Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1023
Article Thumbnail

V-Ray GPU Rendering Platform Comparison: Skylake X, Xeon W, and Threadripper

Written on September 6, 2017 by William George
Share:

Introduction

V-Ray, from Chaos Group, is a hybrid rendering engine that uses both CPUs (processors) and GPUs (video cards). This allows it to take full advantage of all available hardware in a workstation, in order to deliver amazing rendering performance. We have already tested current CPUs on V-Ray, but since GPUs are utilized as well it is important to look at how different platforms handle scaling across multiple video cards.

However, care does have to be taken when putting more than one video card into a workstation: there needs to be space in the chassis, the right type of PCI-Express slots in the correct layout on the motherboard, a large enough power supply, and plenty of cooling. In this article we are going to look at several platforms - motherboard chipsets along with a matching CPU - to see how well they handle the scaling of performance in V-Ray Benchmark.

Test Setup

To see how the different platforms scale in V-Ray 3.57.01, we used the following configurations:

These test configurations cover six major CPU platforms that are currently available, four from Intel and two from AMD. With the tested motherboards they support either 2, 3, or 4 GPUs at maximum - with a mix of x8 and x16 PCI-Express lanes per card, depending on the board and card layout. In our past testing with GPU rendering we found that running cards at x8 vs x16 made no substantial difference.

For the video cards themselves we used NVIDIA's GeForce GTX 1080 Ti, a well-respected GPU for this application due to its high speed and fairly large RAM capacity with a reasonable price tag. Multiple cards were run on each platform, up to 2-4 depending on the board.

To determine performance, we ran V-Ray Benchmark 3.57.01 and recorded the results from each hardware combination. That test returns the number of seconds it takes to render a scene, so lower scores are better. Our findings are shown in the chart below, separated by platform and the number of GPUs - which are color coded for easy cross-platform comparisons.

Benchmark Results

Here are the results, in seconds, from V-Ray Benchmark for the platforms and number of GPUs we tested:

V-Ray 3.57.01 Multi-GPU Platform Performance Comparison

The main takeaway here is that the platform does not make any substantial impact on performance. There is a small amount of variance, within a second or two, but that is well within margin of error. Any of these platforms will do very well with multi-GPU rendering in V-Ray, though keep in mind that the chipset and motherboard will determine the total number of video cards that you can use in a system.

Conclusion

Based on our test results, the platform itself isn't a huge concern for V-Ray when it comes to GPU performance. All the chipsets we tested perform within a few percentage points of each other, though the number of video cards supported does vary by motherboard.

Since V-Ray uses both GPUs and CPUs, though, this needs to be balanced with performance results from the processors themselves. We've done that testing in a separate article, and it is well worth looking at if you are considering the purchase of a new workstation. We have put together recommended systems that take into account both CPU and GPU performance, and offer a range of options so that you can tailor them to your specific needs and budget.

Recommended Systems for V-Ray

1 CPU / 1-2 GPU
Compact

Configure


1 CPU / 1-4 GPU
Tower

Configure


2 CPU / 1-4 GPU
Tower

Configure


1 CPU / 1-4 GPU
1U Rackmount

Configure


Tags: GPU, Rendering, V-Ray, RT, Benchmark, Chipset, Motherboard, Performance, Intel, AMD, Core, i7, i9, Skylake X, Xeon W, Ryzen, Threadripper, X99, X299, C422, X399, Z270, X370, Video, Card
George

Thanks, very informative!
It also seems that Vray doesn't take well advantage of the 3rd and 4th GPU compared to Octane.
I mean with two 1080 it's twice faster, but with the 3rd and 4th it's not as fast as we would expect.

Posted on 2017-09-07 15:27:57

Yeah, scaling to a 3rd and 4th card isn't as effective as one might hope in V-Ray 3.5. There is a newer version of V-Ray out now, 3.6, which does not yet have a stand-alone benchmarking tool... but I am hoping to do some testing on it soon anyway. I want to see if they've improved the GPU scaling at all, and I also want to look at a new feature called Hybrid Rendering. It utilizes both the CPU and GPU(s) at the same time, which sound promising.

Posted on 2017-09-07 15:47:42
George

Thanks for the reply, waiting to see then.
Maybe with longer render it would be better to see if the render time scale well with GPU (here the render only took 67 seconds).

Posted on 2017-09-08 09:35:55
Petar J Petrovic

Thank you for this William. One question though. Apparently Ryzen doesn't create an impact on speed while using 2 cards (due to smaller number of PCI-e lanes, compared to cpu-s with higher count). With that in mind, does Ryzen support 3 gpus and would there be a speed impact in that configuration? Looking at these tests, I can't justify buying a Threadripper platform for lager PCI-e count, as it obviously doesn't have any impact on rendering speeds.

Posted on 2017-12-06 00:16:05

A lot of motherboards for Ryzen don't even have a third slot where another GPU could fit, and those that do (like the Gigabyte GA-AX370-Gaming K7) seem to run that third slot at only PCI-E x4 speed. The testing I have done recently didn't include any GPUs running in that slow of a slot - I only had systems using x16 (normal) and x8 speeds. I suspect that there might be a small performance impact from using an x4 slot, but for rendering it probably won't be a huge factor.

If you do go that direction, make sure you get a chassis and power supply large enough to handle the third GPU. Some motherboards put that at the very bottom of the board, meaning you'd need a chassis with an extra expansion slot as well (assuming dual-width GPUs). That also leads to a potential reason to go Threadripper as well: it would let you fit four GPUs (provided the correct motherboard), so even if the CPU itself isn't having a huge impact you could gain more GPU performance that way. The CPU may also come into play in other ways, depending on what exactly you are running. For example, V-Ray RT 3.6 now uses both the CPU and GPUs together for rendering (see our more recent articles on that topic for more info).

Posted on 2017-12-06 15:42:12
Petar J Petrovic

And there (x8/x8x/x4) lies the devil for me. Knowing whether there is a considerable loss of performance in that last slot (or not) possibly saves me a good chunk of funds (for a better GPU) I would otherwise spend on more expensive cpu and mb. Just the ability to add that third gpu down the road would make a future-proof rig out of Ryzen (considering they will be updating the cpu itself). I more or less understand the chassis and psu issues involved. This (and other articles you wrote) already helped me a great deal, and thank you for writing them in a clear and concise manner. If you happen to know any other info on the subject I'd be very happy to read about it as this current conundrum really stretches me beyond my limits of creating a good workstation I wouldn't need to touch in the next 3-5 years.

Posted on 2017-12-06 20:39:35