AMD’s Ryzen 9 5950X is the newer Zen 3 based version of their 16 core 3950X processor, so which should you pick, or is it worth upgrading? I’ve compared both in games and applications to show you the differences.
CPU spec differences
There doesn’t appear to be that much difference between these two processors based on the specs. Both have 16 cores and 32 threads, the same TDP and same amount of cache, though the 5950X has a 200MHz higher boost clock, so most of the differences are going to be due to IPC improvements present with Zen 3.
Both processors were tested in the same system. I’ve tested with 32gb of DDR4-3200 memory running in dual channel at CL14 and with MSI’s GeForce RTX 3090 Gaming X Trio graphics card to minimize bottlenecks. Neither processor comes with a cooler, so I’ve done all testing with my Fractal S36 AIO with Noctua NT-H2 thermal paste for comparable results.
I’ve tested both CPUs at stock, and with the following all core overclocks applied, so I was able to push the 5950X up to 4.7GHz, while I couldn’t get my 3950X above 4.3GHz, but this will vary based on silicon lottery and cooling. With that in mind we’ll first check out the differences in various applications, as well as power draw and thermals, followed by gaming tests afterwards, then finish up by comparing some performance per dollar metrics.
Starting out with Cinebench R20 we can see the 5950X offering some nice improvements. At stock, its single core performance was almost 19% faster than the 3950X, a pretty great improvement in just one generation. With all 16 cores loaded up in the multicore test though, the 5950X was scoring around 9% higher than the 3950X at stock. The single core scores actually lower with the overclocks in place because locking all cores at these speeds together prevents the higher single core boost speeds that happen at stock. The 5950X was now 20% higher in multicore score with both overclocked, compared to the just 9% boost at stock. I’ve also tested the older Cinebench R15 as a lot of people still use it so you can compare my results, the margins weren’t too different to R20 so let’s move on. I’ve tested Blender with the BMW and Classroom benchmarks. This is another multicore workload. With both at stock speeds, the BMW test was only 2% faster on the 5950x, while the longer classroom test was nearly 6% faster, so not that much difference. With both overclocked though, the 5950X pulls out further in front again, with a 12% lead in the BMW test, and a 16% lead in the Classroom test, so it really seems that the 5950X is getting better gains with the overclocks in place. The V-Ray benchmark is another core heavy rendering workload, and unlike blender which had below average differences, this test had some of the biggest multicore differences out of all applications tested. At stock, the 5950X was scoring 19% higher than the 3950X, and then with both overclocked the 5950X is now almost 29% ahead. The Corona benchmark also uses the processor to render out a scene, and the differences were also larger here compared to most other tests. At stock the 5950X was almost 15% faster than the 3950X, then with both overclocked the 5950X is now completing the task 27% faster. Handbrake was used to convert one of my 4K laptop review videos to 1080p. Not quite as much of an improvement with the 5950X here, but still a reasonable 9% quicker time at stock, or 14% faster with the overclocks. Adobe Premiere was used to export one of my laptop review videos at 4K. I’ve tested with both VBR 1 and 2 pass. In both of these, at stock the 5950X was completing the export around 11% faster. And then a little faster with both overclocked. I’ve also tested Adobe Premiere but with the Puget Systems benchmark tool, as this tests for more things like live playback rather than just raw export times. There was less of a difference between the two chips here, with the 5950X scoring just 4% higher whether at stock or overclocked. Adobe Photoshop was also tested with the Puget Systems benchmark tool. This test tends to favour single core performance, and as we’ve seen Zen 3 is certainly delivering in that regard, so it’s no surprise that the differences are much larger in this test. At stock the 5950X was scoring 18% higher than the 3950X, and 24% higher with both overclocked. I’ve used 7-Zip to test compression and decompression speeds, and Ryzen chips typically do quite well here, especially compared to Intel processors. Zen 3 was still able to offer nice improvements over Zen 2 though, 11% higher in compression and 15% higher for decompression at stock. VeraCrypt was used to test AES encryption and decryption speeds, though both processors were scoring lower with the overclocks in place, something I’ve noticed with other tests too. Microsoft Excel was tested using the Hardware Unboxed large number crunch test, and this test had one of the biggest differences out of all apps tested. At stock the 5950X was completing the task 23% faster than the 3950X, and then a massive 44% faster with both overclocked, so if you’re after a hardcore spread sheet machine Zen 3 could be the way to go. Geekbench was seeing excellent single core improvements, the 5950X was 21% faster here at stock, and 23% faster when overclocked. The multicore difference was still nice, but not as large. The 5950X was 9% faster at stock, and 15% faster with the overclocks.
Application differences (stock vs OC)
Here’s how the 5950X compares against the older 3950X from prior generation with both processors running at stock in all of the applications just tested. The 5950X was faster in all instances, as expected, though I found it interesting that the top of the graph is littered with single core tests, it really appears that this is an area where Zen 3 shines compared to Zen 2.
Once both chips are overclocked, the margins in most of the tests increase as the 5950X pulls further ahead of the 3950X, and this is because it’s able to overclock better. This will of course come down to the silicon lottery, and some tests like Veracrypt don’t seem to interact with the overclocks well.
