Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1898
Article Thumbnail

Adobe Premiere Pro - NVIDIA GeForce RTX 3080 & 3090 Performance

Written on September 24, 2020 by Matt Bach
Share:

TL;DR: NVIDIA GeForce RTX 3080 & 3090 performance in Premiere Pro

While the average editor likely won't notice much of a performance gain with the new NVIDIA GeForce RTX 3080 or RTX 3090 video cards, for certain situations they can provide a healthy boost to performance. In an ideal situation where you are using multiple GPU-accelerated effects, the RTX 3080 10GB is around 10% faster than the more expensive RTX 2080 Ti, or 20-40% faster than the RTX 2080, 2070, and 2060 SUPER cards. The RTX 3090 24GB is even faster, beating the RTX 3080 by 10%, the RTX 2080 Ti by 21%, and 30-60% faster than the 20-series SUPER cards.

These new GPUs are also significantly faster than AMD's current offerings, with the RTX 3080 beating the AMD Radeon RX 5700XT and Vega 64 by around 75% and the RTX 3090 being almost 2x faster in our GPU Stress project.

Introduction

On September 1st, NVIDIA launched the new GeForce RTX 30 Series, touting major advancements in performance and efficiency. While gaming is almost always a major focus during these launches, professional applications like Premiere Pro should also see some significant improvements as well. Adobe has been adding more and more GPU-accelerated effects and features over the last few years, which has made the GPU an increasingly important component in editing workstations.

Premiere Pro GPU Performance Benchmark - NVIDIA GeForce RTX 3080 10GB & RTX 3090 24GB

If you want to see the full specs for the new GeForce RTX 3070, 3080, and 3090 cards, we recommend checking out NVIDIA's page for the new 30 series cards. But at a glance, here are what we consider to be the most important specs:

VRAM CUDA Cores Boost Clock Power MSRP
RTX 2070S 8GB 2,560 1.77 GHz 215W $499
RTX 3070 8GB 5,888 1.70 GHz 220W $499
RTX 2080 Ti 11GB 4,352 1.55 GHz 250W $1,199
RTX 3080 10GB 8,704 1.71 GHz 320W $699
Titan RTX 24GB 4,608 1.77 GHz 280W $2,499
RTX 3090 24GB 10,496 1.73 GHz 350W $1,499

While specs rarely line up with real-world performance, it is a great sign that NVIDIA has doubled the number of CUDA cores compared to the comparable RTX 20 series cards with only a small drop in the boost clock. At the same time, the RTX 3080 and 3090 are also $500-1000 less expensive than the previous generation depending on which models you are comparing them to.

While it is a bit odd that the RTX 3080 has less VRAM than the 2080 Ti, all three of these new cards should all be capable of working with 4K timelines in Premiere Pro. The RTX 3080 and 3090 (with 10GB and 24GB of VRAM respectively) should also have no trouble with 8K timelines. However, keep in mind that the actual performance in those kinds of workflows depends more on the power of your CPU than anything else.

With the launch of the RTX 3090, we can update our previous Adobe Premiere Pro - NVIDIA GeForce RTX 3080 Performance article with results for the 3090, but since the RTX 3070 is not launching until sometime in October, we cannot include it at this time. However, we are very interested in how the RTX 3070 will perform, and when we are able to test that card, we will post another follow-up article with the results.

Premiere Pro Workstations

Puget Systems offers a range of powerful and reliable systems that are tailor-made for your unique workflow.

Configure a System!

Labs Consultation Service

Our Labs team is available to provide in-depth hardware recommendations based on your workflow.

Find Out More!

Test Setup

Listed below is the specifications of the system we will be using for our testing:

*All the latest drivers, OS updates, BIOS, and firmware applied as of September 15th, 2020

To test each GPU, we will be using the fastest platform currently available for Premiere Pro - most notably the AMD Threadripper 3970X. Since Premiere Pro utilizes the CPU so heavily, this should minimize the impact of the processor and allow each GPU to perform at their fullest potential.

One thing we need to specifically point out is that we are using the latest beta build of Premiere Pro because the currently available production build (14.3.2) has a bug preventing it from properly processing R3D media with the RTX 30-series cards. This is a known issue, with the fix being implemented in the Premiere Pro beta. However, while one of the major features of the current beta is the addition of GPU-based H.264/HEVC decoding, we actually saw significantly lower performance with this feature enabled. This is likely due in part to how powerful of a CPU we are using, but we plan on investigating why hardware decoding is giving us worse performance in a future article. For now, we left hardware decoding disabled which, from what we can tell, results in nearly identical performance to Premiere Pro 14.3.2.

