A week ago we looked at the slightly disappointing AMD Ryzen 9600X and 9700X, which boasted modest gaming performance gains in some titles, significantly better content creation performance, and slightly better heat dissipation. Now it’s time to take a look at the second half of AMD’s Zen 5 quartet, the £459/$499 Ryzen 9 9900X and £609/$649 Ryzen 9 9950X. These are powerful 12-core and 16-core parts that will likely appeal more to content creators than gamers. So do they make a better case for Ryzen 9000? And does either manage to beat the 7800X3D’s top-level gaming performance?
Unfortunately, after four days of frantic benchmarking and debugging, I’m not sure AMD succeeded on any of these counts. Our testing of the Ryzen 9900X and 9950X was marked by bafflingly poor gaming performance, including performance drops from the 7900X and 7950X, alongside some examples of genuine improvements that still don’t go far enough to make these CPUs worth recommending.
This review marks the debut of at least two improvements to our benchmarking suite compared to last week’s testing, which itself marked the launch of an entirely new automated benchmarking system for Digital Foundry – and therefore featured a more limited selection of games than we’ve offered in the past.
First, we added 720p data – in addition to the 1080p, 1440p, and 4K data – which some people prefer for CPU benchmarking as it takes the GPU out of the equation even more. Second, we added two new games in new genres: Counter-Strike 2 representing competitive shooters and Starfield representing Bethesda-style RPGs. Both pose a tough challenge for modern CPUs, especially with the higher refresh rate screens that are common today. Of course, we’ll continue to expand our testing in the future, and there are some exciting possibilities on the horizon.
Based on our 9600X and 9700X results from last week, while Games While performance doesn’t match the pace of improvement seen in previous Ryzen generations, we expect at least a reasonable performance boost in content creation. This is thanks to an increase in IPC (instructions per clock, an indicator of single core performance), while clock speeds, power consumption, and temperature remain unchanged or improve slightly. As with the 9600X and 9700X, the single-core gains are a combination of architectural improvements (such as an improved branch predictor, better AVX-512 support, and doubled data bandwidth between L2 and L1 caches) and the move to a more efficient 4nm CCD.
In that sense, the 9900X’s 120W TDP is lower than the 7900X’s 170W – although the 9950X’s 170W performance is the same as the 7950X it replaces. Improved thermal resistance also means temperatures should be lower for the same TDP – AMD claims a 7°C reduction with adjusted TDP, which is arguably more important in these high core count products that are more often used for all-core workloads where temperatures are maximized.
CPU design | Thrust | base | L3 cache | TDP | RRP | |
---|---|---|---|---|---|---|
Ryzen 9 9950X | Zen5 16C/32T | 5.7GHz | 4.3GHz | 64MB | 170W | £609/$649 |
Ryzen 9 9900X | Zen5 12C/24T | 5.6GHz | 4.4GHz | 64MB | 120W | £459/$499 |
Ryzen 7 9700X | Zen5 8C/16T | 5.5GHz | 3.8GHz | 32MB | 65W | £339/$359 |
Ryzen 5 9600X | Zen5 6C/12T | 5.4GHz | 3.9GHz | 32MB | 65W | £269/$279 |
Ryzen 9 7950X3D | Zen4 16C/32T | 5.7GHz | 4.2GHz | 128MB | 120W | 699 USD/699 GBP |
Ryzen 9 7950X | Zen4 16C/32T | 5.7GHz | 4.5GHz | 64MB | 170W | 699 USD/739 GBP |
Ryzen 9 7900X3D | Zen4 12C/24T | 5.6GHz | 4.4GHz | 128MB | 120W | 599 USD/599 GBP |
Ryzen 9 7900X | Zen4 12C/24T | 5.6GHz | 4.7GHz | 64MB | 170W | 549 USD/579 GBP |
Ryzen 9 7900 | Zen4 12C/24T | 5.4GHz | 3.7GHz | 64MB | 65W | 429 USD/519 GBP |
Ryzen 7 7800X3D | Zen4 8C/16T | 5.0GHz | 4.2GHz | 96MB | 120W | 449 USD/375 GBP |
Ryzen 7 7700X | Zen4 8C/16T | 5.4GHz | 4.5GHz | 32MB | 105W | 399 USD/419 GBP |
Ryzen 7 7700 | Zen4 8C/16T | 5.3GHz | 3.8GHz | 32MB | 65W | 329 USD/349 GBP |
Ryzen 5 7600X | Zen4 6C/12T | 5.3GHz | 4.7GHz | 32MB | 105W | 299 USD/319 GBP |
Ryzen 5 7600 | Zen4 6C/12T | 5.1GHz | 3.8GHz | 32MB | 65W | 229 USD/249 GBP |
Ryzen 5 7500F | Zen4 6C/12T | 5.0GHz | 3.7GHz | 32MB | 65W | 200 USD/255 GBP |
As suggested by AMD, we’re using a similar physical setup to our previous Ryzen 7000 tests, including an ASRock X670E Taichi motherboard, G.Skill Trident Z5 Neo DDR5-6000 CL30 RAM, and an Eisbaer Aurora 240mm AiO. The big difference is that our RTX 3090 has been swapped out for an RTX 4090 Founders Edition, which offers additional graphics power to push modern CPUs even harder.
