Premiere Pro CPU performance: AMD Ryzen 9 3950X
Written on November 14, 2019 by Matt BachIntroduction
Ever since the launch of the 3rd generation Ryzen CPUs back in July of 2019, AMD and Intel have been trading blows in consumer CPU space when it comes to Premiere Pro performance. The Intel 9th Gen CPUs edged ahead in some areas, while the AMD Ryzen CPUs took the lead in others.
Now, AMD is launching one more 3rd generation Ryzen CPU - the AMD Ryzen 9 3950X. This processor features a staggering 16 CPU cores which is really starting to blur the line between "consumer" and "HEDT" (High End Desktop) processors. However, the increase in core count comes with a fairly large MSRP price of $749. For comparison, both the AMD Ryzen 9 3900X 12 Core and Intel Core i9 9900K 8 Core have a MSRP of $499. If you want more information on the specs of this new processor, we recommend checking out our New CPU Announcement: AMD Ryzen 9 3950X post.
Article updated 11/19/2019 We discovered that even though integrated graphics was enabled on the 9900K, Premiere Pro was not properly utilizing Quicksync to accelerate H.264/H.265 performance. Clearing preferences and re-installing Premiere Pro 2020 fixed the issue, and we have updated the results to reflect the performance of the Core i9 9900K with hardware acceleration enabled. This does not dramatically change the conclusion, but it did reduce the performance gap between the Core i9 9900K and the new Ryzen 9 3950X.
In this article, we want to see whether the increase in core count (and price) is worth it for Adobe Premiere Pro. However, since Intel is launching their new Core X-10000 series processors and AMD is launching their new 3rd Gen Threadripper processors in the near future, we are only going to compare the 3950X to a handful of Intel and AMD CPUs. If you want to see how it stacks up against a wider range of Intel and AMD processors, check back in the coming weeks for articles that will include the AMD Ryzen 3rd Gen, AMD Threadripper 3rd Gen, Intel Core 9th Gen, and Intel Core X-10000 series processors in a number of applications.
If you would like to skip over our test setup and benchmark sections, feel free to jump right to the Conclusion.
Looking for a Premiere Pro Workstation?
Puget Systems offers a range of workstations that are tailor-made for your unique workflow. Our goal is to provide the most effective and reliable system possible so you can concentrate on your work and not worry about your computer.
Test Setup & Methodology
Listed below are the specifications of the systems we will be using for our testing:
| AMD Ryzen Test Platform | |
| CPU | AMD Ryzen 9 3950X AMD Ryzen 9 3900X |
| CPU Cooler | Noctua NH-U12S |
| Motherboard | Gigabyte X570 AORUS ULTRA |
| RAM | 4x DDR4-2933 16GB (64GB total) |
| Intel 9th Gen Test Platform | |
| CPU | Intel Core i9 9900K |
| CPU Cooler | Noctua NH-U12S |
| Motherboard | Gigabyte Z390 Designare |
| RAM | 4x DDR4-2666 16GB (64GB total) |
| AMD Threadripper Test Platform | |
| CPU | AMD TR 2950X |
| CPU Cooler | Corsair Hydro Series H80i v2 |
| Motherboard | Gigabyte X399 AORUS Xtreme |
| RAM | 4x DDR4-2666 16GB (64GB total) |
| Intel X-Series Test Platform | |
| CPU | Intel Core i9 9960X |
| CPU Cooler | Noctua NH-U12DX i4 |
| Motherboard | Gigabyte X299 Designare EX |
| RAM | 4x DDR4-2666 16GB (64GB total) |
| Shared Hardware/Software | |
| Video Card | NVIDIA GeForce RTX 2080 Ti 11GB |
| Hard Drive | Samsung 960 Pro 1TB |
| Software | Windows 10 Pro 64-bit (version 1903) Adobe Premiere Pro 2020 (version 14.0) PugetBench V0.8 BETA for Premiere Pro |
*All the latest drivers, OS updates, BIOS, and firmware applied as of November 11th, 2019
A few notes on the hardware and software used for our testing: First, we have decided to standardize on DDR4-2933 memory for the Ryzen platform. The officially supported RAM speed varies from DDR4-2666 to DDR4-3200 depending on how many sticks you are using and whether they are dual or single rank, and DDR4-2933 is right in the middle as well as being the fastest supported speed if you want to use four sticks of RAM. In fact, this is the speed we are planning on using in our Ryzen workstations once JDEC DDR4-2933 16GB sticks are available.
