy) will check the results for correct resolution, correct settings, captures that prove the test completed correctly, image quality, and general errors. The technician performs re-runs as needed
• The Subject Matter Expert performs a secondary quality control pass on data. Any data which looks suspect of a bad pass, which requires manual filtering & removal prior to retests, will be deleted and flagged for the technician to rerun
• The Technician performs re-tests and then re-evaluates the results
• The Subject Matter Expert reviews them again. They are typically resolved at this stage and occasionally rejected entirely to move forward
• A Writer & Technician (typically Jeremy) performs a full pass on CPU export names, the hierarchical stack of the data and whether the hierarchy makes sense, and identifies potential areas where additional data may be necessary to confidently come to conclusions, then passes it to Steve. If Steve authorizes any of the re-evaluations from the Writer & Technician, they go back to the lab again for testing.
• Steve performs final intensive review, including running a variety of formulae for data consistency and evaluating results against archival results for consistency of scaling
• The Writer & Technician double-checks the finals. Any disagreement with what Steve passed is voiced and revisited. Once both are in agreement that it makes sense (or the suspect data is removed until validated), the data is sent to publication
• The video editor, who is often technical enough to identify obvious oversights, performs a ‘passive’ QC while editing and flags any issues. If any are found, they go through Steve for review, analysis, and either sign-off as accurate or correction

You’ll notice that a big part of the process is passing the results through multiple team members -- typically 4-5 people look at it before publishing. This is slow, but important. 

Long-Term Support Chart Vetting Process

The Long-Term Support charts will contain data which is not in reviews, but was used as internal gauges for where parts should sensically land. As a result, this is the process it has followed since it has not previously been published:

• (Same) The Team Lead for the tests (typically Steve) establishes a strict SOP for the test suite to ensure data consistency. The Team Lead also determines which software is tested, performs benchmark analysis for standard deviation and consistency, and determines the testing methodology
• (Same) The Team Lead (typically Steve) establishes a Test Matrix containing all CPUs to be tested, which tests they get, and data exporting practices
• (Same) The Subject Matter Expert (Patrick) builds software in collaboration with Steve for the test suite and builds the operating system configurations, establishing a “pre-flight checklist” for testing
• (Same) The Technician who runs the tests (primarily Mike, with assistance from Patrick and Steve and occasionally Jeremy) will check the results for correct resolution, correct settings, captures that prove the test completed correctly, image quality, and general errors. The technician performs re-runs as needed
• (Removed) The Subject Matter Expert performs a secondary quality control pass on data. Any data which looks suspect of a bad pass, which requires manual filtering & removal prior to retests, will be deleted and flagged for the technician to rerun
• (Removed) The Technician performs re-tests and then re-evaluates the results
• (Removed) The Subject Matter Expert reviews them again. They are typically resolved at this stage and occasionally rejected entirely to move forward
• (Removed) A Writer & Technician (typically Jeremy) performs a full pass on CPU export names, the hierarchical stack of the data and whether the hierarchy makes sense, and identifies potential areas where additional data may be necessary to confidently come to conclusions, then passes it to Steve
• (Removed) Steve performs final intensive review, including running a variety of formulae for data consistency and evaluating results against archival results for consistency of scaling
• (Removed) The Writer & Technician double-checks the finals. Any disagreement with what Steve passed is voiced and revisited. Once both are in agreement that it makes sense (or the suspect data is removed until validated), the data is sent to publication
• (Removed) The video editor, who is often technical enough to identify obvious oversights, performs a ‘passive’ QC while editing and flags any issues. If any are found, they go through Steve for review, analysis, and either sign-off as accurate or correction
• (New) Steve performs final quick review, including ad-hoc comparisons between CPUs from which we have a known relative % scaling, to rapidly vet additional information. Data with low confidence is removed rather than investigated. Not every single data point is inspected, unlike reviews

This allows the team to continue work on important next reviews without forcing us to skip more important upcoming projects, but still allows us to get helpful data out. You’ll notice that this approach cuts at least 2 potential internal reviewers from the process, including reducing the amount of times the SME looks over the data, and reduces the review of every single data point down to ad-hoc glances to see if anything “jumps out” as obviously bad.