While Ryzen 9000 is the star of the show, we also tested a selection of previous Ryzen 3000 and 5000 processors, as well as a selection of Intel’s 14th Gen processors. Our Intel CPU tests were conducted on the Gigabyte Aorus Z790 Master using the same DDR5-6000 CL30 RAM, while AM4 CPUs were tested on the classic Asus ROG Crosshair 8 Hero using a Trident Z Royal DDR4-3600 CL16 kit.
For storage, we use a 4TB Lexar NM790 PCIe 4.0 NVMe SSD and 4TB Kingston KC3000. Our rig is completed by a 1000W Corsair RM1000x power supply. The tests were performed on a fresh installation of Windows 11 with the latest Windows updates (23H2), chipset drivers (6.06.28.910) and BIOS revisions (3.06 for the ASRock AM5 board).
Cinebench | 2024 (1d) | 2024 (MT) | R20 (1T) | R20 (MT) |
---|---|---|---|---|
Ryzen 5 3600X | 77 | 578 | 485 | 3654 |
Ryzen 7 5800X3D | 95 | 915 | 546 | 5746 |
Ryzen 9 5900X | 98 | 1171 | 610 | 8393 |
Ryzen 5 7600X | 114 | 845 | 744 | 5814 |
Ryzen 7 7700X | 118 | 1127 | 758 | 7609 |
Ryzen 7 7800X3D | 112 | 1074 | 688 | 6988 |
Ryzen 9 7900X | 116 | 1605 | 776 | 11196 |
Ryzen 9 7950X | 121 | 2004 | 784 | 14272 |
Ryzen 5 9600X | 132 | 935 | 850 | 6358 |
Ryzen 7 9700X | 130 | 1172 | 862 | 7851 |
Ryzen 9 9900X | 135 | 1784 | 879 | 12617 |
Ryzen 9 9950X | 138 | 2235 | 866 | 15850 |
Intel Core i5 14600K | 120 | 1400 | 777 | 9420 |
Intel Core i7 14700K | 127 | 1987 | 818 | 13614 |
Intel Core i9 14900K | 133 | 2107 | 875 | 15297 |
Before we get into the gaming benchmarks, it’s worth looking at the results of our content creation tests. These serve as synthetic benchmarks for gamers, and suggest certain levels of performance we can expect in a best-case scenario. Of course, they can also be a useful indicator for people who actually intend to use a Ryzen 9000 CPU for 3D modeling or video production. These 12- and 16-core parts tend to deliver impressive performances, and thankfully, the same is true for Zen 5.
We’ll look to expand this in the future, but right now we have three workloads: Cinebench 2024 and R20 simulating the 3D modeling and animation program Cinema 4D, while Handbrake is literally just transcoding an early DF Patreon video to H264 and H265 formats – a commonplace task for anyone who works with video.
First, let’s look at the Cinebench 2024 results, which are recorded after a 10+ minute loop to ensure that the CPUs can’t quickly get a good result before thermal or power limits kick in. Here we see single-core improvements in the range of ~15 percent for both CPUs over their predecessors, while multi-core gains are a bit more modest: 11 percent for both. For comparison, Intel’s 14700K is 11 percent faster than the 9900X in this test, but the 9950X outperforms the 14900K by about six percent.
Cinebench R20 is a legacy test for us, having been run in a large number of previous CPU reviews, so we thought it was worth running as well in case you want to go back and compare results. We saw similar margins here gen-to-gen (+11 percent multi-core for both Ryzen 9000 CPUs).
Handbrake | H264 (images/s) | HEVC (fps) |
---|---|---|
Ryzen 5 3600X | 26.66 | 10.80 |
Ryzen 7 5800X3D | 42.00 | 18.71 |
Ryzen 9 5900X | 57.59 | 23.83 |
Ryzen 5 7600X | 41.29 | 18.31 |
Ryzen 7 7700X | 53.27 | 23.65 |
Ryzen 7 7800X3D | 49.63 | 21.54 |
Ryzen 9 7900X | 78.35 | 32.59 |
Ryzen 9 7950X | 98.58 | 41.68 |
Ryzen 5 9600X | 42.51 | 19.77 |
Ryzen 7 9700X | 51.80 | 23.79 |
Ryzen 9 9900X | 82.96 | 35.33 |
Ryzen 9 9950X | 103.25 | 44.97 |
Intel Core i5 14600K | 59.42 | 25.39 |
Intel Core i7 14700K | 80.26 | 31.07 |
Intel Core i9 14900K | 85.06 | 35.08 |
Our final test is the Handbrake transcoding test, where we convert an 822MB 4K video file using the H264 and H265 codecs, setting the Production Default preset to CRF 18. We’re using the latest version of Handbrake here, 1.8.1, which conveniently displays its statistics in the application window rather than hiding the average encoding frame rate in a log file.
These transcoding results show a much smaller advantage for Ryzen 9000, something we also saw with the 9600X and 9700X. That’s about five percent in H264 and eight percent in H265 for the 9900X and 9950X over their immediate predecessors – possibly due to those architectural improvements in AVX.
We didn’t get a chance to experiment with PBO and other performance improvements this time, but we did see a decent jump in the 9700X at the expense of power efficiency, and I expect the 9900X and 9950X to be significantly faster beyond the standard power consumption and thermal budget.
Now that we’ve completed our content creation benchmarks, it’s time for the gaming benchmarks. Unfortunately, it’s mostly downhill from here.