The second thing to note is that we are using an unreleased version of our Premiere Pro Benchmark. This new version includes a few minor changes to the tests themselves - primarily running the CPU/GPU effects tests with a 59.94FPS sequence only rather than with 29.97 and 59.94FPS sequence. For this reason, the overall scores are not going to exactly line up with what you would see if you ran the version of the benchmark that is available for download, but they will be close. We hope to have this new version available for download within a few weeks.
Benchmark Results
While our benchmark presents various scores based on the performance of each test, we also wanted to provide the individual results. If there is a specific codec you primarily use, examining the raw results for that task is going to be much more applicable than the total scores.
Feel free to skip to the next section for our analysis of these results if you rather get a wider view of how each CPU performs in Premiere Pro.
Premiere Pro Benchmark Analysis
Premiere Pro is pretty decent at taking advantage of a higher number of CPU cores, and it really shows with the AMD Ryzen 9 3950X. Overall, it is about 17% faster than the Ryzen 9 3900X or about 13% faster than the Intel Core i9 9900K. Against the HEDT offerings from Intel and AMD, it edges in front of the Intel Core i9 9960X by a few percent and beats the Threadripper 2950X by 9%.
It is worth noting that for live playback, the performance gain is a bit less than it was overall. Here, it essentially matches the 9960X and 2950X, but is still faster than the 3900X and 9900K by about 5%.
Exporting is really where the Ryzen 9 3950X really shines. It isn't hugely faster than the (more expensive) Intel Core i9 9960X, but it is a solid 14% faster than the (also more expensive) Threadripper 2950X, 18% faster than the Core i9 9900K, and 27% faster than the Ryzen 9 3900X.
Is the AMD Ryzen 9 3950X good for Premiere Pro?
All in all, the AMD Ryzen 9 3950X is an absolutely terrific CPU for Premiere Pro. Unlike other applications like Photoshop and After Effects where the 3950X is only marginally faster than the Ryzen 9 3900X or Intel Core i9 9900K, in Premiere Pro the 3950X is around 5% faster for live playback, and a whopping 15-30% faster for exporting. For many users, that is well worth the 50% increase in MSRP, and makes it excellent in terms of both overall performance and performance for your dollar. It even solidly beats the AMD Threadripper 2950X and sneaks in just ahead of the Intel Core i9 9960X.
Keep in mind that the benchmark results in this article are strictly for Premiere Pro. If your workflow includes other software packages, you need to consider how the processor will perform in all those applications. Currently, we have articles for Photoshop, Lightroom Classic, After Effects, DaVinci Resolve, and a number of other applications.
In addition, both Intel and AMD have new processors coming out in the near future which may change the price to performance picture. We are very interested to see how the 3950X stands up against these new processors, and we will be publishing more articles as they launch. So be sure to keep a close eye on our list of Hardware Articles in the coming weeks.
Looking for a Premiere Pro Workstation?
Puget Systems offers a range of workstations that are tailor-made for your unique workflow. Our goal is to provide the most effective and reliable system possible so you can concentrate on your work and not worry about your computer.
Great test! Will you also update/test in Resolve?
Yea, we just didn't get all the Resolve testing done quite in time. Hopefully that post will be up later today.
Can't wait. Thanks for the work. You are always a great source of knowledge, when it comes to productivity :)
Is there a way to get the v0.8 benchmark or is v0.3 still the only available public beta?
V0.8 isn't up yet, mostly because we haven't verified that it works properly on MacOS yet. With the new app versions from Adobe and all the product launches going in right now, getting the new benchmark version up just has to take a back seat for a little bit. I'm hoping to have it done by the end of the month though.