If you see anything that looks out of order, you are welcome to email us at team at gamersnexus dot net.

We are hopeful that this will allow us to publish more data to help people make upgrade decisions, with a middle-ground understanding going into it as to the limitations of the process.

With all the helpful information on how to use these charts and the disclosures out of the way, let’s continue to the data set.

Long-Term Support CPU Charts (1H24)

LTS DATA SET: Zen 5 Review Cycle (pre-Arrow Lake)

The below CPU charts are those found from our long-term support list. These will be updated only a few times a year, but contain the most data for some of the older CPUs.

Production Benchmarks

This section contains the “production” benchmarks for workstation, creation, and development applications.

Blender 3D Rendering on the CPU

The above chart ranks CPU 3D rendering performance in Blender by best to worst (faster, or lower time, is better). This should help you identify the best CPUs for rendering in 2024; however, GPUs are heavily relied upon and would be a separate benchmark.

7-Zip File Compression & Decompression Benchmarks

• 

• 

The above charts contain our tests for 7-Zip file compression and decompression. If you frequently work with compressing and decompressing data, such as for large file transfers in compressed formats or for certain game loading tasks, these will give an idea for performance.

Adobe Premiere Video Editing & Rendering CPU Benchmarks

This chart uses the Puget Suite to benchmark Adobe Premiere for video editing and rendering tasks, focusing on CPU performance. Adobe Premiere cares both about the CPU and the GPU, and relies upon a strong combination (rather than biasing toward one component, like 3D rendering might do) for reduced scrubbing or playback ‘lag’ and improved rendering and encode performance.

Adobe Photoshop CPU Benchmarks & Comparison

This chart is for Adobe Photoshop and compares some of the best CPUs currently out for Photoshop. Testing is done with the Puget Suite and includes warps, transforms, scales, filters, and file manipulation.

Chromium Code Compile CPU Benchmarks

This chart looks at a Chromium code compile. It’s a CPU benchmark for programmers and developers, with the caveat that (like any of these tests), we can’t realistically represent all compile scenarios. Because we try to represent a wide range of possible use cases, we don’t cater too much to any one specialty. This should give you an idea for at least how this specific compile behaves. If your workloads resemble this, you may be able to assume some level of linearity.

Gaming CPU Benchmarks & Best CPUs

Even if your specific game isn’t represented here, the best way to use our charts for determining CPU differences is to look at the relative ranking across a number of games. Our intent with this approach is that you can determine the best-value upgrade (our reviews are very value-oriented) by comparing the typical or average gap between two parts.

In most scenarios, the CPUs will scale similarly from game-to-game, barring any unique behaviors of a particular game. If CPU A is better value in Game A, B, and C, it is very likely also better value in Game D (though not always). 

For individual per-game benchmarks, we’d recommend our Game Benchmarks section that deep-dives on new releases whenever we get a chance.

Dragon’s Dogma 2 CPU Benchmarks

This is for Dragon’s Dogma 2, which is one of the newest games in our CPU test suite. Dragon’s Dogma 2 is still getting regular updates, so the CPUs shown on this chart were all run on the same game version. The game remains CPU-heavy in the cities with dense NPC populations, which is where we test. We published a separate deep-dive benchmark of Dragon's Dogma 2 here.

Stellaris Simulation Time CPU Comparison

This chart evaluates simulation time in Stellaris using a late game save file. The number is represented in time, with lower being better. Faster simulation (shorter time required) is noticeable in 4X or Grand Campaign / Grand Strategy games where AI turn processing requires significant CPU effort. 

Baldur’s Gate 3 CPU Benchmarks

Baldur’s Gate 3 is tested at 1080p/Medium in the above chart, which allows us to see CPU scaling all the way up to the top of the stack for the best gaming CPUs.

We test in Act III in the city itself, near a market with a lot of NPCs.

F1 2024 1080p & 1440p CPU Benchmarks

• 

• 

This chart gallery is for F1 24 and includes both 1080p and 1440p results. Typically, in scenarios which remain primarily CPU-bound, we see roughly the same hierarchical lineup across both resolutions. The top of the chart truncates maximum FPS, which means that CPUs which occasionally bounce off of the GPU bottleneck will be less reliable to differentiate from one another (as they are externally bound).

