Table of Contents
Introduction
So far, AMD’s Ryzen 9000 series of processors have launched to largely cool reviews (including our own AMD Ryzen 9000 Content Creation Review). Although they are noticeably more efficient than their last-gen Ryzen 7000 counterparts in all-core workloads, they otherwise offer little incentive for most users to upgrade: Relatively modest performance gains of 5-15%, slight increases in memory frequency of 400 Mbps (in 2-DIMM configurations), and improved AVX512 support. AVX512 can be very important for specific workloads, but rarely for the general public. All this comes with, at present, higher street pricing than the CPUs they are replacing.
However, for many users, CPU upgrades are not an every-generation affair. To that end, we wanted to supplement our typical performance testing with some looking at a variety of CPUs from the last three Ryzen generations. Although we could only test the top-end SKUs for Ryzen 5000 and 3000 Series processors, relative performance within a generation tends to be relatively stable over time. Checking back on our reviews for that generation will let you find out the present performance gap between a lower-end SKU and the top-end model.
To keep this article brief, we are not listing all the configurations tested. However, we have standardized test platforms to ensure consistency between benchmarks. For details on the Ryzen 9000 and 7000 testing, check out the “Test Setup” section of our full AMD Ryzen 9000 Content Creation Review. The 5950X and 3950X were tested on a similar X570E platform with an RTX 4080, 32 GB of DDR4 RAM at JEDEC 3600, and all the latest drivers.
If you are interested in seeing how your exact configuration compares to the newest workstation parts, you can also run many of the tested benchmarks yourself. Most of the benchmarks we use in our articles are freely available for you to download and run, and you can find a list (with links) below. Unfortunately, we couldn’t gather results for some of the applications we tested in our full review, such as Unreal Engine and RealityCapture, so those will be absent from this article.
Download Benchmarks
Photoshop 25.9.1 – PugetBench for Photoshop 1.0.1 |
Premiere Pro 24.5 – PugetBench for Premiere Pro 1.0.1 |
After Effects 24.5 – PugetBench for After Effects 0.96 |
DaVinci Resolve 18.6.6.7 – PugetBench for DaVinci Resolve 1.0 |
Unreal Engine 5.3 |
RealityCapture |
Cinebench 2024 |
Blender 4.0.0 |
V-Ray 6.00.01 |
Graphic Design: Adobe Photoshop
We started our look at gen-to-gen performance with Adobe Photoshop. Photoshop is relatively lightly threaded and tends to prefer single-core performance. Historically, Intel has been dominant in our Photoshop benchmark, but between AMD’s recent performance advancements and an overhaul of our benchmark to make it more representative of typical Photoshop workflows, AMD has leaped ahead in our scores.
One quirk we found with Photoshop testing this generation is that some of the lower-core processors outperformed the higher-end CPUs. We suspect some of this is due to the single-CCD design of the 9700X and 9600X, but it does mean that some of the previous generations’ Ryzen 7s and Ryzen 5s could outperform their Ryzen 9 counterparts.
Nonetheless, the Ryzen 9 9950X has a modest 7% performance improvement over the 7950X, but a 24% improvement over the 5950X. Compared to the 4-year-old 3950X, the 7950X has a commanding 50% performance lead.
Video Editing: Adobe Premiere Pro
Moving on to Premiere Pro, we expect to see a smaller overall difference in performance, as the application has a lot of portions that are not CPU-bottlenecked. Workloads like processing GPU Effects and RAW media rely primarily on the GPU, so the CPU has less impact on those subscores. Overall, this tends to decrease the effect of upgrading a CPU on the high-level results.
This effect is seen in that the 9950X is not actually faster than the 7950X; instead, both score within the margin of error of each other. The results for the 5950X and 3950X are also both fairly close, with 9950X having a 13% lead over the former and a 20% lead over the latter.