Awesome, thanks for the update Matt, love the benchmark so far and seeing the 3950x numbers got me excited to start putting some other machines I have against this!
Slightly off topic, but when is the lightroom classic benchmark going to be available for download? Really wanted to bench my system, but the link is down for at least a month
We are working on getting the benchmark up for download, but it will probably be another 2-3 weeks. We were close about a month ago, then we realized Lightroom 9.0 was going to launch during Adobe MAX so we held off. It is definitely one of the more "finicky" of our benchmarks (none of these apps are made for benchmarking, so we have to do some "creative" things to get them to work). Even though we are still calling all our benchmarks BETA, I still want to make sure it at least should work for most people.
Test H264 on Intel 9900K was benchmark with QuickSync or not ?
Quicksync was enabled. From what I've seen, it actually doesn't make that big of an impact on higher-end CPUs like the 9900K. Small improvement for exporting, but not much for live playback performance. On lower-end CPUs or laptops I believe it makes a much bigger impact.
Literally just built the 3900x system for my editing computer and gaming rig.
It is so hard to make that 250$ jump when in many cases it is a 5% if lower increase. This premiere 15% increase is also nice, but it's like ok... how much does it increase productivity for price.
Tough decisions. I feel like saving the 250 (and I am one of those people who buys the ULTIMATE BEST so I don't have to buy 100 more times) is a better investment for the rest of the computer let alone, most apps don't truly take advantage of the 16 cores and run idle. The time you see it the most is export, but in live playback seems to be nominally increased if I am reading this right.
Plus with lower base clocks... I dunno am I crazy for thinking the 3900x is the best bang for buck currently?
Yep, you are exactly right. It's always hard to try to decide whether a faster component is worth the price or not, and the answer is completely different between one person and the next. Some people churn out a ton of content (or produce really long videos), and a 15% faster export can mean the difference between being able to land a specific job or not. For other people, it isn't a big deal because it only equates to a few minutes faster of a render.
I think it is always interesting to think of the cost increase in different ways as well. Going from a $500 CPU to a $750 one is a 50% increase in price, but it is a lot less in terms of your total system cost. If you are looking at a $5,000 computer, that $250 is actually only a 5% increase in price overall. Which in exchange for 5% better live playback and 15% faster exports is looking like a pretty good return.
So many different ways to look at it, and really none of them are wrong. Just unique to each situation.
Would you expect there to be any significant gains with the 3950x when using faster RAM? Another question that is kind of off topic, what do you guys think about the ASRock X570 Creator? Internal 10gbe AND 2x Thunderbolt ports would be amazing for the PCs I'm trying to build for a client.
My best guess is, that Matt thinks that you should send your client to PS if you have to ask them for advice. ;-)
I don't disagree with you. I tried... because I don't want to have to support systems when the users break things.
We won't say no if you do convince them! But to answer your questions, I would check out our "Does RAM speed affect video editing performance?" post: https://www.pugetsystems.co...
Basically, for Premiere Pro, 3200MHz RAM will probably get you another 3% performance with the 3950X, and 3600MHz would get you less than another percent. But definitely keep in mind the speeds that are officially supported by AMD: 2933MHz max with 4 sticks or 3200MHz max with 2 sticks. From our experience when doing that RAM speed testing, going beyond those specs absolutely makes the system less stable overall. We had to run our Premiere Pro benchmark several times with 3200MHz memory to get it to complete without Premiere Pro crashing, and something like 6 times to get it to complete successfully with 3600MHz RAM. If you ask me, 3% more performance isn't worth having Premiere Pro crash any more often than it will do on it's own. Honestly, I think using RAM that is beyond spec is one of the leading reasons why a lot of home-built PC's have stability issues, although I can't really collaborate that with any solid evidence beyond my own experience with higher frequency RAM.