For the testing itself, we will be using the "Standard" preset of our PugetBench for Premiere Pro benchmark. This tests a number of different codec types (H.264, ProRes 422, and RED RAW), but also includes a dedicated "Heavy GPU Effects" test that is meant to put as much load on the GPU as possible while still staying within the realm of what someone might actually do in the real world.

If you wish to run our benchmark yourself, you can download the benchmark and compare your results to thousands of user-submitted results in our PugetBench database.

Raw Benchmark Results

While we are going to go through our analysis of the testing in the next section, we always like to provide the raw results for those that want to dig into the details. If there is a specific codec or export setting you tend to use in your workflow, examining the raw results for that task is going to be much more applicable than our more general analysis.

Overall Premiere Pro Performance Analysis

In applications like Premiere Pro where the GPU is secondary to the CPU, many reviewers like to focus on extreme situations that heavily load the GPU. However, we first want to start off by looking at the overall performance we saw from our Premiere Pro benchmark with each GPU in order to show what most users would likely experience in their day-to-day work.

Looking at the Overall Standard Score, the new RTX 3080 and RTX 3090 do top the charts, but they effectively score the same as the Titan RTX, 2080Ti, and dual RTX 2080 Ti. This is a pretty good result for the RTX 3080 considering that it is less expensive than any of those options (often significantly so), but the 3% performance gain from the RTX 3090 is going to be hard for anyone to justify.

Even compared to the RTX 2080 SUPER, the 3080 and 3090 are only about 5-8% faster overall, which is just beyond the margin of error for this test. However, if you are still using an old GTX 1080 Ti, the performance gain is close to 30% which should result in noticeably better performance for most users.

Keep in mind that the scores shown in the charts above include quite a few tests that are heavily CPU limited. Playing ProRes and H.264 footage (currently) does not utilize the GPU, and neither does exporting to ProRes, which easily covers more than half the tests. This is still a fairly good indication of what the average Premiere Pro user may experience overall, but we can dive into specific situations where a more powerful GPU should net you more significant performance gains.

Exporting to H.264 Analysis

NVIDIA GeForce RTX 3080 10GB & RTX 3090 24GB Premiere Pro GPU Encoding performance benchmark

The chart above is not an official part of our Premiere Pro benchmark at the moment but is a calculated score from the H.264 export tests. As we showed in our recent Premiere Pro 14.2 H.264/H.265 Hardware Encoding Performance article, the amount of speedup you can see by utilizing the recently added hardware encoding feature varies greatly depending on your source media and export settings. In this case, we are exporting 4K media to UHD H.264 40mbps, which, coincidentally, is where we saw the largest performance gains.

Here, the performance advantage of the new RTX 30-series cards is a bit larger, with both cards beating the RTX 2080 SUPER by 12%, the RTX 2080 Ti by 6%, and even edging out over the dual RTX 2080 Ti configuration by a small 3%. Once again, however, both the RTX 3080 and 3090 perform almost the same. They both are certainly much faster than the AMD Radeon GPUs or the older GTX 1080 Ti, but this is a feature that has a fairly dramatic soft cap to GPU performance so there is less to gain as you go up in raw GPU performance.

Heavy GPU Effects Analysis

NVIDIA GeForce RTX 3080 10GB & RTX 3090 24GB Premiere Pro GPU Effects benchmark performance

Our heavy GPU effects test looks at live playback and export performance with sequences that have a large number of GPU-accelerated effects including Lumetri Color, Ultra Key, Sharpen, Gaussian Blur, Basic 3D, Directional Blur, VR Digital Glitch, and VR De-Noise. Because it is designed to maximize the load on the GPU and minimize the load on the rest of the system, this should be close to the maximum possible performance difference you will ever experience between different GPUs in Premiere Pro.

Here, the RTX 3080 beats the RTX 2080 SUPER by a solid 20%, the RTX 2080 Ti by 10%, and even the dual RTX 2080 Ti configuration by 5%. The RTX 3090 was even faster, out-performing the new RTX 3080 by 10%, the RTX 2080 Ti by 21%, and the dual RTX 2080 Ti by 15%

Compared to the more modest GPUs we tested, the performance gains can be very significant. The RTX 3080 and 3090 were more than 40% and 60% faster than the RTX 2060 SUPER respectively, and the RTX 3090, in particular, was close to twice the performance of the AMD Radeon RX 5700XT and Vega 64.

