Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1264
Article Thumbnail

DaVinci Resolve 15: NVIDIA GeForce RTX 2070 Performance

Written on October 22, 2018 by Matt Bach
Share:

Introduction

In many ways, DaVinci Resolve is the poster child for how video editing applications should leverage the power of the GPU. While other core components like the CPU are certainly important, if you want the best FPS possible while color grading, a significant portion of your workstation budget will likely be spent on either a single powerful GPU or multiple GPUs.

NVIDIA's new RTX series cards are especially interesting since not only should they perform extremely well from the get-go, DaVinci Resolve is listed as already implementing (or planning to implement) the new features found on the RTX platform by using "Turing Tensor Cores in Resolve 15 to accelerate AI inferencing for graphics enhancement". This is a fairly generic bit of text, and it really only makes sense if you understand the two major new features found in these RTX cards: Tensor cores and RT cores.

What are Tensor Cores?

While already available on the more expensive Titan V GPU, the RTX line introduces tensor cores at a more reasonable price point. These tensor cores operate alongside the normal CUDA cores that traditionally do the heavy lifting, but are designed specifically for machine learning inference (running already created and trained machine learning models). Blackmagic has already announced that they will be using these cores, but exactly how they do so and where exactly it will improve performance is still to be seen.

What are RT Cores?

RT cores are brand new in this generation of graphics cards, and are specialized for a single type of operation: ray tracing. It is possible that Blackmagic may utilize these cores for ray tracing in the Fusion tab, but if or when they will take advantage of these RT cores is currently unknown.

If you would like to skip over our test setup and benchmark result/analysis sections, feel free to jump right to the Conclusion section.

Test Setup & Methodology

Listed below are the test platforms we will be using in our testing:

Test Hardware  
Motherboard: MSI MEG X399 Creation
CPU: AMD Threadripper 2990WX 3.0GHz
(4.2GHz Turbo) 32 Core
CPU Cooler: Corsair Hydro Series H80i v2
RAM: 8x DDR4-2666 16GB (128GB total)
Hard Drive: Samsung 960 Pro 1TB M.2 PCI-E x4 NVMe SSD
OS: Windows 10 Pro 64-bit
Software: DaVinci Resolve 15 (ver. 15.1.0.25)

To see how the new RTX cards and the RTX 2070 in particular perform in DaVinci Resolve, we tested it against a selection of cards from NVIDIA as well as AMD's Vega 64 GPU.

Our testing for DaVinci Resolve primarily revolves around the Color tab and focuses on the minimum FPS you would see with various media and levels of grading. The lowest level of grading we test is simply a basic correction using the color wheels plus 4 Power Window nodes with motion tracking. The next level up is the same adjustments but with the addition of 3 OpenFX nodes: Lens Flare, Tilt-Shift Blur, and Sharpen. The final level has all of the previous nodes plus one TNR node.

We kept our project timelines at Ultra HD (3840x2160) across all the tests, but changed the playback framerate to match the FPS of the media. For all the difficult RAW footage we tested (CinemaDNG & RED), we not only tested with the RAW decode quality set to "Full Res" but we also tested at "Half Res" ("Half Res Good" for the RED footage). Full resolution decoding should show the largest performance delta between the different cards, but we also want to see what kind of FPS increase you might see by running at a lower decode resolution.

Codec Resolution FPS Bitrate Clip Name Source
H.264 3840x2160 29.97 FPS 80 Mbps Transcoded from RED 4K clip
H.264 LongGOP 3840x2160 29.97 FPS 150 Mbps Provided by Neil Purcell - www.neilpurcell.com
DNxHR HQ 8-bit 3840x2160 29.97 FPS 870 Mbps Transcoded from RED 4K clip
ProRes 422 HQ 3840x2160 29.97 FPS 900 Mbps Transcoded from RED 4K clip
ProRes 4444 3840x2160 29.97 FPS 1,200 Mbps Transcoded from RED 4K clip
XAVC S 3840x2160 29.97 FPS 90 Mbps Provided by Samuel Neff - www.neffvisuals.com
XAVC Long GOP 3840x2160 29.97 FPS 190 Mbps Transcoded from RED 4K clip
Blackmagic RAW 4608x1920 24 FPS 210 Mbps A001_08122231_C008 Blackmagic RAW
RED (7:1) 4096x2304 29.97 FPS 300 Mbps A004_C186_011278_001 RED Sample R3D Files
CinemaDNG 4608x2592 24 FPS 1,900 Mbps Interior Office Blackmagic Design
[Direct Download]
RED (7:1) 6144x3077 23.976 FPS 840 Mbps S005_L001_0220LI_001 RED Sample R3D Files
RED (9:1) 8192x4320 25 FPS 1,000 Mbps B001_C096_0902AP_001 RED Sample R3D Files

With the addition of the "Fusion" tab in Resolve, we are also going to be including some basic tests for that tab as well. At the moment these are relatively easy projects that specifically test things like particles with a turbulence node, planar tracking, compositing, and 3D text with a heavy gaussian blur node. These projects are based on the following tutorials:

If you have suggestions on what we should test in the future, please let us know in the comments section. Especially if you are able to send us a sample project to use, we really want to hear from you!

