Puget Systems print logo
https://www.pugetsystems.com
Read this article at https://www.pugetsystems.com/guides/2051
Article Thumbnail

DaVinci Resolve Studio - NVIDIA RTX A6000 48GB Performance

Written on January 28, 2021 by Matt Bach
Share:

TL;DR: NVIDIA RTX A6000 48GB performance in DaVinci Resolve Studio

While the higher cost of workstation-class hardware like the A6000 may put it outside the budget of many, for those that need 48GB of VRAM or where system uptime is worth the investment, the NVIDIA RTX A6000 48GB is a terrific choice. When the GPU is being heavily used in DaVinci Resolve, the RTX A6000 is about 32% faster than the previous generation RTX 6000/8000, and more than 2.25x faster than the Quadro P6000 from two generations ago! It is also around 75% faster than the fastest AMD Radeon Pro GPU, making it easily the fastest workstation card currently available for DaVinci Resolve.

Introduction

While the launch of NVIDIA and AMD's consumer GPUs have been a major topic in recent months, NVIDIA is also starting to release the successor to their Quadro RTX line - starting with the RTX A6000. The NVIDIA RTX A6000 is a very powerful card, featuring 48GB of VRAM and significantly higher specs than the Quadro RTX 6000/8000 from the previous generation.

One thing we always like to point out when testing workstation cards like the RTX A6000 is that in many cases, straight performance is not the primary reason why you would invest in one of these cards over an NVIDIA GeForce or AMD Radeon GPU. In fact, performance tends to be lower than a comparable "consumer" GPU since most of those cards are designed to push the boundaries of performance.

Workstation cards, on the other hand, are all about specialty features not necessary for most consumer workflows (higher VRAM capacity, ECC VRAM, vGPU support, Quadro Sync, etc.), and more importantly for many, have a strong focus on stability. Not only do workstation drivers get more validation for professional applications, but the hardware itself is also more reliable. In fact, over the last three years (Jan 2018 to Jan 2021), we have seen about 3x higher reliability in the field with Quadro cards over GeForce (.33% failure rate versus .91%).

In addition, another advantage for Workstation cards that can be a factor for DaVinci Resolve users is that they are designed with multi-GPU configurations in mind. While there are some blower-style NVIDIA GeForce cards, having more than one or two GeForce cards installed in a system is not something that NVIDIA specifically designs them for.

DaVinci Resolve Studio GPU Performance Benchmark - NVIDIA RTX A6000 48GB

If you want to see the full specs for the NVIDIA RTX A6000 as well as the other NVIDIA and AMD workstation GPUs we will be testing, we recommend checking out the NVIDIA Quadro and AMD Radeon Pro product pages. But at a glance, here are what we consider to be the most important specs:

VRAM Cores FP32 Performance Power ~MSRP
Quadro RTX 4000 8GB 2,304 7.1 TFLOPS 160W $449
Radeon Ro WX 9100 16GB 4,096 12.3 TFLOPS 230W $1,499
Radeon Pro VII 16GB 3,840 13.1 TFLOPS 250W $1,899
Quadro RTX 5000 16GB 3,072 11.2 TFLOPS 265W $2,300
Quadro RTX 6000 24GB 4,608 16.3 TFLOPS 295W $4,000
RTX A6000 48GB 10,752 38.7 TFLOPS 300W $4,650
Quadro RTX 8000 48GB 4,608 16.3 TFLOPS 295W $5,500

While specs rarely line up with real-world performance, the biggest thing to note is that the RTX A6000 includes 48GB of VRAM at a lower cost than the Quadro RTX 8000, and theoretically has more than twice the performance of any of the previous generation cards.

48GB of VRAM is not necessary for most Resolve users, but if you work with especially high-resolution media (12K+) and make heavy use of OpenFX and noise reduction, having 48GB of VRAM can be the difference between a smooth workflow and one plagued with "Your GPU memory is full" errors. Considering that the RTX A6000 48GB is $850 less than the RTX 8000 48GB (which adds up quickly if you want multiple GPUs), the A6000 is already attractive from a straight price perspective - not to mention any performance improvements there may also be.