Compared to the older GTX 1080 Ti, the results get a bit... odd. This is one of the first times we have run this exact test on the GTX 1080 Ti, and we were surprised to see how poorly that specific card performed. We verified this result with multiple cards, drivers, etc., but it is likely that one of the effects we are using simply does not run very well on the older GTX series cards. Because this is likely an artifact caused by a single effect, you probably won't actually see a nearly 5x increase in performance going from a GTX 1080 Ti to an RTX 3080 or 3090 unless you happen to be using one of these "problem" effects. But if you use a large number of GPU-based effects, at least a 2x increase in performance is not an unreasonable expectation.

How well does the NVIDIA GeForce RTX 3080 & 3090 perform in Premiere Pro?

While the average editor likely won't notice much of a performance gain with the new NVIDIA GeForce RTX 3080 or RTX 3090 video cards, for certain situations they can provide a healthy boost to performance. In an ideal situation where you are using multiple GPU-accelerated effects, the RTX 3080 10GB is around 10% faster than the more expensive RTX 2080 Ti, or 20-40% faster than the RTX 2080, 2070, and 2060 SUPER cards. The RTX 3090 24GB is even faster, beating the RTX 3080 by 10%, the RTX 2080 Ti by 21%, and 30-60% faster than the 20-series SUPER cards.

These new GPUs are also significantly faster than AMD's current offerings, with the RTX 3080 beating the AMD Radeon RX 5700XT and Vega 64 by around 75% and the RTX 3090 being almost 2x faster in our GPU Stress project.

Overall, this makes the new RTX 3080 and 3090 an excellent option if you are configuring a new editing workstation, and can be a solid upgrade option depending on your workflow and what GPU you currently have. In general, the more GPU accelerated effects you use, the more of a benefit you will see with the more expensive RTX 3090 24GB card, but the extra VRAM also makes it a good idea if you want to work with 8K media. If you only work with 4K media, you may want to wait for the upcoming RTX 3070 which is slated to launch in October.

As always, keep in mind that these results are strictly for Premiere Pro. If you have performance concerns for other applications in your workflow, we highly recommend checking out our Hardware Articles (you can filter by "Video Card") for the latest information on how a range of applications perform with the new RTX 3080 and 3090 GPUs, as well as with different CPUs and other hardware.

Premiere Pro Workstations

Puget Systems offers a range of poweful and reliable systems that are tailor-made for your unique workflow.

Configure a System!

Labs Consultation Service

Our Labs team is available to provide in-depth hardware recommendations based on your workflow.

Find Out More!
Tags: Hardware Acceleration, hardware encoding, NVIDIA, NVIDIA vs AMD, AMD, Vega 64, Radeon RX 5700 XT, RTX 2060 SUPER, RTX 2070 SUPER, RTX 2080 SUPER, RTX 2080 Ti, Titan RTX, RTX 3080, RTX 3090
Gerardo Gherman

hey guys im currently using a premiere pro custom build pc and using a 1080ti thinkin... 1080ti raw power is great because its has lots of cuda cores etc etc,
but now that i see your charts, it looks like it has lots of power but old power. so a 2060/2070/2080 will outperform a 1080ti in premiere and danvinci is that right?

Posted on 2020-09-25 16:01:23

Yes it will, but remember that it will really only do so when using GPU accelerated effects and exporting with hardware encoding. If you look at the individual live playback scores for example, (https://www.pugetsystems.co..., the 1080 Ti scores about the same as the 2080 Ti in everything except the GPU effects test and the 4K RED test (since that uses the GPU for debayering).

If you use multiple GPU-accelerated effects, want faster export times, or work with RED footage, upgrading to a newer card should bring a noticeable increase in performance.

Posted on 2020-09-25 17:21:25
Gerardo Gherman

currently owning a i9 7920x and i only do adobe premiere! im thinking on upgrading!!! cpu also
i dont care about exporting times, just.. good playback h264-h265 prores and raw files
would u recomend me to stay in x299 and maybe upt a 14-16 cores?
or switch to ryzen 3950x 16 cores? i dont want threadripper. i dont kno why i had a bad experience playing h264 files on the original 1950x and it made me buy the i9 7920x
what would be the best choise for live playback performance?

Posted on 2020-09-25 17:47:35

Check out this post for CPU performance: https://www.pugetsystems.co... . You can also run our benchmark on your system and compare it to all the results in our public database: https://www.pugetsystems.co...

