Performance testing graphics cards.

[jhengel]

We are in the process of testing workstations. We are testing the Dell T3500 with the OCUS, a customized OCUS with our data and our own benchmark.

We are testing four different graphics card with this T3500. The Nvidia Quadra FX 580, 1800, 3800 and 4800. One would think that you would getdrastically different results.We are seeingthat the graphics relatedresults within the benchmarks seem to be pretty much the same.

So if we take the OCUS5 benchmark as an example we find that the graphics data only differs by 5-15 seconds. I'm wondering if the OCUS stresses the graphics card at all...

I was wondering what everyone else does for testing and stressing graphics cards or how they evaluate new workstations. Any help would be greatly appreciated.

Thanks,

John Hengel

-

925-422-7608

[BlaineProut]

John, in my experiences with Nvidia cards, the mid-level fx quadro cards (980, 1100, 1300, 1400, 1500, 1700 and now the 1800) are more than sufficientfor std CAD work. Even if you have adv. assemblies, and higher feature count parts...I have not seen higher level cards return value other than outside of MCAD usage...I have seen value in going above this level for non-proe usage, such as heavier animations in simulation (Flow and structural) in other software apps, and that should be included in your study as well..

Bottom line, the fx1800 is the card to go with unless you have other non-MCAD tools that can benefit from the higher capabilities of the higher level cards.

Hope this helps.