Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1324
Article Thumbnail

OctaneRender 4.00: NVIDIA GeForce & Titan RTX GPU Rendering Performance Roundup

Written on January 14, 2019 by William George
Share:

Introduction

OctaneRender, from OTOY, is a real-time, physically correct 3D rendering engine that uses GPUs instead of CPUs for processing. This is a relatively new approach, as traditional graphics rendering was done on CPUs instead. Graphics processors are ideal for highly parallel tasks like rendering, though, and it is easier to fit multiple video cards in a single computer than multiple CPUs.

OctaneBench, the free benchmark for OctaneRender, was recently updated to version 4.00. While this does not yet have support for the new features in NVIDIA's RTX series of graphics cards, it does bring a performance boost compared to previous versions. As such, we thought it was time to do a round-up of GeForce and Titan cards from the last two generations.

Test Setup

For our test platform this time around, we wanted to use what we recommend to folks looking for a multi-GPU rendering workstation. That means a board and processor capable of running four video cards, with high clock speed being a bigger factor that core count - and also minimizing the cost of the CPU, so that funds can be focused on getting the most powerful GPUs instead. That combination leads naturally to a Xeon W based system, with the quad-core Xeon W-2125, ECC memory, and a fast SSD to avoid potential bottlenecks.

If you would like full details on the hardware configuration we used for this round of testing, just .

Benchmark Results

Here are the OctaneBench 4.00 scores for each of the GPUs we tested, sorted from fastest to slowest and with the RTX series cards highlighted in darker green:

OctaneBench 4.00 GeForce GTX, RTX, and Titan GPU Performance Roundup

And here is another way of looking at the results, as percentages relative to the GeForce GTX 1080 Ti's performance:

OctaneBench 4.00 GeForce GTX, RTX, and Titan GPU Performance Roundup as Percentage Compared to GTX 1080 Ti Result

Analysis

The new Titan RTX didn't unseat the Titan V for OctaneRender's top performance spot, but it is a little faster than the RTX 2080 Ti and has twice the video memory of either of those cards. That massive VRAM capacity seems tempting, especially for folks rendering large, complex scenes - but there is a caveat. While the Titan RTX offers solid GPU-based rendering performance as a single card, it is limited to one or maybe two cards at most in a single workstation.

Why is that? NVIDIA started using dual-fan coolers on their RTX series (both GeForce and this Titan model). Using dual fans means a single card stays cooler and quieter than a more traditional, single-fan design - but it also means greatly increased heat within the computer, which is compounded with each additional card. Most systems can handle the heat from one GPU, and some can handle two with high airflow fans and ample space between the two cards, but beyond that there is just too much heat inside and the fans on the video cards start recycling that hot air which can no longer properly cool the cards - leading to overheating and throttling. Unfortunately, it looks like the Titan RTX is only built by NVIDIA directly... meaning there are no 3rd party models with different cooling layouts.

What Are the Best GPUs for OctaneRender?

NVIDIA's Titan V is still the fastest card that can be effectively used in multi-GPU workstations for OctaneRender 4. The newer Titan RTX does have more video memory, and costs less, but it is crippled by its dual-fan cooling layout and unsuitable for multi-GPU.

The RTX 2080 Ti is not far behind, though, and costs far less than either Titan - so those are probably the best option right now, as long as you use single-fan models. If you need more video memory, to rival the 24GB that comes on the Titan RTX, NVIDIA's Quadro RTX 6000 should have similar performance and is equipped with the right type of cooler... but also substantially more expensive.

Recommended Systems for OctaneRender

Tags: GPU, Rendering, Octane, Render, OTOY, OctaneBench, Benchmark, NVIDIA, GeForce, GTX, 1070 Ti, 1080 Ti, Pascal, Titan, Volta, RTX, 2070, 2080, 2080 Ti, Turing, Performance, Video, Card

What if you water cool multiple
Titan RTX's?

