We just finished running the OCUS and Spec benchmarks against three different graphics cards. We tested the Nvidia Quadro 600, 2000, and 4000.We see around an 18% difference between the 600 and 2000 but only around 5% between the 2000 and 4000. Needless to say we were disappointed we didn't see better performance in the 4000. Any one else run the benchmarks? Are there profile setting that will help with the perfomance? We are getting ready to run the same tests against three ATI cards next week. Does Creo use any CPU for graphics related tasks? If so, why is it not using the GPU for all graphic related tasks? I'm assuming it has to do with how PTC codes OpenGL with Creo.
Any input is appreciated.
I would also appreciate if I get detailed information concerning this topic. I guess that most of a load is put on CPU when user manipulates with shaded model. So why PTC recommends users to buy expensive graphic cards ?
My machine is a quad-core Xeon W3550 with nVidia FX3800, and running WF4.
Using GPU-Z, the GPU load hits maybe 30% while spinning a reasonable complex assembly, in shaded with edges on, whilst the CPU sits at 25% (i.e. one core maxed out).
With edges off, it spins rather more fluidly and the GPU reaches perhaps 40% while the CPU maxes one cor.
In hidden line, it's even slower than the shaded with edges, and now the GPU is only around 10%.
Even viewing shaded with a transparent gearbox casing only runs the GPU around 15%, although there's a spike to 100% when first switching to transparent and again when I stop spinning the model.
Would be interesting to hear whether Creo 2.0 shifts more of the load to the graphics card...
Maybe you should try ati graphics. I saw this whitepaper and it looks really interesting...
I have seen that whitepaper and we will be testing the ATI cards soon. Our biggest slowdown is in hidden line mode and I'm not sure the OIT technology will help that.