FFXIV: Dawntrail CPU Benchmarks (1080p & 1440p)

This set of 1080p & 1440p charts is for Final Fantasy XIV: Dawntrail, tested at Maximum quality settings.

Rainbow Six Siege CPU Benchmarks

Rainbow Six Siege is a problematic benchmark as it updates frequently, causing entire wipes of our dataset. This is a list of results that were all run on the same game version. Unfortunately, we don’t have as much data for it as a result of the regular wipeout.

Starfield CPU Comparison

This chart is for Bethesda’s Starfield at 1080p/Low. Despite the game’s memeable launch, it remains a useful benchmark for CPU performance comparisons.

Total War: Warhammer III CPU Benchmarks

This gives us a look at a Grand Strategy / Grand Campaign Total War game, which tend to be CPU heavy. We’re using a battle for the benchmark.

Active CPU Benchmark Charts

ACTIVE DATA SET: AMD R9 9900X CPU Review

The below list of charts is our most heavily-vetted, most recently-published data set. Due to maintenance overhead and our focus on new content, we won’t updated it after every single review, but we will update this upload after major review cycles are fully complete. For example, if both AMD and Intel are launching CPUs across a 2-3 month spread, we’ll update this list at the end when we can breathe again.

These will lack some of the data found on the LTS charts; however, they may contain more recent data (such as newer CPUs).

There may be discrepancies between the LTS and Active charts. They are not necessarily comparable, and often are not. This is for reasons such as Windows version differences, game updates, and test platform changes.

Active Charts: Production

Rather than individually break them out into headings like above, we’ll just dump all the production charts below:

• 

• 

• 

• 

• 

• 

• 

• 

• 

Active Charts: Gaming Benchmarks

And the same for gaming. You can find each game at the top of the chart:

• 

• 

• 

• 

• 

• 

• 

• 

CPUs Present on These Charts & Their Reviews

This table includes a simplified list of all CPUs on these charts, including links to the original GN reviews where present. Be advised that CPUs often have several follow-up pieces of content in rapid succession as the landscape changes. To keep things simple, we’ll just link the original reviews. You can still find the follow-ups on the channel.

