Article Thumbnail

Premiere Pro CC 2017 NVIDIA Quadro (Pascal) Performance

Written on April 18, 2017 by Matt Bach
Share:
Table of Contents:
  1. Introduction
  2. Test Setup
  3. Render Previews
  4. Exporting
  5. Conclusion
  6. Premiere Pro Workstations

Introduction

While the Quadro P5000 and P6000 have been available for quite some time, NVIDIA has only recently launched the rest of their Pascal-based Quadro cards. Now that the full line is available, we thought it was time to take a look at how they perform in Premiere Pro.

To see how the different Pascal Quadro cards compare to each other, we will be examining the following tasks:

  1. Rendering previews (standard and VR projects)
  2. Exporting to 4K H.264 (standard and VR projects)
  3. Exporting to 4K DNxHR HQ
  4. Exporting to 8K H.265

If you would like to skip over our test setup and individual benchmark results, feel free to jump ahead to the conclusion.

Test Setup

To see how the new Quadro cards perform in Premiere Pro, we will be testing with the following hardware:

Testing Hardware
Motherboard: Asus X99 Deluxe II
CPU: Intel Core i7 6950X 3.0GHz
(3.4-4GHz) 10 Core
RAM: 4x DDR4-2400 32GB ECC Reg.
(128GB Total)
GPU: NVIDIA Quadro
P4000 8GB

MSRP ~$920
NVIDIA Quadro
P5000 16GB

MSRP ~$2,500
NVIDIA Quadro
P6000 24GB

MSRP ~$5,500
NVIDIA Quadro
GP100 16GB

MSRP ~$7,000
NVIDIA
Titan Xp 12GB

MSRP ~$1,200
Hard Drive: Samsung 960 Pro 1TB M.2 PCI-E x4 NVMe SSD
OS: Windows 10 Pro 64-bit
Software: Premiere Pro 2017.0.2

Our base test platform uses some of the best hardware currently available for Premiere Pro including the Core i7 6950X CPU and 128GB of RAM. For storage, we will be using just a single Samsung 960 Pro 1TB NVMe drive. Be aware that based on our Premiere Pro Storage Optimization testing, we typically recommend having at minimum one separate SSD for your media cache and scratch files as that greatly improves the time it takes to import media, conform audio, and generate peak files. However, since we are not testing those tasks in this article (because they are not GPU accelerated) we opted to stick with a single drive to cut down on the number of hardware variables that might affect our results.

While we will not be testing any particularly low-end video cards, we will be looking at the P4000, P5000, P6000, and GP100. Note that the GP100 is a bit of a specialty card that includes some features like NVLink and HBM2 memory which shouldn't make much of a difference in Premiere Pro. This card is designed more for engineering and simulation, but we had one on hand so we decided to go ahead and throw it into the mix. We will also be including the Titan Xp video card which is a bit of an oddball from NVIDIA because in many ways it doesn't fit into either the Quadro or GeForce line  - they recently even dropped the "GTX" term from the name entirely. However, it is a very popular card for GPU-intensive applications so it should work well as a point of reference.

Most of the media we will be using is available from the Sample R3D Files and were transcoded to the various codecs we wanted to test.

4K

EPIC DRAGON
3840x2160
23.976 FPS

6K

WEAPON 6K
6144x3077
23.976 FPS

8K

WEAPON 8K S35
8192 x 3456
50 FPS

To test exporting and rendering previews we used a moderately complex timeline involving multiple clips, Lumetri Color correction, multicam footage, and some other effects like a logo overlay, Gaussian blur, and cross dissolves. If you want a more in-depth look at what our timelines are like, we recorded a short video explaining our test process:

Our 4K VR testing was performed using the "Sample 1 - Ring road motorbike ride" footage from the Autopano Video Benchmarking page. For this testing, we limited ourselves to only using built-in Premiere Pro effects such as Lumetri Color correction, text overlay, and cross dissolves.

Render Previews

Rendering previews is something that you hope never to have to do since it can interrupt your workflow, but if you do complex editing it is often unavoidable. Because of this, being able to render previews as quickly as possible can be an important part of an efficient Premiere Pro workstation.

NVIDIA Quadro Pascal Premiere Pro 2017 Benchmark Render Previews

 

Standard Footage

VR Footage

 

Overall, the results were fairly consistent between our standard and VR footage testing, but there was some variation between the different codecs and resolutions we tested. This means that if you happen to primarily use one of the codecs we tested, we recommend looking at the individual result rather than the average.