Edit: Forgot to mention that the first gen Threadripper CPUs we pretty bad for most things actually. Tons of cores (for the time), but the per-core performance was terrible. Almost all apps are going to be fairly lightly threaded for "active" tasks like playing footage, navigating around the UI, etc., so while those first gen Threadrippers were great for some tasks, they weren't very good for a lot of stuff. The 3rd gen Threadripper, however, are really good at pretty much everything. Just quite a bit more expensive than Ryzen or the Intel options.

Posted on 2020-09-25 18:01:55
Rob Englander

Great review Matt.. I currently run a newly upgraded editing machine... with (X299 Deluxe II - i9 10940X - 64Gb - ASUS RTX2070 Super) ... I'm curious as to whether to ditch my newly acquired 2070S for the 3080 for what appears to be (in the video creation dept.) not a lot of performance increase .. 3D CGI etc., gaming etc. seems to be the main focus here.. 3090 is a tad pricey.. cheers Rob

Posted on 2020-09-27 06:30:17

If you stay in the Adobe suite, it definitely isn't worth an upgrade unless you want that 15% faster H.264/HEVC export performance or really like your GPU accelerated effects. If you also use Resolve, however, if could be worth it, but it completely depends on what you do. Even Resolve will only be able to take advantage of the 3080 if you are actually loading up the GPU with OpenFX, noise reduction, etc.

Posted on 2020-09-28 16:51:57
Lauri Pesur

What about the 16series GPU-s ? Where would they slot in the charts?

Posted on 2020-09-30 09:59:55

We don't regularly test older cards like that, but there are plenty of results in the benchmark database you can check out and compare: https://www.pugetsystems.co...

Posted on 2020-09-30 16:02:37
Andrew Salmon

Perhaps I missed it -- this tech is still new to me; I'm researching to build a video-editing workstation for my wife -- but could you explain how the RTX 3000 series does with h.265? Is this the encoding/decoding section? She's got her eye on the Canon EOS R5, but from what I understand, the codecs are extremely hardware-intensive. Do the new GPUs handle the demanding codecs better than the old GPUs? Thanks for these helpful articles!

Posted on 2020-10-03 15:43:47
Tom P

Doing a google on 'h265 editing system' will bring up lots of useful guidance. If I shot h265, I would transcode to prores 422 to make it less taxing on my editing system.

Posted on 2020-10-04 13:27:39
Andrew Salmon

Yes, transcoding seems to be the general response. If that's the answer, so be it. :) No one seems to have a clear answer for whether or not the RTX 30xx series does a better job of handling demanding codecs, however. Perhaps it's not designed for that. or perhaps I'm asking the wrong question? I wish I knew more! Some people have pointed to an iPad pro as having the proper hardware support, but I'm not sure if that's with the full 10-bit file. I'm trying to educate myself, but I'm definitely coming from a place of profound ignorance when it comes to video editing and hardware requirements. Thanks for taking the time to reply!

Posted on 2020-10-05 13:57:26

Encoding is fairly straight-forward, performance for that is in this section: https://www.pugetsystems.co...

Decoding of H.265 (HEVC) and H.264 is a bit harder right now. In the current live version of Premiere Pro, the GPU isn't used at all to accelerate performance - only Intel Quicksync can be used. But, in the beta, GPU accelerated decoding of H.264/HEVC was added and the performance is terrific from what we have seen so far. In fact, we are in the midst of a bunch of testing for that and should have an article/video up in the next few weeks.

I don't think the GPU itself will make a huge difference for decoding performance - although a 3000 series card will certainly be better than a 1000 or 2000 series card. I think for most people, this update will make working with H.264/HEVC directly more than possible. Transcoding and proxies may still be worthwhile for some workflows, but for others it should be fine to skip that.

Posted on 2020-10-05 17:13:17
Andrew Salmon

Thank you so much for your reply! I'll keep an eye out for the article. I appreciate your work. :)

Posted on 2020-10-05 17:17:32

Hi Matt

Thank you so much for all the testing you do. Having discovered it fairly recently, I'm so happy to have done so and follow it avidly :)

I am researching building a new PC at the moment and have a few questions which you might have insight into. I am looking at: CPU - 5950X vs TR 3960X vs waiting for next gen Threadripper vs Intel i9-10 series. GPU - 3080 vs 3090. Ram 64GB VS 128GB.

I am 90% Premiere, 10% Resolve (for grading with noise reduction). So 3080 seems more logical as the extra Resolve performance would only affect me 10% of the time.