CPU	Architecture	Release Date	GN Original Review*
AMD Ryzen 7 7800X3D	Zen 4	2023	AMD R7 7800X3D CPU Review
AMD Ryzen 9 9950X	Zen 5	2024	AMD R9 9950X CPU Review
AMD Ryzen 9 9900X	Zen 5	2024	AMD R9 9900X CPU Review
AMD Ryzen 7 9700X	Zen 5	2024	AMD R7 9700X CPU Review
AMD Ryzen 5 9600X	Zen 5	2024	N/A
Intel i9-14900K	Raptor Lake Refresh	2023	Intel i9-14900K CPU Review
Intel i7-14700K	Raptor Lake Refresh	2023	Intel i7-14700K CPU Review
Intel i5-14600K	Raptor Lake Refresh	2023	Intel i5-14600K CPU Review
AMD Ryzen 9 7900	Zen 4	2023	AMD R9 7900 CPU Review
AMD Ryzen 5 7600	Zen 4	2023	AMD R5 7600 CPU Review
AMD Ryzen 7 7700	Zen 4	2023	AMD R7 7700 CPU Review
AMD Ryzen 5 7600X	Zen 4	2022	AMD R5 7600X CPU Review
AMD Ryzen 7 7700X	Zen 4	2022	AMD R7 7700X CPU Review & Benchmarks
AMD Ryzen 9 7900X	Zen 4	2022	AMD R9 7900X CPU Review
AMD Ryzen 9 7950X	Zen 4	2022	AMD R9 7950X CPU Review
AMD Ryzen 9 7950X3D	Zen 4	2023	R9 7950X3D CPU Review
Intel i5-13600K	Raptor Lake	2022	Intel i5-13600K CPU Review
Intel i7-13700K	Raptor Lake	2022	Intel i7-13700K CPU Review
Intel i9-13900K	Raptor Lake	2022	Intel i9-13900K CPU Benchmarks
Intel Pentium G7400	Alder Lake	2022	Intel Pentium G7400 Review
AMD Ryzen 5 5600	Zen 3 Vermeer	2022	AMD R5 5600 Review
AMD Ryzen 5 5600X	Zen 3 Vermeer	2020	AMD R5 5600X Review
AMD Ryzen 5 5600X3D	Zen 3 Vermeer	2023	AMD R5 5600X3D CPU Review
AMD Ryzen 9 5950X	Zen 3 Vermeer	2020	AMD R9 5950X Review
AMD Ryzen 9 5900X	Zen 3 Vermeer	2020	AMD R9 5900X Review
AMD Ryzen 7 5700X	Zen 3 Vermeer	2022	AMD R7 5700X CPU Review
AMD Ryzen 7 5700X3D	Zen 3 Vermeer	2024	AMD R7 5700X3D CPU Review
AMD Ryzen 7 5800X	Zen 3 Vermeer	2020	AMD R7 5800X CPU Review
AMD Ryzen 7 5800X3D	Zen 3 Vermeer	2022	AMD R7 5800X3D CPU Benchmarks
Intel i3-12100F	Alder Lake	2022	Intel i3-12100F CPU Benchmarks
Intel i5-12400	Alder Lake	2022	Intel i5-12400 CPU Benchmarks
Intel i5-12600K	Alder Lake	2021	Intel i5-12600K CPU Benchmark
Intel i9-12900K	Alder Lake	2021	Intel i9-12900K Review
AMD Ryzen 5 3600	Zen 2 Matisse	2019	AMD R5 3600 CPU Review
AMD Ryzen 7 3700X	Zen 2 Matisse	2019	AMD R7 3700X CPU Review
AMD Ryzen 9 3900X	Zen 2 Matisse	2019	AMD R9 3900X Review
AMD Ryzen 9 3950X	Zen 2 Matisse	2019	AMD R9 3950X Review
Intel i3-9100F	Coffee Lake Refresh	2019	Intel i3-9100F CPU Benchmarks
Intel i5-9600K	Coffee Lake Refresh	2018	Intel i5-9600K CPU Tests
Intel i7-8086K	Coffee Lake	2018	Intel i7-8086K Binning Test
Intel i7-8700K	Coffee Lake	2017	Intel i7-8700K CPU Review
Intel i5-8600K	Coffee Lake	2017	Intel i5-8600K Review
Intel i7-9700K	Coffee Lake Refresh	2018	Intel i7-9700K CPU Review
Intel i9-9900K	Coffee Lake Refresh	2018	Intel i9-9900K CPU Review
AMD Ryzen 7 2700	Zen+ Pinnacle Ridge	2018	AMD R7 2700 Review
AMD Ryzen 5 2600	Zen+ Pinnacle Ridge	2018	AMD R5 2600 Review
AMD Ryzen 5 1600	Zen	2017	AMD R5 1600 Review
AMD Ryzen 7 1700	Zen	2017	AMD R7 1700 Benchmarks
AMD Ryzen 7 1700X	Zen	2017	AMD R7 1700X Review
AMD Ryzen 3 1300X	Zen	2017	AMD R3 1300X Review

Update Log

The below is an update log of changes to this page. The format is MM/DD/YYYY:

• 10/13/2024: Created page with initial dataset following Zen 5 reviews and preceding Arrow Lake reviews

Update process / house cleaning (externally visible, but used internally):

• Apply changes
• Append to update log
• Modify "Data Set" for both LTS and Active charts to identify the replacement dataset
• Modify current recommendations at the top of the page
• Append table of tested CPUs
• Update to link the latest Best CPUs article

---

Table of Contents

• AutoTOC

Intro

Today we’re reviewing the $68 SilverStone FARA 515XR with 4 included rainbow fans, not traditional cycling RGB. It was intended as an Asia-exclusive case, but SilverStone is experimenting with bringing it to the US as a budget option. We’re also reviewing the SilverStone 514X, this one has 4x traditional ARGB fans and is priced at $100 to $110.

There was a period of time where Silverstone made our #1 recommended sub-$100 case with the RL06 (watch our review). It’s been a few years, but we’re back to see if they can repeat.

Editor's note: This was originally published on October 3, 2024 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.

