Quadro FX 1000 Viewperf results

[GregHess]

Hey Lee!

What was the spec system's cpu rated at?

Remember that even if your testing the same two cards, cpu speed is still a variable in the performance of 3d acceleration.

[CgFX]

More specifically, the memory bandwidth of the CPU plays a huge roll as this is what is used to send all those millions of polygons in "immediate mode" to the graphics processor for each and every frame.

This is why the P4 is always used in workstation submissions vs. an Athlon. P4's may spin their wheels needlessly for higher marketable clock rates compared to Athlons but they have traditionally had higher memory bandwidth which is good for the graphics.

[Amyd]

If he tested his stuff on the system in his signature (Dual Xeon 2.8), there are very few more powerful alternatives that could have give Spec scores (actually none, as most of the current 7.1 Viewperf results in the database are made on single processor 3Ghz Pentium IVs).

Indeed, the scores are very very low, from 2 times up to almost 5 times lower than they should be.

What drivers did you use? And what OS?

Cheers,
Andrei

[CgFX]

He may have something wrong but...

Viewperf is not multi-threaded and a 3.06 GHz P4 has more memory bandwidth than a single 2.8 GHz Xeon.

[LeeTN]

The Specviewperf systems were all single cpu Pentium 4's ranging between 3 GHz and 3.2 GHz.

And yes, the system in my sig is the one I ran the tests on.

However, I think that I may have screwed the test by opening some other programs while it was running, a client called and wanted some info, so....also there were a lot of "file not found" errors during the test. The reason I ran the test I did is because I do not have Maya, Lightwave, etc... installed on the system and the viewperf test is supposed to simulate results of performance for those apps without the need to have them installed.

I will run the tests again, this time without interruption.

[Amyd]

Originally posted by CgFX
He may have something wrong but...

Viewperf is not multi-threaded and a 3.06 GHz P4 has more memory bandwidth than a single 2.8 GHz Xeon.

Yes, however not 2 to 5 times more. The difference should be pretty small.

For instance, take the Dell 360 vs. IBM M Pro results. Dell uses a 3 Ghz FSB 800 chip with PC3200, while M Pro uses a 3.06 Ghz FSB533 chip with the same kind of memory Lee's system has, ECC PC2100. The differences when tested with the FX1000 are usually less than 1% between the systems, despite the big difference in memory bandwidth.

LeeTN, that sounds like a very probable cause for the low results (even though hehe the dual processor should have handled that more gracefully). Also try disabling any background activity, like antivirus programs, firewalls & so on.

Cheers,
Andrei

[LeeTN]

Yeah, the biggest benefit of the multiprocessor setup is the ability to run multiple programs without any slowdown, gotta love it. I frequently am working with InDesign, PhotoShop, Illustrator and GoLive all open simultaneously, transferring images and text between them.

Will disable everything and try the tests again. Will post results afterwards.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.