Posted on 2019-01-17 17:30:47

That is a potential option for those with the expertise and patience for putting together such a build. We don't offer liquid-cooling here at Puget anymore, because of the greatly increased risk of shipping damage and the added complexity for things like upgrades later on - but folks who build their own systems could opt to go that way. Given the heat these cards put off, though, you would likely need multiple radiators... at least two, looped between pairs of cards (or on completely separate loops) for that to work.

Posted on 2019-01-17 17:42:30

Thank you William. Did the person testing see any benifits in the VRAM while working in the viewport with the RTX Titan? And if so how much compared to the 2080 Ti?

Posted on 2019-02-22 00:52:02

This test was just using OctaneBench to look purely at rendering speed - it did not test modeling or viewport speed in any applications. That is something we'd like to build testing for, with any or all of Cinema 4D, Maya, and 3ds Max, but it is harder to measure and we don't have experienced designers on staff at this time.

Posted on 2019-02-22 16:46:04
Darryl

Great article. How do you think the Titan RTX will compare to the upcoming Quadro 5000?

I'm using Maya, Octane and Daz3D. Will bring props and models into Maya for animation but will also do stills in Daz Studio as well.

I know the Quadro 5000 has less Vram but I've heard the Quadros are optimized for viewport performance. Just wondering if the RTX Titan has those driver optimizations as well. In general does the TITAN get most of the driver updates to run nicely on pro apps such as Maya.

Note I will not do any CAD work so that's not an issues. Thx

Darryl

Posted on 2019-01-18 13:34:09

The Titan series has been all over the board - sometimes they get more professional drivers and certification with applications like Maya, sometimes they don't. The same goes for other "pro" features, like good FP64 performance: some Titans have them, some don't. Because NVIDIA has tried to straddle the line between mainstream / gamer and professional video cards with this series it has really led to a mixed bag in that regard. Unfortunately, I don't currently have a good benchmark in place for Maya to try and answer that question directly either... but that is something we are hoping to work on this year.

I can say, from just a specs standpoint, that the Titan RTX is more akin to the Quadro RTX 6000. The 5000 is more on par with a GeForce RTX 2080, I think, though of course with more VRAM onboard. Certified drivers can make a big difference with applications like Maya, though, as you alluded to. I doubt any of the RTX cards (even the Quadros) have been certified by Autodesk yet... so if that is a crucial factor to you, you might want to wait and see what cards they do end up testing.

Posted on 2019-01-18 17:50:24
Machausch

Great work! For the Titan V did you unlock its full power for compute workload? We have seen 400+ scores from Otoy‘s RenderToken community: https://www.reddit.com/r/Re...

Posted on 2019-01-22 11:06:07

For the purposes of this article, the Titan V (and all the other cards) were run at their stock / factory settings. I did experiment with enabling full power (P0) state on the Titan V for compute, but found that it quickly ran into thermal throttling and ended up with an OctaneBench score around 380. The fan wasn't at full speed, though, so there would likely be further tweaks one could make to force the fan higher and get a score around 400 like you mentioned... but since that is not something we would be comfortable doing on systems we sell, in this article I stuck with the scores which a customer would see if they purchased one of these from us :)

Posted on 2019-01-22 17:03:44
Morgan Brown

Are you able to confirm if NVLink works for memory pooling in Octane?

Posted on 2019-02-26 17:59:14

This is actually something I would like to check, but I am not sure how to go about it using OctaneBench. The test scenes are not large / complex enough for memory pooling to show a performance benefit, if it was functioning, and the GPU list shown on the left-hand side of the OctaneBench window shows all installed GPUs with their normal amount of VRAM - whether NVLink is enabled or not.

However, in the current preview version of OctaneBench 2019, I can say that having NVLink enabled causes the test to crash. I messaged OTOY about that and it sounds like they haven't enabled NVLink functionality in the RTX-enabled version of OB 2019... but that makes sense, as it is still in development and the benchmark they released is just a preview. There is no such error in OctaneBench 4.00b.

Do you have any suggestions about how to check for memory pooling? Is there a command-line instruction for Octane that might list details like this?

Posted on 2019-02-26 19:29:13