Power draw, thermals, clock speeds
When we look at the total system power draw from the wall with the blender test running, the 5950X was actually using less power at stock, good stuff considering it was also completing this specific task 6% faster too. With both overclocked though, the 5950X starts using more power, as this is needed to support higher clockspeeds.
Higher power draw typically results in more heat, and for the most part the temperatures align with what we just saw in the power draw graph. The 5950X was a little cooler at stock, but then with the overclocks and higher power draw the 5950X was a little warmer.
These are the average clockspeeds being reached during these same tests, so at stock both were actually averaging the same speed, again despite the 5950X using less power and performing better. We can see that by overclocking both though, the 3950X only gets a 400MHz boost, while the 5950X on the other hand is now around 800MHz higher than itself at stock, and this is where the higher power draw and temps come from.
Let’s get into the gaming results next, I’ve tested 10 games at 1080p, 1440p and 4K resolutions.
Shadow of the Tomb Raider was tested with the games benchmark tool. No difference between the two processors at 4K as it’s just less of a factor at such high resolutions. At 1440p though, the 5950X was reaching 15% higher average FPS over the 3950X, and then a larger 36% higher average FPS at 1080p. Microsoft Flight Simulator was tested in the Sydney landing challenge. Again no real difference at 4K here either, then at the other resolutions even the 1% low from the 5950X was ahead of the average FPS from the 3950X. In terms of average frame rate though, the 5950X was 22% faster at 1080p and 20% at 1440p. Assassin’s Creed Odyssey was tested with the games benchmark, and this game has one of the smallest differences between the two processors at all resolutions. The 5950X was just 4% ahead at 1080p, and 3% at 1440p. Death Stranding on the other hand is a game that I’ve found to generally see big increases with Zen 3. The 5950X was 22% faster than the 3950X in average FPS at 1080p, but then just 5% ahead at 1440p, and like the other games no real differences once we get to the highest 4K resolution. Battlefield V was tested in campaign mode, and this was another game where no major differences were recorded regardless of resolution, however this may be because the game has a 200 FPS cap, which both processors were reaching at 1080p. I had to put CS:GO to the test as AMD were claiming some big gains with Zen 3, and I was seeing this too. This game had the largest improvement with the 5950X at 1080p out of all 10 games tested, with a 42% higher average FPS. This game also had the biggest difference in average FPS at 4K too, despite the higher resolution the processor still has an important role in this test. Red Dead Redemption 2 was tested with the games benchmark, and unlike CS:GO, we’re back to seeing no major changes at 4K between the two CPUs. The 5950X was 13% faster at 1440p though, and 23% faster at 1080p when compared to the older 3950X. Rainbow Six Siege was tested using the games benchmark with Vulkan. Only small differences here, and even at 1080p the 5950X was just 2% ahead in average FPS, and honestly probably not a difference you’re actually going to notice when playing given both are near 500 FPS anyway. Far Cry New Dawn was also tested with the games benchmark, and as a game I’ve found to depend more on the CPU, it wasn’t too surprising to see a similar improvement with the 5950X even at 4K, where it was still reaching 8% higher average FPS compared to the 3950X. Call of Duty Modern Warfare was another with no major differences, at least in terms of average FPS. At 1080p the 5950X had a larger boost to the 1% low.
Over all 10 games tested, at 1080p the 5950X was around 19% faster than the 3950X in average FPS, though this may be held back a little by the 200 FPS frame cap of battlefield V, which is why it’s down the bottom. In any case, there are some nice gains on offer here with Zen 3 compared to what was available just last year. Stepping up to 1440p, the 5950X was now 10% faster than the 3950X, so still some nice gains, but at higher resolutions the processor starts to matter less and less when compared to the choice of graphics card. This is further illustrated at 4K, where the 5950X was now just 2% faster than the 3950X on average out of these 10 games tested. Only CS:GO and Far Cry New Dawn are really seeing a benefit from Zen 3 here, which makes sense given both are titles I’ve found to be primarily dependent on processor performance. 4K doesn’t really make sense to test in a CPU comparison, I just did it to illustrate that if you plan on gaming at higher resolutions then it might not be worth spending more on the 5950X, and this is why the cost per frame differences at 4K actually see the 5950X as worse value. It costs more, but wasn’t performing too much better.
Cost per frame value
Things go the other way at 1080p though, even though the 5950X costs more, the higher levels of performance on offer at this lower resolution result in a lower cost per frame value.
Honestly at this higher price point, if you’re buying today I don’t think it’s too much extra money to fork out for the improvements on offer with the 5950X. I’d be willing to pay the almost 10% more money for those single core boosts alone.
That said, the 3950X still does well in multi core workloads, especially compared to what Intel’s offering. If you already have a 3950X, upgrading could be worth it especially if time is money and you do a lot of rendering work for example, it’s really going to depend as the Zen 2 based 3950X still has a lot to offer. Upgrading might even be a simple process with just a BIOS update needed depending on your motherboard. Anyway let me know which you’d pick down in the comments.