---

Credits

---

Test Lead, Host, Writing

Steve Burke

Testing, Writing

Patrick Lathan

Testing

Mike Gaglione

Camera, Video Editing

Vitalii Makhnovets

Camera

Tim Phetdara

Writing, Web Editing

Jimmy Thang

---

• 

• 

The 514X is a conventional mesh-fronted budget case, the kind SilverStone has been selling for years. 

As for the 515XR, SilverStone has been reluctant to bring it over to the US because the company feels this market has different expectations -- primarily for size, as this isn’t as deep as typical towers. This is a unique opportunity for us to compare a normal budget case versus the absolute minimum viable version and see what it takes to shave off that final $30.

Specs

	514X	515XR
Model No.	SST-FA514X-BG (Black)SST-FA514X-WG (White)	SST-FA515XR-BG (Black)SST-FA515XR-WG (White)
Material	Steel, plastic, tempered glass	Steel, plastic, tempered glass
Motherboard	ATX (12" x 11"), Micro-ATX (9.6" x 9.6"), Mini-ITX (6.7" x 6.7")	ATX (12" x 11"), Micro-ATX (9.6" x 9.6"), Mini-ITX (6.7" x 6.7")
Drive bay	Internal: 3.5" x 2, 2.5" x 2	3.5"/2.5" x 2, 2.5" x 2
Cooling system	Front: 120mm x 3 (120mm x 3 ARGB fans included)Rear: 120mm x 1 (120mm x 1 ARGB fan included)Top: 120mm x 3 / 140mm x 2	Front: 120mm x 3 / 140mm x 2 (120mm x 3 rainbow fans included)Rear: 120mm x 1 (120mm x 1 rainbow fan included)Top: 120mm x 3 / 140mm x 2Internal: 120mm x 2
Radiator support	Front: 120mm / 240mm / 360mmRear: 120mmTop: 120mm / 140mm / 240mm / 280mm / 360mm	Front: 120mm / 140mm / 240mm / 280mm / 360mmRear: 120mmTop: 120mm / 140mm / 240mm / 280mm
Limitation of CPU cooler	168mm	159mm
Expansion slot	7	7
Limitation of expansion card	Length : 378mm (Front mounted fans installed inside the front chassis)394mm (Front mounted fans installed outside the front chassis)Width : 174mm	Length : 350mm
Power supply	Standard PS2 (ATX)	Standard PS2 (ATX)
Limitation of PSU	217mm (Drive cage installed at the front position)185mm (Drive cage installed at the rear position)	160mm
Front I/O port	USB Type-C x 1USB 3.0 x 2Audio x 1Mic x 1	USB 3.0 x 2Audio x 1Mic x 1[Ed. Note: There's only one audio jack]
Dimension	220mm (W), x 492.7mm (H), x 458.6mm (D), 49.7 liters8.66" (W), x 19.4" (H), x 18.06" (D), 49.7 liters	204.8mm (W), x 484.5mm (H), x 402.5mm (D), 39.94 liters8.06" (W), x 19.07" (H), x 15.85" (D), 39.94 liters
Net weight	8.6 kg	SST-FA515XR-BG: 6.18 kgSST-FA515XR-WG: 6.2 kg
MSRP	SST-FA514X-BG: $99.99SST-FA514X-WG: $109.99	SST-FA515XR-BG: $67.99SST-FA515XR-WG: $73.99

Specs copied from manufacturer materials, please read review for our own measurements and opinions

The Build (514X)

• 

• 

We’ll start with the 514X and move to the 515XR to look at what cost-saving they did.

The 514X is doing what the majority of competitive cases were back in 2019, which is shoving four fans into a $100 ventilated box. This is done with careful cost saving and has become more difficult in recent years, though there are still good options: We’ll soon be reviewing the Lancool 207 as a budget high airflow case, as another new example.

• 

• 

In the 514X, there are several areas of obvious cost-saving, and SilverStone directly acknowledged some of them to us: the expansion slots have punch-out disposable covers and there are no rubber grommets on cable cutouts. SilverStone pointed out that it still made an effort to ventilate the slot covers, even though they're disposable, but disposable slot covers on a $100 case does seem a little bit too cheap even for this configuration.

