AMD Threadripper 3990X: Does Windows 10 for Workstations improve rendering performance?

Introduction

When AMD launched the 64-core Threadripper 3990X, some reviewers reported that performance of this 128-thread beast was hindered by running a normal version of Windows 10 Pro – and that using Windows 10 Pro for Workstations or Windows 10 Enterprise instead gave better results. My initial round of testing with this processor was done on Windows 10 Pro, but we also offer the Windows 10 Pro for Workstations edition as an option on PCs we build here at Puget Systems… although we had never used that in our Labs testing.

Because the 3990X performed so well with CPU-based rendering applications, and became my top recommendation for that type of workload, I felt it was important to investigate the claims about how the choice of operating system might be impacting multithreading. With rendering specifically in mind, I re-ran both current and previous-generation versions of Cinebench and V-Ray benchmarks (the same tests I had used on the 3990X previously) to see if the version of Windows being used would impact our customers.

In addition to looking at Windows 10 Pro for Workstations as a potential way to improve performance, I also wanted to check on the impact of SMT (simultaneous multithreading… AMD's equivalent to Hyper-Threading) top see what turning it off might do to rendering. Some applications are known to benefit from turning SMT / HT off, but in the case of CPU-based rendering I expect that doing so will only slow things down… but it is worth checking, especially since it was specifically raised as another issue in the article linked to above.

As a side note, I also took this opportunity to bump up the CPU cooler from Noctua's U12S to the U14S. Our product qualification team had found that the U12S was borderline for cooling the 3990X, and in some situations could lead to very slight thermal throttling. For this reason, some of the normal Windows 10 Pro results presented here may be slightly higher than they were in our older articles.

Test Hardware

Here is a list of the hardware components that made up my test configuration, along with the OS versions and benchmarks I used:

Benchmark Details

As previously mentioned, I used two versions of Maxon's Cinebench utility for this comparison: both the current R20 release and the previous R15. They both render a scene using Cinema 4D's built-in engine, fully utilizing all cores until the render is completed, but the newer version includes a more complex scene that takes longer and thus gives a better comparison for high-performance processors. Based on how long the render takes, an artificial "CB score" is assigned – with higher scores being better, and representing shorter render times. There is also a single-core mode available, performing the same task with just one active thread – which can give a decent idea of single-threaded performance. It isn't applicable to rendering but can give a rough idea of how CPUs should stack up in modeling.

I also utilized two benchmarks from Chaos Group, makers of the V-Ray rendering engine. Their V-Ray Next Benchmark includes tests for both CPU-only and GPU+CPU rendering. This is somewhat novel, as most GPU rendering engines do not use the CPU at all, but in V-Ray Next they implemented CUDA emulation on the CPU to improve performance even further. V-Ray Benchmark 1.0.8 is older, but some folks may still be utilizing V-Ray plug-ins from a year or two ago – and the test is short enough that we decided to go ahead and run it this time around to have an extra data point. It includes both CPU and GPU tests, but we are only listing the CPU score here since GPU performance is not our focus nor is it impacted by the CPU in this version.

Benchmark Results

Each of the four benchmarks was run on Windows 10 Pro and Windows 10 Pro for Workstations – both with SMT enabled (the default setting) and disabled. That results in either 128 or 64 threads being available to the operating system, and the claim which needed verifying was that the normal version of Windows 10 Pro would not handle >64 threads as well as the Pro for Workstations variant.

To help keep things easy to track, I have separated the two versions of Windows 10 by color: Windows 10 Pro is shown in red, while Windows 10 Pro for Workstations is in blue. I also shaded the bars showing SMT enabled darker and SMT disabled lighter. The order of the four different combinations is kept the same throughout the charts, to avoid any confusion that could arise from sorting the results.

Here is a gallery of charts showing Cinebench performance, where higher CB scores indicate faster performance:

And here is a similar gallery, using the same order and color scheme, with V-Ray benchmark results:

Analysis

Looking first at the performance difference between operating systems with SMT enabled – Windows 10 Pro with darker red line and Windows 10 Pro for Workstations in dark blue – it looks like there is no substantial benefit either way. If anything, the normal (non-Workstation) version is *very slightly* ahead in many of the tests, but by less than half a percent… so well within the margin of error.

Regardless of the OS version, though, turning off SMT (the lighter colored bars) resulted in a substantial drop in performance. The 10 to 15% slower rendering times come with no upside, as single-threaded performance (shown in the last of the Cinebench R15 charts) stayed the same across all four configurations. For CPU-based rendering, then, you definitely want to leave SMT enabled!

Does the version of Windows 10 Pro impact rendering speed with the 3990X?

My findings in this round of testing contradicts what Anandtech posted in their launch-day article: as far as I can tell, there is no difference between using Windows 10 Pro or Windows 10 Pro for Workstations with AMD's TR 3990X for CPU-based rendering.

Closing Thoughts

Between this data, and our other articles looking at the same situation in Adobe CC programs and photogrammetry, it seems that something else must have been throwing off the performance measurements that Anandtech took. Other review websites, and indeed AMD themselves, have also come out with similar conclusions. This is a great example of why testing by multiple sources is important!