RIP Intel: AMD Ryzen 7 9800X3D CPU Review & Benchmarks vs. 7800X3D, 285K, 14900K, & Morejimmy_thang November 7, 2024
The Highlights
Buy a GN 4-Pack of PC-themed 3D Coasters! These high-quality, durable, flexible coasters ship in a pack of 4, each with a fully custom design made by GN's team. You'll get a motherboard-themed coaster with debug display & reset buttons, a SATA SSD with to-scale connectors, RAM sticks, and a GN logo. These fund our web work! Buy here.
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.
Steve Burke
Mike Gaglione
Vitalii Makhnovets
Tim Phetdara
Jimmy Thang
The 9800X3D is a $480 MSRP part for the AM5 platform. Let’s get into the review.
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.
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.
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.
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.
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.
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.
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.
Grab a GN15 Large Anti-Static Modmat to celebrate our 15th Anniversary and for a high-quality PC building work surface. The Modmat features useful PC building diagrams and is anti-static conductive. Purchases directly fund our work! (or consider a direct donation or a Patreon contribution!)
Let’s get into gaming benchmarks.
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.
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 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.
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.
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 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 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.
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.
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 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 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 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.
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.
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.
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.
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.
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.
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.
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.
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 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 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.
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 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 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.
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.
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.
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.
Visit our Patreon page to contribute a few dollars toward this website's operation (or consider a direct donation or buying something from our GN Store!) Additionally, when you purchase through links to retailers on our site, we may earn a small affiliate commission.
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.
Continue reading...
We test the 9800X3D’s frequency, thermals, gaming performance, efficiency, and more
The Highlights
- The 9800X3D is a follow up to the 7800X3D but runs on Zen 5 and changes the location of its cache within its die stack
- It’s not the most competitive CPU when it comes to production workloads
- The 9800X3D is the new gaming king CPU
- Original MSRP: $480
- Release Date: November 7, 2024
Table of Contents
- AutoTOC
Buy a GN 4-Pack of PC-themed 3D Coasters! These high-quality, durable, flexible coasters ship in a pack of 4, each with a fully custom design made by GN's team. You'll get a motherboard-themed coaster with debug display & reset buttons, a SATA SSD with to-scale connectors, RAM sticks, and a GN logo. These fund our web work! Buy here.
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 | GamersNexusEarly 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
Grab a GN15 Large Anti-Static Modmat to celebrate our 15th Anniversary and for a high-quality PC building work surface. The Modmat features useful PC building diagrams and is anti-static conductive. Purchases directly fund our work! (or consider a direct donation or a Patreon contribution!)
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.
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
Visit our Patreon page to contribute a few dollars toward this website's operation (or consider a direct donation or buying something from our GN Store!) Additionally, when you purchase through links to retailers on our site, we may earn a small affiliate commission.
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.
Continue reading...