1) Are you able to share anything from your tests so far on the beta H.264 hardware GPU decoding in terms of either playback, timeline responsiveness or export? Especially in terms of what you said in the 14.2 encode piece: “This means that if you work with a lot of H.264/H.265 media and want to minimize export times, you should stick with an Intel-based system with Quicksync - at least until Adobe adds GPU-based H.264/H.265 decoding.”

In regard to export, the beta guys at Adobe told me they aren’t really focussed on exports because the decode part is not the bottleneck, though if anyone could test that it’s you. Responsiveness is obviously hard to put a figure on, but still if you have any impression I’d be glad to hear it. Playback I imagine does show in your benchmark.

2) Referencing the 2015.4 Storage Optimization article (which still gets shared on forums & is so helpful) - I am wondering if you’ve done tests on the different types of NVMe drives - SLC/MLC/TLC/QLC and PCI-E 3.0 vs 4.0 on both the boot drive and the cache drive (assuming a 2 drive set up for now) and how they affect Premiere? I am wondering at what point the extra speed is not worth paying for in a system and even whether it’s fine to get the slowest NVMe for the sake of more storage space. (Re reliability, even with a QLC I don’t see getting close to its rated endurance.)

3) If you did have two different NVMe drives, would you put the faster of the two as boot or cache drive?

4) If I’m looking at CrtstalDiskMark tests to compare drives, is there one of the tests that most represents what Premiere would be doing to say the cache drive? (e.g. RND4K Q32T1)

[A couple of reference CDM tests for one range of NVMe at both ends of the scale:

Sabrent Rocket 4.0 $200 for 1TB https://www.legitreviews.co...
Sabrent Rocket Q $120 for 1TB https://www.legitreviews.co...]

5) Do you have a guesstimate on the availability of the 5960X TR?

Sorry for so many questions!

Posted on 2020-10-10 11:52:20

Hey Nicholas, I'll try to give you a quick answer on each of your questions:

1) GPU decoding is a huge deal in Premiere Pro! We have some testing done, and are just waiting for it to be out of beta to post. It is pretty much the same performance as Quicksync, so it really just means you don't need to go with Intel to get great performance with H.264/5 media anymore. The biggest differences we saw were in reverse playback, 2-4x forward playback (so J/K/L editing pretty much), scrubbing, and multi-cam performance.

2) Once you get above a normal SATA SSD in terms of speed (say around 500MB/s), most things are going to be CPU bottlenecked rather than storage. I think it is still a good idea to go with an NVMe for a cache drive if you can just for that little bit of performance, but I wouldn't stress too much about PCIe Gen 3 vs 4, or the different types of drives. Honestly, I would just recommend going with a Samsung drive and calling it good. Their drives have been insanely reliable for us, and that more than anything else is top priority for storage IMO.

3) I would have it as the boot drive unless you also use After Effects (since any bit of performance you can get for disk cache helps)

4) Not really sure, but I would guess random 4K with a que depth of 1

5) Sorry, anything that could be taken as NDA info we have to keep under wraps - even if we don't actually know anything and are just guessing. Rumors and "leaks" are already a big problem and we don't want to contribute to them.

Posted on 2020-10-12 20:06:21

Thanks so much Matt, that's v helpful and totally undestand re Threadripper next gen. I'm glad that 2x forward playback is improved - that's my preffered way of watching interviews. If I didn't have to make proxies of H.264 footage on every job it would be a massive time saver.

Always hard to know when to leap in with a new computer as there's always one thing round the corner. I've started keeping my own spreadsheet of performance vs cost from your benchmarks and others like Cinebench. Interestingly, on UK pricing the i9 10850K is leading the pack and I can see you have that in your 4K system, though it is quite marginal over the 3950X and the estimated/leaked 5950X results (https://www.cpu-monkey.com/.... Looking forward to your proper Premiere benchmarks for the 5950X.

Going to the 3960X is a bit of a drop in price/performace, even with current reduced pricing (unless working with a lot of RED footage which I don't) - but then again sometimes that's well worth it when the cost is a once off and you're saving time day in day out doing exports and proxies.

I was wondering if you're planning on introducing hardware H.264 export to the benchmark? I can see that it's tough - you need to have consistecny for the whole thing to work & I appreciate that for the time being you include the results separately. I was also wondering about H.264 -> ProRes i.e. making proxies from H.264 which I do for most jobs and I think is a very common activity for Premiere users. I don't think that's a part of the test at the moment is it? This also might be improved by the new H.264 decoding which would be interesting to see.

Posted on 2020-10-14 07:23:35