MONUMETRIC Video mmt-volt

**colliert** · July 14, 2015, 06:52 PM

Originally posted by Gordon Price

"Heavy" test: This test will be identical to the current one, but the model will be built in the middle of a 3x3 or maybe a 5x5 grid, where the other squares in the grid are occupied by linked versions of the final model. This will (I hope) produce a test that puts more load on both the CPU and the GPU, while still keeping the download reasonable. My hope is this version will provide some differentiation between hardware that is getting lost now in the ever decreasing times. It also means that processing time will go up, especially with the 5 test average option, so the old "light weight, single test" option will remain.

Thank you for creating this tool, we have been using it this past week to determine the level of our machines across the office, and finding that we're in great shape, but the size of the model in this test doesn't get near the level of models being used in our projects, so I'm not getting enough variation between the levels of the machines. Ones with SSDs aren't scoring any better than the HDD boxes, and the video cards aren't separating themselves enough either.

With scores to prove our needs, it's easier to get the bosses to approve better hardware seeing numbers that show the difference.

I would love to use models that are over 150MB for this test, is there any way to adapt the current test to this level of model creation easily?

Or am I just looking for something that's not necessary to do?

**mmetz** · July 21, 2015, 06:14 PM

This is fantastic! Good work! I would like to see how our machines perform with Revit 2016 installations. Is there a date when the RFOBenchmark 2016 will be published?

Best Regards,
Matt

**Gordon Price** · August 8, 2015, 12:30 AM

Originally posted by mmetz

Is there a date when the RFOBenchmark 2016 will be published?

Thanks Matt! And, with any luck I'll finally be able to publish Monday. I hope to be looking for a few testers over the weekend. Just running some of my own tests now, and I have the basics working, plus the new "Heavy" graphics test, and a new "Update from previous version" test. I need to get the Heavy model creation journal done, and a few polish items, but it's finally darn close. Working vacation in Oregon after RTC has been productive.

**lwignall** · August 8, 2015, 01:06 AM

Gordon, happy to help with tests in our lab if that is useful. Also working with Autodesk on a request from a customer with their workloads that my team can run this against if you allow custom models? These are larger files as well.

**Gordon Price** · August 8, 2015, 01:12 AM

Originally posted by lwignall

Gordon, happy to help with tests in our lab if that is useful. Also working with Autodesk on a request from a customer with their workloads that my team can run this against if you allow custom models? These are larger files as well.

Luke, the new benchmark is modular, so one can build custom journals, plug them in to the benchmark (it's all basic XML driving it) and the script will manage things from there. I'll check in with you first thing Monday morning with some final code to test, unless you want to play in the benchmark sandbox over the weekend?

**Twiceroadsfool** · August 8, 2015, 01:13 AM

Hey Luke!

One of my clients just stood up a new VDI environment, with GRID cards as well. They will be opened for business for testing a week from Monday.

I snuck a peek at Gordons revised benchmark. Its pretty badass.

**lwignall** · August 8, 2015, 01:25 AM

Originally posted by Twiceroadsfool

Hey Luke!

One of my clients just stood up a new VDI environment, with GRID cards as well. They will be opened for business for testing a week from Monday.

I snuck a peek at Gordons revised benchmark. Its pretty badass.

If you can share results, builds, etc. we are always looking for field validation. We can only do so much in the lab to offer guidance, real users on real data is the ultimate set of results. Also curious about end user acceptance criteria, if any, used. We are being asked for this from almost everyone.

**lwignall** · August 8, 2015, 01:58 AM

Originally posted by Gordon Price

Luke, the new benchmark is modular, so one can build custom journals, plug them in to the benchmark (it's all basic XML driving it) and the script will manage things from there. I'll check in with you first thing Monday morning with some final code to test, unless you want to play in the benchmark sandbox over the weekend?

Working with team at Autodesk testing builds, this will go along way to helping, thanks for all the hard work! If you send it over I can't swear the NVIDIA Perf Eng will be able to leave it untouched until monday...LOL!

**Twiceroadsfool** · August 8, 2015, 02:31 AM

Originally posted by lwignall

If you can share results, builds, etc. we are always looking for field validation. We can only do so much in the lab to offer guidance, real users on real data is the ultimate set of results. Also curious about end user acceptance criteria, if any, used. We are being asked for this from almost everyone.

I'll share everything, when I have data. We've already brute force loaded it with a three modeling teams, one in a 1.2 GB model. The limiting factor per blade was ram, not graphics. But we weren't running lumion anywhere on the VDI yet. Hehe

**Gordon Price** · August 8, 2015, 04:46 PM

Luke, it's still Beta, and a little rough around the edges, but PM me with a direct email and I'll get you a link to a dropbox download.

Thanks!
Gordon

MONUMETRIC Video mmt-volt

RFOBenchmark

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Related Topics

MONUMETRIC Anchor B sticky footer - 7ba6