DaVinci Resolve 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 are the specifications of the system we will be using for our testing:

*All the latest drivers, OS updates, BIOS, and firmware applied as of Jan 20th, 2021

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 its fullest potential.

To see how the RTX A6000 performs, we will be comparing it to most of the mid to high-end NVIDIA Quadro GPUs from the previous generation as well as the Quadro P6000 from two generations ago. One thing to note is that we are not going to test the Quadro RTX 8000 since, in terms of performance, it is actually identical to the Quadro RTX 6000. The RTX 8000 does have more VRAM than the RTX 6000 (48GB vs 24GB), but since our DaVinci Resolve benchmark doesn't need more than 8GB of VRAM, that would not affect the results for our benchmark.

In addition, we are only including two Radeon Pro GPUs: the Radeon Pro WX 9100 and Radeon Pro VII which are currently the top-end workstation GPUs from AMD. NVIDIA tends to hold a commanding lead in applications like DaVinci Resolve, but including these two cards will let us see if there are any surprises since the last time we tested workstation cards in DaVinci Resolve.

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 is actually not a good idea as we generally recommend 20GB+ of VRAM when working with 8K media in Resolve, 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.

NVIDIA RTX A6000 48GB DaVinci Resolve Studio benchmark results

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 Score, the NVIDIA RTX A6000 tops the chart but only manages to sneak past the RTX 6000 from the previous generation by 8%. Compared to the Quadro P6000 from two generations ago, however. the RTX A6000 is a large 33% faster. Considering that we are looking at the Overall Score which includes a number of tests that are not directly affected by the GPU, this is a very respectable increase in performance.

Compared to the AMD Radeon Pro cards, the A6000 scores about 28% higher than the Radeon Pro VII and 46% higher than the Radeon Pro WX9100. The A6000 is of course significantly more expensive than either of these cards, but this serves to show the kind of performance advantage NVIDIA currently has over AMD.

Looking at the results in more detail, it is interesting to see that the RTX A6000 only manages to tie the RTX 6000 for the 4K Media tests - suggesting that even with the best CPU available for DaVinci Resolve, we are likely hitting a CPU limitation for many of those tests. Another oddity is that the performance in Fusion is a bit worse with the RTX A6000 compared to the RTX 5000 and RTX 6000. We are not sure why that would be, but Fusion has always been a bit odd for GPU performance. For example, many of our Fusion projects actually show a drop in performance as you add more GPUs to the system.

Bear in mind that this looking at the Overall Score which includes the performance for a number of tests 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

DaVinci Resolve Studio GPU Score benchmark performance NVIDIA RTX A6000 48GB

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.

For this test, the RTX A6000 again tops the chart, but this time by a significantly larger margin. It ended up performing about 32% faster than the RTX 6000 or just over 2.25x (!) faster than the Quadro P6000 from just two generations ago. The performance gain over the RTX 6000 (which has the same performance as the RTX 8000 48GB) is big enough by itself, but if you have an older GPU and looking for a workstation-class upgrade, the A6000 can give you a huge boost to performance.

We again want to point out that workstation cards like the RTX A6000 are not solely about performance, but for those that want to know how it performs relative to the NVIDIA GeForce line, you can compare it to the results in our recent DaVinci Resolve Studio - AMD Radeon RX 6900 XT Performance article. In terms of just straight-up performance, the GPU score of ~141 for the RTX A6000 48GB puts it right between the GeForce RTX 3080 10GB and the RTX 3090 24GB (only with more VRAM, higher reliability, and all the other workstation-only features).

How well does the NVIDIA RTX A6000 48GB perform in DaVinci Resolve Studio?

While the higher cost of workstation-class hardware like the A6000 may put it outside the budget of many, for those that need 48GB of VRAM or where system uptime is worth the investment, the NVIDIA RTX A6000 48GB is a terrific choice. When the GPU is being heavily used in DaVinci Resolve, the RTX A6000 is about 32% faster than the previous generation RTX 6000/8000, and more than 2.25x faster than the Quadro P6000 from two generations ago! It is also around 75% faster than the fastest AMD Radeon Pro GPU, making it easily the fastest workstation card currently available for DaVinci Resolve.

