Puget Systems print logo
Read this article at https://www.pugetsystems.com/guides/1803
Kelly Shipman (Puget Labs Technician)

Progress Comes in Many Forms

Written on June 12, 2020 by Kelly Shipman

It's been a few weeks since my last post, and a lot of things have happened. The ongoing pandemic has continued to grind on me, and I continue to tweak my workstation and workflow to continue to get better and better at working from home, my allergies turned into a sinus infection, and I continued to struggle to learn programming. All of which continue to improve, even if just small amounts each day.

Then the world erupted. Two of my coworkers, Cameran and Arianna have already covered it better than I could ever hope to. I ask that you go read their thoughts if you haven’t already. I feel very fortunate to work at a company that takes these matters very seriously. We are a small company, and we feel every employee is essential to our success. That includes their emotional well being. Just like a family, if one person is hurting, we all stop and listen.

Back to the benchmarks. I’m sure you all are curious how the 3ds Max Benchmark is coming along. Well, good and not so good. Due to how Autodesk’s licensing works, for the time being, I’ve had to put a halt to the part of the benchmark that we would run on a customer’s computer before shipping it out. The sign-in process is complicated, and the uninstall process is very difficult. We have begun talks with Autodesk to see what options we have.

The good news is that I’ve been able to successfully run the benchmark on all of our test platforms in Labs. I’m currently in the process of sorting through the data now. Some things aren’t much of a surprise, like Threadripper is AMAZING at CPU rendering, and the 2080 Ti gives better viewport performance than the 2060.

Other things surprised me, such as loading and saving. Using a scene file that is 4GB, they all loaded in about 30 seconds, give or take a second. The exception to this is the 9900K loaded in an average of 24.6 seconds, and the 10900K averaged 26.7.

This is the average load time of multiple passes. I expected a bit more spread than we are seeing.

All of the test beds are using the same NVMe drives, so I expected the 9900K and 10900K to be faster, but then the rest of the CPUs plateaued. So either my test was flawed, or it doesn’t make much of a difference what CPU you use. A bit of research is in order.

I’m hoping to have the full article out soon detailing all the tests, so keep your eyes out for that.

Tags: 3ds Max, Autodesk, benchmark, GPU, CPU, performance

This is an incredibly bizarre comment to type and I apologize in advance. I initially tried reaching out to technical support after I noticed I was simultaneously receiving abnormally high scores *and* a number of expressions errors on the After Effects beta and wanted to validate my scores since they were literally topping the charts when I KNOW your TRX40 line is superior but first received this reply:

Thank you for reaching out to Puget Systems Support but unfortunately we are unable to provide technical support for systems not built by Puget. If you have a Puget computer please let us know the Order number. The benchmark also does not come with Technical Support so our best recommendation is to reach out in the Disqus comment section on the benchmark page.

And now my results are being flagged as "possibly using pre-launch hardware and has been hidden from the public listing page pending further review." (https://www.pugetsystems.co... despite having tried to contact you about the software issue ahead of time. So second attempt at contact, unfortunately in a public comment section which bums me out a little. Could you elaborate on the meaning of "Pre-launch hardware", and if the expressions errors would effect my scores / if you've been receiving them as well? Original email with expression screencaps and HWmonitor recordings for validation included below.

"Hi! I started benchmarking on my updated workstation today and noticed I was getting what seemed to be abnormally high results compared to most others (1106 on After Effects and 946 on Premiere) and while hubris wanted me to chalk it up to a perfect build and between testing, resting and writing you did manage to knock me down a few pegs with your TRX40 line - again I dream about tinkering with one of them day and night <3 - but because my results did seems a bit abnormal I wanted to screencap two expression errors in the after effects beta and double check if y\'all were aware of them and getting them on your own tests. (Only the right screen matters, left was just for hw monitoring)

Error 1: https://drive.google.com/fi...
Error 2: https://drive.google.com/fi...

I enjoy benching but dislike sacrificing actual practicality for the sake of a number so if the score is mud, that\'s fine, please do just let me know!
And if you have the time, please let me know some of the other standards you personally operate by when running 24/7 workstation benchmarks! (Nvidia settings, disk cache / memory mediacache settings etc etc.)
All the best,
Posted on 2020-06-14 06:01:27

Hey Vincent, like our support department mentioned, we don't provide support for our benchmarks unless you purchase a license (and that is through a different support contact than our workstation support), and posting in the comments section for the benchmark is the most likely place to get help ( https://www.pugetsystems.co... ), but I can answer this here.

The AE score you are getting looks pretty much spot-on with what I would expect and is within a percent or so of our latest testing: https://www.pugetsystems.co... and in line with the other public results with a 3950X (at least those with a decent GPU): https://www.pugetsystems.co... .

Your Pr score is also about what I would expect compared to the other public results: https://www.pugetsystems.co... . There are some really low results uploaded with a 3950X for some reason (probably due to old drivers or other system problems), but your results are in line with most of the results.

As for the one result being flagged as having pre-launch hardware, it looks like something messed up while gathering the system specs. We are still working on making that more reliable (everything is done via Javascript and as far as I know, there is no way to access the Win32 API directly through JS), but it is getting flagged because it has a spec with "0000" in the text. "0000" is a common designation for early edition engineering sample Intel CPUs, so we have a filter in place in case someone accidentally uploads results using a CPU that is still under NDA.

The expression error is a non-issue. For some reason, the last couple of versions of After Effects shows that error even though there is nothing wrong with the expression and it works fine.

Posted on 2020-06-16 22:14:25