Color Tab FPS - Raw Benchmark Results

Color Tab FPS - Benchmark Analysis

To analyze our benchmark results, we are going to break it down based the three different levels of color grading we tested. The easiest - a basic grade with 4 power windows - is not too difficult and every GPU we tested should be able to give full playback FPS in everything but RED 8K (Full Res Premium). However, each level up should show more and more of a difference between the different cards.

The "Score" shown in the charts is a representation of the average performance we saw with each GPU for that test. In essence, a score of "80" means that on average, the card was able to play our project at 80% of the tested media's FPS. A perfect score would be "100" which would mean that the system gave full FPS even with the most difficult codecs and grades.

In terms of pricing, the new RTX 2070 8GB should be just a bit more expensive than the GTX 1080 8GB, but quite a bit cheaper than the GTX 1080 Ti. Because of that, we feel that the most valid comparison between the RTX 2070 and the previous GTX line is to look at it versus the GTX 1080. If you also want to look at the RTX 2080 and RTX 2080 Ti, the most straight-forward comparison is against the GTX 1080 Ti and Titan Xp respectively.

While there really isn't much of a difference between any of the cards we tested with our lightest level of grading (since they all achieved near full FPS), the RTX 2070 did extremely well with the two higher grades. Where the RTX 2080 and 2080 Ti only saw a 10-25% increase in performance compared to the previous generation cards with the more difficult color grades, the RTX 2070 was about 40% faster than the GTX 1080. This means that it is a much closer match to the GTX 1080 Ti than it is to the GTX 1080.

Fusion Tab FPS - Raw Benchmark Results

Fusion Tab FPS - Benchmark Analysis

Fusion is relatively new to our DaVinci Resolve testing, and so far we haven't been too impressed with how well it takes advantage of the GPU. To be fair, we are not using media footage in these projects that is particularly difficult to process, but given the FPS we saw in each project we doubt that that having multiple GPUs would significantly improve performance even if you are using 8K RED media.

Whether it is due to our test projects or simply how much more CPU dependent Fusion is, we really didn't see much of a difference with any of the cards we tested. However, we will again point out the future potential of the RTX cards for Fusion. The new RT cores in particular could be very interesting and while we have no idea if/when they will be used, it is certainly something to keep in mind.

Is the RTX 2070 good for DaVinci Resolve?

Compared to the previous generation GTX 1080 8GB, the new RTX 2070 8GB is on average about 20% faster although that increases to roughly 40% faster with more difficult levels of color grading. This makes the RTX 2070 a terrific entry option for DaVinci Resolve, even ignoring the new Tensor and RT cores which may be utilized in the future.

NVIDIA GeForce RTX 2080 & 2080 Ti DaVinci Resolve 15 Benchmark

There isn't really much mystery here, the RTX cards are simply really good for DaVinci Resolve. Especially if you consider the fact that Blackmagic has already stated that they will be taking advantage of the Tensor cores in Resolve 15, it makes these cards incredibly attractive for use in a color grading workstation. However, not everything is perfect and there is one big issue that you may have to work around: the cooler most RTX cards use.

To put it bluntly, the style of cooler used on the reference cards from NVIDIA and most 3rd party manufacturers is not good for multi-GPU configurations. They can be excellent for a single GPU, but if you want two or more cards the design is sub-optimal. The issue is that the cooler does not exhaust out the back of the system so the hot air generated by the cards is simply recycled inside the system over and over. We didn't see a significant performance drop in our Resolve testing, but we are also testing in an ideal environment with relatively short clips. In GPU-heavy applications like OctaneRender and Redshift, however, we have seen up to a 30% performance drop over time using multiple reference RTX cards. This doesn't mean you cannot use the RTX cards in multi-GPU configurations, but rather that you should try to use cards with a "blower" style cooler that is designed to vent the heat directly outside of the chassis.

If you are interested in how the RTX cards perform in other applications, be sure to check out our recent Video Card articles as we have (or are working on) a number of other articles for the RTX 2070, 2080, and 2080 Ti.