Motion Graphics: Adobe After Effects
Like Photoshop, After Effects is primarily single-threaded for CPU-bound scenarios. However, certain heavy workflows may use its Multi-Frame Rendering (MFR) feature, which can take advantage of CPUs’ multi-core capability. This somewhat muddies the overall score, as it measures both lightly and heavily threaded scenarios.
Looking at the Overall score, the 9950X has a small 5% lead over the 7950X—just on the edge of our standard margin of error for these benchmarks. This extends to a large 31% performance improvement over the 5950X and 50% over the 3950X. The latter two are large and definitely make the 9950X (or 7950X) a promising upgrade for After Effects.
However, the performance improvements are even larger when only the multi-core score is considered. There, the 9950X was still only a touch faster than the 7950X, but 42% faster than the 5950X and more than 100% of the 3950X.
Video Editing / Motion Graphics: DaVinci Resolve Studio
To an even larger degree than Premiere Pro, much of our DaVinci Resolve benchmark is GPU accelerated, meaning that we see relatively smaller performance changes from CPUs. With our recent 1.0 version of the benchmark, we have added a number of additional tests looking at the various “AI” features within Resolve, which further biases the overall score away form pure CPU performance. However, this is reflective of how real-world workflows work in these types of applications.
Looking at the scores, we see that the 9950X and 7950X perform identically to each others. However, the 9950X does have a solid 16% performance lead over the 5950X. This gap widens when comparing to the 3950X, with the 9950X being 21% faster. These represent solid improvements over CPUs older than last gen, but no more than what we saw from Ryzen 7000.
CPU Rendering
Moving from the photo/video suite to CPU rendering, we see more of a performance gain for Ryzen 9000—although it depends a lot on the application. CPU rendering has traditionally been an area of strength for AMD, although desktop processors like this are not typically used for it. Instead, CPU rendering tends to be done on Threadripper or EPYC systems with much higher core count and memory capacity, while desktop platforms usually make use of GPU-based renderers.
V-Ray is the first of the CPU renderers we looked at, and in it, the 9950X gained a solid 14% over the 7950X. This positions it 70% faster than the 5950X and 92% faster than the 3950X. In Blender, we see similarly strong performance, with the 9950X 48% and 56% faster than the older CPUs, respectively.
However, in Cinebench multi-core, the 9950X once again performs identically to the 7950X. This means that it is 40-50% faster than the 5950X and 3950X but not necessarily a compelling upgrade over the 7950X. Single-core performance is better, with the 9950X seeing a 13% uplift over the 7950X, 35% over the 5950X, and 65% over the 3950X.
Conclusion
We typically expect that a new generation of CPU will handily outclass anything older than one generation back, as typical performance improvements of 15-25% compound. However, in this case, the Ryzen 9000 series saw relatively little performance gains over the Ryzen 7000 series. This has resulted in relatively muted performance gains over the two-generation old Ryzen 5000 series of processors compared to some of the other times we have looked at performance gains over multiple generations. The AMD Ryzen 9000 series of processors still handily outclass any 5000 or 3000 CPUs, but given the performance of the Ryzen 7000 processors, they may not represent the best upgrade for the money.
Of course, this only looks at AMD Ryzen CPUs and largely ignores other factors, such as what a modern GPU can give you. If you want to see exactly how your existing system compares to a new system, and whether an upgrade is worth the investment, the best thing to do is download a benchmark that best matches your workflow (such as our PugetBench for Creators) and compare your results directly to what we show in our hardware articles, or to results in public databases like PugetBench, V-Ray, and Blender. Just bear in mind that the performance gains of the Ryzen 9000 CPUs tend to be very application-specific, so even a similar application may not be a good proxy for your workflow.
If you need a powerful workstation to tackle the applications we’ve tested, the Puget Systems workstations on our solutions page are tailored to excel in various software packages. If you prefer to take a more hands-on approach, our custom configuration page helps you to configure a workstation that matches your exact needs. Otherwise, if you would like more guidance in configuring a workstation that aligns with your unique workflow, our knowledgeable technology consultants are here to lend their expertise.>>>