3 Replies Latest reply on Feb 15, 2011 10:18 AM by Greg Jankowski

benchmark test results

Michael Bonitatibus Feb 8, 2011 3:06 PM

I have two questions about the benchmark tests in SW2011.

Can some one explain what the "render" test results under 1 mean?

Specifically why would one instance of an X 920 CPU have a value of 0.6 (with a Quadro FX 2800m) and in another

instance be over 54.3 ( with a Quadro FX 3800m) ?

Both instances are Dell Precision M6500 platforms and both video cards are high end.

Why are sub $200 video cards performing on par with high end cards?

there doesn't seem to be a clear winner on performance from the test results.

given that the high end CPUs would be partnered with high end graphics boards it is supprising

to see a Quadro 600 getting the same test results (31.5) as a Quadro 4000.

Shouldn't the 4000 be 5 times faster?

Re: benchmark test results

Russ Johnston Feb 8, 2011 3:09 PM (in response to Michael Bonitatibus)

Video cards have NO affect on rendering times. Rendering is CPU dependant.
Like Show 0 Likes(0)

Actions
Re: benchmark test results

Florent Dufils Feb 9, 2011 3:51 AM (in response to Michael Bonitatibus)

The test result is the total time, in seconds, spent during the test calculating renderings.
I assume that very short times (like 0.6 you reported) come from computers with no PhotoView license for the test to run.
Like Show 0 Likes(0)

Actions
Re: benchmark test results

Greg Jankowski Feb 15, 2011 10:18 AM (in response to Michael Bonitatibus)

There are two questions here:

1) A score of under 1 is not valid, we will investigate what is happening here.

2) This version of the performance test does not push the graphics card. The first version was focused on CPU and disk performance tests. We would like to better test/differentiate graphics cards and need to find a way to provide valid tests and will work, or handle, card that have advanced capabilities while still allowing lower end cards to be tested and separate these cards from the workstation grade graphics cards. We are looking into how to provide meaningful numbers.

I have also explained more about the test in the post https://forum.solidworks.com/community/administration/hardware_and_os/blog/2011/01/04/solidworks-performance-test.
Like Show 0 Likes(0)

Actions