• 

• 

• 

• 

As for the lack of grommets, SilverStone advertises that a steel cover plate hides the cutouts and the cables routed through them. That's the theory, at least; we couldn't get the cover installed in its stock position. Fortunately there are two options for placing the cover (or it can be removed entirely, although the built-in GPU support relies on it. It's our policy to leave everything in place as much as possible for stock testing, but we were forced to shift the cover forward to get the 24-pin power cable connected. Moving the cover only requires removing one screw.

• 

• 

The HDD cage itself is only compatible with 3.5" drives: 2.5" drives must be mounted behind the motherboard tray.

• 

• 

The fan ARGB hub uses a SATA power connector and takes input from a button on the front panel (for built-in patterns) and an ARGB header (for external control). 

The hub doesn't have PWM speed control (though there are unpopulated pads for it), and even if it did, all four case fans are 3-pin. That’s not something we see very often these days. We always connect case fans to motherboard headers for thermal testing, which always allows speed control via voltage. The only advantage of the hub is that SilverStone uses it to pre-manage the rats' nest of cables.

Each of the fans has a daisy-chained ARGB connector, so if you have one free ARGB header and four free fan headers, you can get rid of the hub and connect everything to the motherboard. 

• 

• 

• 

• 

The front fan mounts are limited to 120mm exclusively. We’d recommend planning to use the stock fans with the 514X. On the plus side, there's space to move the front fans back into the interior of the chassis, which would give additional intake surface area at the cost of GPU clearance.

• 

• 

There's support for radiators up to 360mm on both the front and top mounts, but the front mount is easier to work with due to the case dimensions. We had to angle the top in. 

• 

• 

• 

• 

There are a couple of other minor budget-related points: no thumbscrews for the expansion slots, a loose square of mesh for a PSU filter, and a chassis that was originally built for a different case, evidenced by the unused side panel snaps and slot for a PSU shroud extension that doesn't exist. 

• 

• 

• 

• 

On the other hand, the 514X's front panel is almost entirely metal, including the snaps, and a GPU support is included with the case. The snaps are good, and although a GPU support in that location isn’t the most helpful, it doesn't hurt. 

In spite of the low price, the front I/O includes a USB Type-C port. The Type-C port in particular is still one of the more expensive case features, so we're happy to see one here.

We've spoken to manufacturers before about colors and pricing. Black has been the most broadly popular color for years, so it's the most common and the cheapest. Low-margin budget cases are the most likely to reflect paint price differences in retail prices, like the King 95 PRO, where it’s a little costlier for white version. The white version of the 514X also costs $110 whereas the black version costs $100. 

The Build (515XR)

Let’s move on to the 515XR, which is the cheaper of the two.

The 515XR is much cheaper than the 514X, about a 33% price reduction when comparing the like-for-like color SKUs. It's not necessarily worse than the 514X, but there are a few reasons that SilverStone has been reluctant to bring the 515XR to the US market, like the case’s shorter depth, which won’t fit longer GPUs. The company has been very open with us about this release being a trial run.

SilverStone is also concerned that the rainbow fans are going to lead to confusion and returns in the US market. In Asia, we’re told that rainbow fans are common and are understood. 

The fans in the 515XR aren't ARGB, which is fine, but since most cases are photographed with a rainbow pattern for RGB now, this will lead to confusion. The term "rainbow fans" combined with marketing images of multicolored fans implies baked-in RGB lighting patterns. 

Instead, each of the fans has a set of static non-animated multicolored LEDs that can't be turned off since they're powered by the fan connectors. This is an older school approach for price. We wouldn't mind some old-school solid-color LEDs with that, or no LEDs at all, but we’re not sure about the rainbow choice. According to SilverStone, rainbow fans are more common in the Asian market from which the 515XR originates.

Out of the box, the fans are all connected to a single 4-pin Molex adapter. We recommend leaving this alone: the fan cables are too short to reach anything on their own, and if you drop fan speeds, the LEDs dim (they run on the same circuit). Those of you who built computers in the early 2000s will remember this behavior from the older LED fans.

The 515XR is nearly identical to the sawed-off form factor of the old Fractal Meshify C. The Meshify C was a case that came out back in 2017 (watch our review), in the brief span after optical drive bays and before really massive GPUs. This is 2024, though, and 4090s exist, so SilverStone is worried that the 515XR is "only" compatible with GPUs up to 350mm long. 