Tags: DaVinci Resolve, GeForce, RTX, 2070, 2080, 2080 Ti, 1060, 1070, 1070 Ti, 1080, 1080Ti, Radeon, Vega
Mavanti

Your tests are great but I wondered why you never tested the AMD Radeon
Pro WX 9100 16GB. Its not much more expensive than the RTX 2080 Ti. I
have no personal experience of the WX 9100 16GB but it seems like it out
performs the RTX 2080 Ti in terms of real time noise reduction filters
and multiple grading/FXs.

I reference this video as an example

https://youtu.be/W1RVrIJcxeo

I have been through all of your video card
bench marks test I believe but no card you have tested was able to do
that (unless I have misunderstood some thing).

Personally I am trying to decide between buying the best GPU I can possibly afford and reducing the rest of the cost of my X399 Threadripper build or just getting the RTX 2070. All that really matters to me is Davinci Resolve real time playback at 4k but in your bench marks even the RTX 2080 TI doesn't seem able to do that with a lot of FXs. The only card I have seen that seems to be able to is the WX 9100 16GB but no one seems to be using it for Davinci Resolve. There is also now a Radeon Pro WX 8200 which is another pro workstation card that is also less expensive than a RTX 2080 TI.

The AMD "gamers" cards don't have the horse power/RAM to compete with the RTX 2080/TI but given that there are AMD PRO work station cards that are cheaper or not much more than a RTX 2080 TI it would be interesting to see how good the Vega architecture pro work station cards are in Davinci Resolve.

Posted on 2018-10-31 05:19:56

Always, always, take claims made by hardware manufactures (whether it is AMD, Intel, NVIDIA, or anyone else) worth a grain of salt. They all love to say things like "real time" without really getting into what they are doing. Notice how he talks about adding noise reduction and blurs, but never mentions the resolution, codec, or FPS of the footage they are using.

That said, we are planning on testing the Radeon Pro series, we are just waiting for the RTX Quadro cards to become available so we can do a full "workstation" class GPU roundup. Something to keep in mind though, the Radeon Pro WX 9100 is almost identical to the Radeon Vega 64. In fact, according to AMD, the Vega 64 should actually be a hair faster. All you get from the WX9100 is more VRAM (which is useful for 8K footage, but won't impact raw performance) and presumably higher reliability and driver/software support. We'll have to wait for our benchmarks, but from everything I see I can't imagine how the WX9100 will be able to beat the RTX 2080 Ti - it should be closer to the RTX 2070 than anything else.

Posted on 2018-10-31 16:10:26
Mavanti

Yes of course you are right. It didn't even occur to me that they could be running any thing other than RAW 4k, as that would be a total cop out and pointless demo but actually they could have been running 1080 at 24fps ! And thinking about it and looking at the stats your probably right about the Radeon Pro WX 9100 stats are basically a match for the Radeon Vega 64. But I still have a little voice wondering if they were running in 4k with a RAW codec etc but the cards were linked some how and that greatly increased how well they work in Davinci Resolve. Just a me wondering, what if. I have seen some GPU boxes, Sonnet I think which I thought were running 2 x Radeon Pro WX 9100 operating together I believe but I could be wrong.

I very much look forward to your workstation roundup and if possible some other dual card set up tests, if you think it worth while. Personally I don't really understand how a second card can accelerate Davinci performance, if they are not linked some how. Except if one is not running any monitors, just dedicated to the interface and so being able to dedicate all of that cards resources just to that.

Posted on 2018-10-31 17:49:12

I don't know all the specifics (most software developers keep their magic secret), but the simple version is that for lots of things, the GPUs don't actually have to talk to each other at all. It all depends on the types of calculations being done and whether they impact each other.

Think of it like this: for 4K footage we have about 8 million pixels that need to have their color tweaked by DaVinci Resolve. Changing the color on each one is completely independent of the other pixels, so the system can easily just split them right down the middle, send half to one GPU and half to the second. Each GPU does all the number crunching for their share of the pixels, then reports it back. This is a perfect scenario, and in real life it is much messier, but you get the idea. And in fact, the GPUs can talk to each other if they need it, it just has to go over the PCI-E bus. There are new technologies like NVIDIA's NVLink that should make this talking way faster if it is needed, but that requires software support that Resolve doesn't have yet.