Another aspect to consider is the fact that the RTX A6000 has a 50W lower TDP than the GeForce RTX 3090 (300W vs 350W). This is important because as we discovered in our Quad GeForce RTX 3090 in a desktop - Does it work? post, a workstation with four RTX 3090 cards ended up using ~1700 watts of power which is more than most 120V electrical circuits is able to reliably provide over long periods of time. If the RTX A6000 is able to drop that down to 1500 watts with four cards, that could make a quad GPU setup with RTX A6000's the fastest desktop solution you can get for Resolve without needing dual power supplies or other non-standard power configurations.

Unfortunately, at the moment we do not have access to four RTX A6000 cards, but looking at the performance with four of these cards in DaVinci Resolve is something we certainly want to test as soon as we are able to!

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 new GPUs from AMD and NVIDIA, as well as with different CPUs and other hardware.

DaVinci Resolve 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: NVIDIA, NVIDIA vs AMD, AMD, RTX A6000, RTX 4000, RTX 5000, RTX 6000, WX 9100, Radeon Pro VII, DaVinci Resolve
grokker

Why not put the 3090 in the comparison? It's not like Resolve requires a "Quadro-like" card for anything.

Posted on 2021-01-28 23:49:09

Couple of reasons for that, but the biggest is that NVIDIA/AMD/Intel all don't like it much when we (or any other reviewer) combine workstation and consumer hardware in the same article. The idea behind it is sound, however, so this is one of those requests that we have zero issue following

The reason why we agree that it usually makes sense to separate workstation and consumer hardware in articles like this is that the reason why someone would want a workstation GPU is only partially driven by performance - as we talked about in this article, there is so much more to why you might invest in a workstation GPU. People tend to just look at the charts and not read the surrounding text, and unless someone already understands what the differences are between a GeForce and Quadro card, they will probably just see that a RTX A6000 is beat by an RTX 3090, and at a fraction the price. They might actually be a user who really would benefit from the larger up front investment of a Quadro card, but it can be hard to overcome the initial reaction seeing the price/performance comparison.

Another reason is that most of the time, the people who go with consumer hardware and those that invest in workstation hardware are two completely different groups of people. It isn't quite to the same level, but it kind of is like if reviewers started combining laptop CPU and desktop CPUs in the same article. It would certainly be interesting to some, but if someone is in the market for a laptop, they probably don't care about desktop CPU performance - and vice versa - so combining them just makes the charts more complex than they need to be.

Posted on 2021-01-29 04:58:33
grokker

As someone with a very long experience in the VFX industry I couldn't disagree more with you. We buy graphics hardware for its raw power and the Studio drivers on the GeForce end (and our own experience) give us enough confidence on the reliability of these components. That's even without considering that any self-respecting studio will have engineers capable of testing hardware reliability by ourselves. The ONLY reason to buy non-GeForce hardware (I refuse to call a 3090 a "non-workstation" or "non-professional" piece of hardware) is the VGPU limitation of that series, which by the way is completely artificial, like many long-standing differences between the two lines.

I do understand though that you might be under pressure from manufacturers and accept that, but I definitely wouldn't call it any other than bullying on their part.

Posted on 2021-01-29 09:48:42

I agree that probably 90% (or more) of the VFX industry is exactly as you explained. When you can get multiple RTX 3090 cards for the same cost as a single RTX 6000, you can literally buy multiple 3090s, throw them in the trash if one breaks, and still come out well ahead. Heck, that is why we tend to lead with GeForce until we get some indication from the customer that Quadro might be a good fit. Some people see the cost as well worth it for that extra bit of reliability. If a GPU fails, it still takes time to troubleshoot and replace the card, and if that lost time ends up costing thousands of dollars...

There are also some who simply have unlimited budgets and want the best possible systems. I'm always a little jealous of those customers, but if they are working on a multi-million dollar project or something, then the difference of a few thousand (or even tens of thousands) dollars is a drop in the bucket compared to everything else. And given how central a workstation is in their workflow (literally everything stops if the system goes down), I can understand why they would want to invest in making their computers as reliable as possible.

Posted on 2021-01-29 18:21:55
grokker

You're right. All of us dream of that day when someone knocks at the door with a multi-million dollar project where "money is no object". :-)

Posted on 2021-01-29 22:41:46