As for the ASrock boards with Thunderbolt, I would avoid them. Thunderbolt on AMD is NOT officially certified by either AMD or Intel (we've checked with AMD directly on that), so this is ASRock doing their own thing without certification. We have a ton of customers who need Thunderbolt, and I can tell you that it can be very finicky on PC. We've ended up deciding that only Gigabyte makes a solid enough Thunderbolt product, and even then only when it is integrated onto the motherboard. Their add-on cards are still better than the other brands we've tried, but nowhere near as stable or reliable as the version they have integrated onto their boards.
Let's say one did a lot of work with H.264 and HEVC footage. Does the fact that AMD lacks QuickSync (which Intel uses for decoding and encoding these formats inside of Premiere) make much of a difference? I noticed this was an issue with some editors that are on AMD, as far as performance goes. Thoughts?
Are you absolutely sure Adobe uses QuickSync for decoding (playback)? They only use software/cpu for playback, as far as I understand. QuickSync only for export/encoding, if iGPU is enabled.
I'll answer both these questions in one reply. First, Quicksync does accelerate playback as well as exporting as long as you have it checked in Preferences -> Media -> "Enable hardware accelerated decoding". I will note that it really doesn't seem to make a difference on CPUs like the Core i9 9900K, but from what I understand it can be significant on less powerful processors like those in laptops.
As for the advantage of QuickSync in general, again not a big deal for live playback - any of these CPUs can play 150mbps single clips just fine, but none are going to be able to handle multicam. Exporting to H.264 you can get a small amount better performance from the 9900K over the 3900X or 3950X, but it isn't going to be huge (and exporting to something like ProRes the Ryzen CPUs have the lead). Also keep in mind that hardware accelerated H.264 exporting in Premiere Pro is limited to ~60mbps and is lower quality than "Software Only" mode. For some people the difference in quality is not noticeable, others think it is very significant.
Thanks for clarifying! I was under the impression it's not used at all because even on my weaker 7700K it doesn't make any difference for ~100mbps 4k60p h264 playback (with iGPU and Premiere setting enabled)
Wow - what an article!!!!
I am thinking on changing my CPU...I do simple stuff (or that is what I believe) like color grading and some stabilization in Premiere and I can never preview it in 1/2 nor 1/4. That's on my Ryzen 5 2600, 16 Gb DDR4 and a MSI Radeon RX 580 DirectX 12 RX 580 ARMOR MK2 8G.
The CPU goes to 100% when using the preview and the GPU never goes past 8%. I have reinstalled Premiere, drivers...even tried a GTX 1050 4Gb Ti and got similar results (GPU when upto 28% for the Nvidia).
I makes me want to leave this computer and get a second hand mac with Final Cut, however I wonder if it's all just down to CPU power. I got the same results after putting everything in the same fps and resolution.
Any advice?
Yea, the GPU is only going to be used in Premiere Pro when you have GPU-based effects applied. Most of those effects (like Lumetri Color) is so easy for the GPU to do, however, that it won't break a sweat even with a relatively basic video card. Even in our "GPU Stress" timelines, I don't think the GPU gets above 30% load or so - the CPU is simply the limiting factor most of the time for video editing. Since your CPU is at 100%, that is definitely your limiting factor. The new Ryzen 3rd Gen CPUs should be able to drop right into your motherboard (probably with a BIOS update first though), so that would be what to upgrade if you are looking for more performance.
I can't really comment on FCP since I haven't used it all that much. I do think that your choice of editing software should come down to which you are the most comfortable in and which offers the features you need, however. Trying to get better performance from one app vs another is really hard to do IMO since things are constantly changing.
Hey Matt! Thanks for the super informative article!
I just ran the 0.8 Beta Standard benchmark on an almost identical system (3950x, Aorus Master, 64gb 3200mhz ram, 2080ti, nvme gen 4 drive, win 10 18362) and I got almost identical results (within a point at max to yours) - except for 4K RED playback - which is around half what you're getting here with the 3950x.
My 4K RED playback score is 38.6 vs 70.9 here. Again, everything else lines up exactly.
Any idea as to why could that happen?
Amazing article. Thank you Matt.