With that said, the Quadro cards all performed about how we expected with the top end card (GP100) being about 15% faster than the lowest end card (P4000). This may not seem like a very big difference considering the huge price disparity, but if you look at our testing for GeForce cards the relative performance between the models is strikingly similar. This means that the relatively small differences are not really the fault of the Quadro cards themselves, but rather simply a sign that performance does not increase at the same rate as the price.

Exporting

Since there are a huge variety of resolutions and codecs you might export to, it simply isn't feasible for us to test every possible combination. However, what we can do is to make sure we touch on the more common combinations as well as the ones we believe will be more widely used in the future. Because of this, we tested exporting to 1080p, 4K, and 8K using H.264, H.265, and DNxHR HQ codecs. Our source footage also uses a range of codecs including H.264, RAW TIFF, ProRes 422HQ, ProRes 4444, DNxHR HQ, and RED.

In addition, since VR has been gaining steam we also wanted to include exporting VR projects to 4K H.264 from a variety of source codecs.

NVIDIA Quadro Pascal Premiere Pro 2017 Benchmark Export

Export to 1080p H.264

Export to 4K H.264 VR

Export to 8K H.265

 

Export to 4K H.264

Export to 4K DNxHR HQ 8-bit

 

Exporting is a bit interesting because we actually saw the largest performance gains from the more expensive cards when exporting to 1080p. Instead of the 14% average difference we otherwise saw between the lowest and highest cards, for 1080p the difference was almost 30%. 

If we look at the results as an overall average, we saw about a 4% increase in performance going from the P4000 to the P5000, then a 10% improvement moving up to either the P6000 or GP100.

Conclusion

Our Premiere Pro testing covers nearly 45 benchmarks with various footage resolutions and codecs and while there is certainly a performance difference between the different cards we tested, the difference is smaller than many might expect. There are certainly times when the higher end cards are much faster, but on average we can sum up our testing with the following statement:

In Premiere Pro 2017, we saw on average a 6% difference between the P4000 and P5000, a 10% difference between the P5000 and P6000, and a 2% difference between the P6000 and GP100.

As we mentioned in the test setup section, the GP100 is a bit of a specialty card designed for more for engineering applications than video editing - but we wanted to see how much of a difference the high bandwidth memory on this card makes. Unfortunately, the answer is: not much. For most users, the small performance difference isn't going to be worth the price premium of this card.

Otherwise, the performance gain from the different cards is about what you would expect. The P4000 stands up surprisingly well to the P5000, although the P5000 also has 16GB of VRAM which would make it the better choice for 8K or higher timelines. The P6000 has even more VRAM (24GB) although this is more than Premiere Pro should ever really need so the main advantage to using this card is simply the fact that it is 10% faster than the P5000.

If you looked closely at our charts, you probably noticed that the NVIDIA Titan Xp was as fast or faster than even the most expensive Quadro card even though it is nearly 5x cheaper than the P6000. We do want to point out that raw performance is not the only consideration here; there are a number of other advantages to using a Quadro card that you might want to consider:

  1. Quadro supports 10-bit displays (GeForce does not)
  2. Quadro cards are a bit more reliable long-term
  3. NVIDIA & Adobe tend to provide better software support for Quadro cards.

Whether those advantages are worth the additional cost it is completely up to you. We offer both product lines on our Premiere Pro workstations so whether you value the best performance for your dollar or maximum stability and reliability, we have you covered.

Tags: Premiere Pro, GPU, Video Card, Quadro, Pascal
guest01

"Quadro cards are a bit more reliable long-term"
I assume this claim rests on some empirical data. Would you mind sharing the exact numbers?

Posted on 2017-04-20 21:06:38

I can't give you the raw numbers, but I'm happy to share some percentages. Over the last year, we saw a total failure rate (including both DOA and in the field) of 2.02% with GeForce and 1.36% with Quadro. We feel that in the field failure rate is most important (since that is the number of our customers that have to deal with an issue) and for that it is .99% with GeForce cards and .45% with Quadro cards. So Quadro is not quite twice as reliable overall, but it is more than twice as reliable from the viewpoint of our customers.

Do keep in mind that this covers everything from video corruption to bad display ports to a noisier than normal fan. So it's not just catastrophic failures but everything that doesn't meet our quality standards.

Posted on 2017-04-20 21:44:34
Pal Pacher

Thanks for this interesting comparison. I would be really curious to know how the new Pascal Quadro cards compare with the older architecture ones in PPCC2017 in the performance (e.g K5200 vs P5000)? Would it worth to upgrade? thanks

Posted on 2017-04-22 23:57:40