The thing with AMD cards is that on the Radeon side, you do get more bang for your buck, but even the Vega 64 is more of a mid-range card so you can get way higher total performance from NVIDIA (although at a higher cost to match). This applies to the Radeon Pro and Quadro lines as well, it is just that the price points are dramatically higher. A pair of WX9100 is pretty good if you need workstation cards (which really you don't for Resolve but that is another topic entirely). You would have to do a pair of Quadro P5000 to get about the same performance which would actually be a bit more expensive. You are talking about ~$3,000 either way, which even for professional colorists is going to bring them to the limits of most budgets. So while you could go up to a pair of P6000 or GV100 NVIDIA cards, I think that is outside the vast majority of budgets.

Posted on 2018-10-31 18:00:45
Mavanti

Thanks for your replies and insight Matt. I am personally just going to buy a RTX 2070 and worry about getting better cards next year. I did see that there was a new AMD work station card and this time they put the testing details on the end of this video below. It doesn't mean much to me but I thought you might be of interest to you.

https://www.youtube.com/wat...

Posted on 2018-11-02 03:15:29

Definitely good to see them putting testing details. It's a big mass of text, but at least you can see what they did. Good job AMD!

That said, it looks like most of what they focused on in this is Blender Cycles which is a GPU-based 3D Rendering and Ray Tracing engine. That type of work is very efficient at using the GPU, but not really applicable for things like DaVinci Resolve or anything else outside of that one application. That video isn't wrong or anything, it just should have said "Up to 30% faster than the competition [in Blender Cycles]". That doesn't sound quite as impressive, however, so it completely makes sense why they shortened it. That is just what everyone does in those kind of hype videos.

Honestly, this is kind of why we do all this in-depth software by software benchmarking. While some software do behave similarly, we've found that the only way to truly know how a piece of hardware will perform for X software is to test it in that specific software. Two applications could be made by the same company and look similar (Premiere Pro and After Effects are a good example), but end up with drastically different performance with the exact same hardware.

All in all, I think you are right to get a RTX 2070 for DaVinci Resolve.

Posted on 2018-11-02 17:14:57
edin

Lol you can see in that same video that he is struggling with 10 serial nodes of small blur and its at 18fps then he keeps deleting nodes and not improving the frame rate. Bad bad .... I think from what I have seen that most cost effective and future proof is RTX2070 the only thing that scares me is 8gig of ram.... i wish it was at 11gig. What do you think Matt? I have to choose between 1080ti or rtx2070 in a z800 workstation. I have tested Vega64 and was not impressed it was slightly faster then gtx1070

Posted on 2018-11-13 10:45:51

I think 8GB of VRAM should be completely fine if you are working with HD or 4K media. Once you get to 6K or 8K, however, I would probably jump up to a 11GB card if you can.

Posted on 2018-11-13 17:50:30
OneCharmingQuark

Could you test the Quicksync and NVENC GPU based rendering from Intel and Nvidia in Resolve? I know you did some limited testing for Quicksync in Premiere already. The quality of NVENC is supposed to be quite good on recent versions, so it would be nice to see how much you are really losing in image quality for the extra performance.

Many people also tend to compare FCP X with Quicksync enabled to Premiere Pro and Davinci Resolve without either Quicksync or NVENC. This doesn't give Windows workstations, like yours, a fair shake. Comparing a Mac product with FCP X in a like-for-like situation with Resolve or PP would give a more accurate benchmark. I think this will make more people consider using a PC. Many people don't realize that the performance gap can be significantly reduced, or that GPU rendering can have negative effects on video quality. This biases them against workstations like yours, and in favor of iMacs or Hackintoshes. You can show how much of the "optimization" advantage of Macs has gone away in recent software updates, or that utilizing Quicksync in FCP X results in an inferior product compared to one of your high-end machines running Resolve or PP.

Posted on 2018-10-31 17:31:24

From what I understand, Resolve supports H.264/HEVC decoding on NVIDIA GPUs, but doesn't support Intel Quicksync for the same. We actually have that on by default in all our Resolve testing since that is the default option Blackmagic sets.

Premiere, however, I believe is CPU decoding only, not through the GPU. That is on our list to look at closer, but to be honest, most of our customers end up wanting the higher-end Intel X-series which doesn't support Quicksync. That just means it is a bit lower on the to-do list than some of the other projects we have going on right now.

Posted on 2018-10-31 17:49:59
OneCharmingQuark

Maybe you're right. Judging from the Blackmagic forums there were people implying that the hardware acceleration included Quicksync, but I can't find anything from Blackmagic themselves.

On another topic, it would be nice to see video stabilization performance as well as more 60 fps footage tested. Video stabilization seems to prefer single core performance, but there isn't much information out there on exact GPU and CPU scaling. Higher frame rate footage seems more useful for benchmarking when so much hardware gives a full 30 fps in many of the tests. I see you've used higher frame rates in some of you older tests.

Posted on 2018-10-31 19:06:20