We don't see this as an issue. Most people buying a $68 case are probably not buying super long video cards. We recommend limiting GPUs to ~300mm to give the case fans some space: if your GPU is bigger than that, there are better case options.

Like the 514X, it doesn't make sense to replace the 515XR's stock fans; however, unlike the 514X, the 515XR has 140mm front mounts. 

• 

• 

• 

The 515XR's drive support is also more flexible, since the drive cage can mount either 2.5" or 3.5" drives. The cage has a removable sled with vibration damping, and the dedicated 2.5" mounts are separate and held in with thumbscrews, all of which are improvements over the 514X. The PSU mount in the 515XR also has foam supports that the 514X lacks, and there's a single reusable expansion slot cover.

The 515XR has two internal 120mm fan mounts on top of the PSU shroud. We generally don't see much thermal improvement from shroud-top fans without careful planning, like in the Antec Flux Pro. The 515XR's shroud is mostly sealed. Still, it's an option that the 514X lacks.

• 

• 

We do not recommend putting a 360mm radiator into the 515XR. There's only a 32mm gap left in the shroud for a radiator (with the stock fans installed), and the drive cage may also interfere with the available space depending on which drives are installed. The top slot technically fits radiators up to 280mm, but this would make it difficult to use any of the cutouts at the top edge of the motherboard.

• 

• 

• 

• 

There are some cheap quality of life improvements we'd like to see. First, the cutout nearest the 24-pin power header is almost too small, and the cutouts along the bottom edge of the motherboard are partially blocked by the PSU. Second, the reset and power buttons aren't marked. Finally, the rear fan is positioned so that it needs to be fully removed before installing motherboards with built-in rear I/O covers.

As a final note, we asked SilverStone why the black version of the 515XR is 20 grams lighter than the white one, assuming it could be a typo. Its product manager says that white paint requires a thicker coat, and that the white fans also weigh slightly more than the black ones, so the 20 gram difference is real.

Thermals

We're covering two different cases in one review, so most of our comparisons here will be between the 514X and the 515XR. We don't have many sub-$100 budget cases on the charts yet, so these will serve as a starting point as we add more, like the upcoming Lian Li Lancool 207. Of the cases that have been tested with our current methodology, the aging Fractal Pop Air RGB is the closest match to the SilverStone cases, so we'll be using that as a representative example of existing budget mesh-fronted enclosures. We've seen the Pop Air drop as low as $50 for the non-RGB variant. Montech's Sky and Air cases are usually strong competitors in this space, and we'll add those to our charts as we test them.

CPU Full Load Thermals - Noise-Normalized

Our first test is for noise-normalized thermals, where we set all cases to the same noise level in our hemi-anechoic chamber. Neither the 514X nor the 515XR offer fan control out of the box, but we were able to perform noise-normalized tests by connecting the fans to motherboard headers as usual.

The 514X averaged 48 degrees Celsius above ambient all-core and 52 degrees for the P-cores, while the 515XR was significantly worse at 53 all-core and 57 P-core. The 515XR's dual-layer front panel is more restrictive, while the 514X's front panel is a single layer of steel mesh. 

The 514X isn't especially impressive outside of its price when compared to the rest of the cases on this CPU thermal chart, but it's tied with the Pop Air RGB (watch our review) -- which has been on heavy sales lately. The 515XR has some of the worst results on the chart, tied with the stock HYTE Y60. This is a letdown, and it's not at all what we expected: we've seen plenty of cheap mesh-fronted cases punch above their weight, like the Lian Li Lancool 215, Montech X3 and Air 1000, and several Phanteks cases like the P400A (read our review) with the mesh front. We’re adding the 216 back to the charts later this week along with the 207, so that’ll help give some perspective as well.

The industry has moved away from using multiple layers of material in mesh front panels, which has led to better performance in many cases. The 515XR could be something special if it had not made this design decision.

GPU Full Load Thermals - Standardized Fans

Our standardized fan test replaces all stock fans in each case with the same set of three Noctua fans: two 140mm intake, one 120mm exhaust. We have explained why there are limitations to this testing, such as worsening performance of cases which include more or larger fans. However, we run this test due to popular demand from the audience. You can find our methodology and limitations of this testing linked here.

