DaVinci Resolve Studio – AMD Radeon RX 6900 XT Performance

Introduction

For the last part of 2020, both NVIDIA and AMD have been slowly releasing a number of new GPUs. One of the most recent was the AMD Radeon RX 6900 XT 16GB which officially launched on December 8th, but due to extremely tight supply, we have not been able to get our hands on one to test until just recently. In fact, we still have not been able to purchase a card ourselves, but our friends at Linus Media Group were able to loan us a card so we could perform our testing.

While gaming is almost always a major focus during these GPU launches, professional applications like DaVinci Resolve are becoming more and more important. In fact, if you view the product page for the 6900 XT, you will note that they are starting to include Content Creation performance from benchmarks like our very own "PugetBench for Premiere Pro".

If you want to see the full specs for the latest GPUs from NVIDIA and AMD, we recommend checking out the NVIDIA GeForce RTX 30 Series and AMD Radeon RX Graphics Cards product pages. But at a glance, here are what we consider to be the most important specs:

While specs rarely line up with real-world performance, the biggest thing to note is that the AMD Radeon 6900 XT 16GB is at the top of the product stack from AMD and has an MSRP that is $300 higher than the NVIDIA GeForce RTX 3080 10GB, or $500 lower than the GeForce RTX 3090 24GB. Based on price alone, this means that we (hopefully) will see better performance from the 6900 XT than the RTX 3080.

However, something else important to notice is that the Radeon 6000 series cards have a ton of VRAM for their price. Unfortunately, even 8GB of VRAM is usually plenty for most 4K timelines, but even 16GB is not quite enough if you want to work with 8K timelines. For that, we typically find that you want a GPU with at least 20GB of VRAM which largely limits you to the RTX 3090 24GB or workstation-class cards like Quadro and Radeon Pro. Still, the extra VRAM can be useful for those that tend to use multiple noise reduction nodes, a lot of OpenFX, and complicated grading node structures on 4K/6K timelines.

Note that the current supply is so poor that you will be lucky to find many of these cards for anywhere near the MSRP. However, we typically use the MSRP as a baseline for price in order to rule out fluctuations due to different brands, sales, and scarcity. The actual cost you will likely need to pay for either an AMD or NVIDIA card is likely to be quite a bit different, so keep that in mind as you read this article.

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 December 1st, 2020

To test each GPU, we will be using the fastest platform currently available for DaVinci Resolve – most notably the AMD Threadripper 3970X. Since Resolve can utilize the CPU heavily in some workloads, this should minimize the impact of the processor and allow each GPU to perform at their fullest potential.

For the testing itself, we will be using the 0.92.1 version of our PugetBench for DaVinci Resolve benchmark. 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.

We will be using the "Extended" preset that includes both 4K, 8K media as well as specific GPU Effects and Fusion tests. Using 8K media with most of these GPUs (including the 6900 XT) is actually not a good idea due to the "out of GPU memory" errors you would likely encounter, but our benchmark does not load the Resolve UI which means that the VRAM load is much lower; allowing GPUs with just 8GB of VRAM to successfully complete the 8K tests.

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 task you tend to perform in your workflow, examining the raw results is going to be much more applicable than our more general analysis.

Overall DaVinci Resolve Performance Analysis

While many reviewers like to solely look at things like temporal noise reduction (often to an unrealistic degree) or OpenFX that heavily utilize the GPU, we first want to start off by looking at the overall performance we saw from our DaVinci Resolve benchmark with each GPU in order to show what most users would likely experience in their day-to-day work.

Looking at the Overall Extended Score, the AMD Radeon 6900 XT is interestingly only about on par with the Radeon 6800 XT. They trade places depending on the exact test but overall ended up scoring almost exactly the same.

Unfortunately, that means that AMD was unable to make up any ground versus the NVIDIA GeForce 3000 series GPUs. Even the RTX 3060 Ti ended up scoring a bit over 10% higher, and the RTX 3080 was closer to 20% faster than the Radeon 6900 XT.

The AMD Radeon 6000 series cards do have the advantage of having more VRAM than the similarly priced NVIDIA models, although because of how VRAM usage breaks down across 4K, 6K, and 8K timelines, that actually isn't as big of a deal for most users as you might expect.

Bear in mind that this looking at the Overall Extended Score which measures the performance in all of our tests – including a number that can be CPU bound. To get a better idea of the maximum performance difference between these cards, we should hone in on the "GPU Effects" portion of our benchmark which looks at tasks like TNR and various GPU-accelerated OpenFX.

GPU Score Analysis

The GPU effects portion of our benchmarks looks at the performance of individual GPU-accelerated effects such as temporal noise reduction, film grain, lens blur, optical flow, face refinement, and more. In our testing, these effects easily show the largest benefit from having a powerful GPU, which means that they should give us the best look at the maximum performance gain you may encounter from each of the GPUs we are testing.

In this test, the Radeon 6900 XT does pull away from the 6800 XT by about 6%, but still only manages to tie with the RTX 3060 Ti. The RTX 3080 (which is cheaper than the 6900 XT by about $300), on the other hand, out-performs the 6900 XT by a large 56%. And if you have a bit of budget to spare, the more expensive RTX 3090 is nearly 80% faster than the 6900 XT.

How well does the AMD Radeon RX 6900 XT 16GB perform in DaVinci Resolve Studio?

Overall, AMD is heading in the right direction with the Radeon 6000 series cards, but just like we saw in many Adobe applications, even the impressive gains they are showing over the 5700 XT simply isn't enough for them to catch up to NVIDIA.

From an overall standpoint, the Radeon 6900 XT is solidly beaten by the significantly less expensive GeForce RTX 3060 Ti by about 10%, and falls behind the more closely priced (although still cheaper) RTX 3080 by a larger 23%. And when looking at especially GPU bound tasks like noise reduction and OpenFX, the results are even more in NVIDIA's favor with the GeForce RTX 3080 scoring a massive 56% higher than the Radeon 6900 XT.

Currently, the only time the Radeon 6000 series cards – including the 6900 XT – would make sense to use for DaVinci Resolve is if you wanted to be able to edit 6K timelines but cannot afford the (significantly) more powerful NVIDIA GeForce RTX 3090 24GB. Having 16GB of VRAM on a sub-$1000 card is very nice, although 6K editing has fallen out of style a bit in favor of 8K – and 16GB is not quite enough for most 8K timelines in Resolve. For that, you ideally want to have at least 20GB of VRAM which will limit you to the GeForce RTX 3090 24GB or a workstation-class card like Quadro or Radeon Pro.

As always, keep in mind that these results are strictly for DaVinci Resolve. 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 latest AMD and NVIDIA GPUs, as well as with different CPUs and other hardware.