I’m experiencing fairly poor results. Your thoughts would be much appreciated.
I built:
3950x / 1080ti / 64gb 3600 CL16 / SSD and NVME as per recommendation article from Puget. (All on different drives - project, software, cache)
I hardly get 10-15% cpu use...
gpu sits at 100% but cpu is dead...
render and export.
Ran the Puget systems Standard bench and only managed 620 overall score. Comparing your tests to mine I see that mine is extremely poor. (you guys managed 800+)
(I’m on the latest edition of Prem Pro cc 14)
Have searched the web far and wide, have tried various codecs. Still the same.
Any ideas?
It really could be anything - BIOS, drivers, thermal throttling, background tasks (especially antivirus) - but the fact that your GPU is sitting at 100% is very odd. Premiere Pro doesn't use the GPU that heavily, and even on our "GPU Effects" test I wouldn't expect to see more than 50-70% load. On the normal tests, the GPU should be only around 10-25%.
You might want to do a full GPU driver reinstall and using DDU https://www.guru3d.com/file... to clean out any GPU driver files between the uninstall and reinstall of the driver. Have you run any GPU-oriented benchmarks like Unigine Superposition or 3dMark to see if your system performs about right for those? That would help you narrow it down to a system-wide issue, or something specific to Premiere Pro.
Thank you for taking the time to reply. Much appreciated.
I have since my post noticed that when I get the above condition (10-25% cpu and 100% gpu I'm working specifically with 4K files (H265 10 bit) in 1080 sequence with Neat video V5, Red giant Colorista IV as well as Lumetri effects applied. If I paste this exact timeline to a 4K sequence the CPU usage would go up to 45-50% and GPU still 100%) At least double the cpu use.
I have updated all drivers, BIOS, VBIOS (1080ti Aorus with 3fp BIOS), Chipset, etc. Ran the Puget Premiere bench and managed a score of 647.5
Maybe this is the best this combo (3950x and 1080ti Aorus) can deliver? (Does any subscriber here have the same rig? What do you get?)
Puget systems 3950x bench gets 815.5 with the 3950x and 2080ti. So maybe the 2080ti is just THAT much stronger than the 1080ti, a whole 168 points?...
Further to the above...
Today I took a M.2 C-drive with the same Windows and PPro version installed from another rig (Samsung 960 EVO - 512gb) and fitted it into my 3950x/1080ti (usually on a HP EX950 2TB M.2) rig and booted up. Ran Puget Premiere pro and manged a near 10% better score: 691.5 with the Samsung drive installed.
It was fascinating to compare the two benches The 4K heavy effects did SO MUCH better on the Samsung M.2 hence the 691.5 score.
Example:
HP EX950 drive: 4K Heavy CPU Effects (59.94FPS),4K Heavy CPU Effects (59.94FPS) Live Playback Score,22.3,Score
HP EX950: 4K Heavy CPU Effects (59.94FPS),4K Heavy CPU Effects (59.94FPS) Export Score,6.7,Score
then on the Samsung 970 EVO:
Samsung 970EVO: 4K Heavy CPU Effects (59.94FPS),4K Heavy CPU Effects (59.94FPS) Live Playback Score,59.6,Score
Samsung 970EVO: 4K Heavy CPU Effects (59.94FPS),4K Heavy CPU Effects (59.94FPS) Export Score,24.8,Score
(The GPU scores were nearly the same)
So either the Samsung is just a better drive working better with my 3950x/1080ti or a possible software issue that exists on my HP EX950 is no longer playing its part and causing better CPU utilization or something. So weird.
So now the difference is down to 815.5 minus 691.5 = 124 points. That could be the 2080ti you guys use? But the above is a CPU test...
I can share the bench data with you?
For the record:
3D mark bench is in line with other users by a couple of points. No problems at all.
Cinebench R15 and R20 completely ion line with other users. No problem at all.
Many thanks again.
Further to my post from above if this is of any help to anyone. I went into my BIOS and disabled SMT. Ran bench again and managed 9% better score.