This is the perfect chance to see how the stock fans affect the performance of the SilverStone cases, versus the designs of the chassis themselves.

GPU temperature in the 514X averaged 43 degrees Celsius above ambient, 49 on the memory, and 56 hotspot. That puts it in the middle of the chart, tied with several other similar mesh-fronted front intake configurations like the Flux Pro, Torrent, and Pop Air RGB. That's confirmation that the 514X's front panel and general layout are at least as good as other cases in the same category. The 515XR averaged 49 degrees, 57 on the memory, and 64 hotspot, which puts it among the worst results on the chart for the second time in a row. Again, all of these results were recorded using the same set of fans, so the difference comes down to the 515XR's front panel. The 514X seems fine. The 515XR is a let-down.

CPU Full Load Thermals - Standardized Fans

In the same test, the 514X's CPU temperature average of 40 degrees above ambient still came close to the other mesh-fronted cases, but the Flux Pro (read our review), Torrent, and Pop Air all averaged 38 degrees with the standardized fans. The 515XR was again worse at 44 degrees, but the bottom of the chart is stacked with bottom-intake configurations, so it's not among the worst results. Front or side intake is significantly better for CPU thermals with our test bench.

GPU Full Load Thermals - Noise-Normalized

Back to the noise normalized results, neither of the SilverStone cases did well for GPU thermals. The 514X averaged 50 degrees above ambient, 58 memory, and 65 hotspot, with only two previously-tested cases doing worse, one of them being the Pop Air RGB. 

The 515XR is the new hottest result on the chart. We express all temperatures as degrees above ambient to account for fluctuations, so the average GPU temperature of 54 degrees above ambient translates to a logged steady state temperature of 75 degrees. That wasn't enough to cause any throttling, but in a warmer room, the GPU might lose some boost headroom.

CPU Full Load Thermals - Full Speed

This is the full speed chart with the stock fans.

The 514X's stock fans topped out at 1300-1400RPM, while the 515XR's were 1100-1200RPM. Those limits are a little low for 120mm fans, which makes them relatively quiet, which is important given that both cases are set up to run all case fans at 100% speed out of the box. With all four fans maxed out, average all-core CPU temperature in the 514X was 44 degrees above ambient while maintaining a noise level of 34.6dBA. It matched some louder cases, like the 36.8 dBA King 95 Pro, but only slightly outperformed the quieter 31.3 dBA Pop Air RGB. 

The 515XR tied the Pop Air for noise but had worse thermals, as usual. The front panel design really prevents it from doing as well as it could.

VRM & RAM Full Load Thermals - Noise-Normalized

In noise normalized testing, VRM and RAM thermals aligned with the CPU thermal results. The 514X averaged 34 degrees above ambient for the VRM and 28 degrees for memory, putting it in the middle of the chart, while the 515XR averaged 39 for the VRM and 32 for the memory, the hottest results on the chart in both categories.

Conclusion

The 514X is an extremely normal case: it costs about $100, it has the basic features you'd expect from a budget case, but it also has a full set of four ARGB fans. Thermal performance doesn't look amazing compared to the rest of our chart, but we haven't yet fully populated our chart with comparable $100 cases. Performance is similar to the Fractal Pop Air RGB, so you can look back at that review for a rough comparison versus a wider selection of older cases. The 514X isn't bad, but we aren't excited about it, and we've seen the Fractal North on sale recently for as low as $110. Anyways, the 514X is okay.

We were excited about the $68 515XR mostly for the price, especially after seeing that it has almost every feature we care about from the 514X (except the Type-C header). Sub-$68 is the domain of DIYPC and SAMA, and there are few name-brand options with reasonable ventilation at that price. Unfortunately, it reminds us strongly of the Thermaltake Versa J24 (watch our review). The J24 was extremely similar to the 515XR in several ways, but most importantly it had a layered front panel that hurt its performance. For that review, we were able to remove a filter and improve the J24's performance without changing its appearance, but that's not an option with the 515XR. The 515XR is almost a really competitive case at its price. We’re just a little hung up on its front panel. We have to admit that there aren't easy alternatives to recommend at this price (other than cases on sale, like the Pop Air), but the 515XR could have had a wholehearted recommendation with some alterations to the front panel.

---