I just unzipped this to try it out, and I see several additional tests - four starting with "Graphics_". The ReadMe doesn't seem to talk about them, so I was wondering if you could shed any light on what the differences are?

Okay, I ran the Graphics_Expanded test and it finished! I have uploaded the resulting text file to Google Drive, in case you want to review it to make sure it seems like it is behaving:

https://drive.google.com/file/d/0B3R...ew?usp=sharing

I'm hoping for the 2018 version of the test soon, since I'd like to run through a full gamut of video cards on the latest release... and in whatever you think the best benchmark mode will be to see differences in GPU performance.

Woot to the test finishing! That's certainly a step in the right direction.

So, yeah, the GPU stuff has been a big part of what changed this year. From a testing standpoint, I think the one that will server you best is the Hammer. This is a larger model, and it does a few more tests as well. I think this test, with the larger model, and with a high screen resolution, will hopefully start to separate the higher end cards from the pack.

The Pile driver is the same model, but it runs a huge bank of tests switching from one view style to the others in every possible permutation. I built this test for my own validations, because I discovered that what style you change FROM hugely affects performance, regardless of what style you are changing TOO. I thought that odd, and spent a lot of time making sure that was real performance, not some odd artifact of the testing process. Eventually Autodesk verified that it is real behavior. I would like some feedback on even including this one in the final version, since on decent hardware a single run takes HOURS!

And lastly, the Acceleration test just compares HA on and off, with the simple test. I think most offices have recognized the value of hardware acceleration, so this is no longer included in the full tests (now there a comparison of sketchy lines on and off!) but I included this for any office that does need to demonstrate to IT that $50 cards and HA off is a miserable option.

And lastly, 2018 is on the agenda for this weekend. If nothing else i hope to get you a similar beta early next week. Stay tuned!

Thanks!
Gordon

Awesome - thanks for that info, Gordon! I will run the Pile Driver test variant tomorrow, to make sure it completes without issue on 2017.2, and then await the 2018 version of the benchmark.

Thanks Gordon,

I will try this out this weekend! Curious how a 1600x vs 7700k does. I mean we know it is all IPC but curious how bad.

All, trying to wrap up the 2018 stuff, but like 2017, ADSK shagged the hound on journal execution in demo mode on a VM with 2018, is I have to do all my testing on my Bootcamp partition. And demo mode expired. And ADSK shaggy us by actually rooting the ^@%#$ing hard drive at the end of the demo time (why is this not a class action lawsuit?!) so now I have to activate my own license on that BootCamp partition to finish. Which I plan to do this weekend so I can wrap up. We're close, and it's still June, which is a record for the last few years. :hide:
Gordon

Thank you for the update, Gordon, and for all the work you are doing on this benchmark!

Well, after a long delay completely rebuilding a BotCamp partition and reactivating Revits, I finally got the chance to do the file upgrades for 2018. I think it's a very good sign that some of the heavy tests are now capable of driving my VM based testing procedure right into the ground. Woot!

So, here is a beta of v3.1 for 2018. If all goes well with a few of you testing, I will reconcile the last few changes with 2017, then update the 2016 PowerShell code and FINALLY issue this darn thing!

Thanks for the patience everyone.
Gordon

Awesome, thanks Gordon! I've downloaded the beta you posted and will start running through the different benchmark variants. If I run into any issues I will post here, or else I will post back once I've had a successful pass through each of them.

If there is anything else you'd like for me to be on the lookout for, please let me know. I'm also a little curious about your earlier comment regarding the Revit free trial - did you say that it messed up your OS installation when it was over? And do these benchmarks even work well with the free trial? I would think the trial pop-up dialog every time you open Revit would mess them up.