Nvidia GT120 v/s Nvidia Quadro 4000 Xbench Performance

I just replaced my GT120 with the Quadro 4000, I compared performance using Xbench 1.3 (probable not the correct tool) and I did not notice any difference in Quartz and Open GL performance, as a matter of fact I notice the results worse for the Quadro 4000 than the GT1200, do I have to use a different benchmarking software ?, I have installed the latest drivers (end of march) from the Nvidia/PNY site, do I have to install something like CUDA drivers ?, thanks

XBench is NOT a CUDA or OpenCL benchmark.

It’s an old benchmark, using 2D and basic OpenGL 3D functions where the bottleneck is more the driver itself than the performance of the GPU in middle-end (and high-end) GPU.
It was devised to show differences between GeForce 5200 and GeForce 4MX, not the actual GPU that we use in current Mac hardware.

Your Quadro 4000 is 5X faster whatever metrics your consider for CUDA or OpenCL development, and if I’m wrong, let me send you a GeForce GT120 in echange of your Quaro 4000 ;)

Beside that XBENCH an real old (bad) benchmark for modern macs, i recommend :
Try CUDA-Z, an really nice CUDA Tool (Mac OS X) which shows all informations about CUDA on your gpus.
Also has an TEST (benchamrk) tab which does some basic benchmark tests like bandwith test and basic integer/flonting point becnmarks.
YOu will see , with CUDA-Z huge diff in speed between those cards!


If you want to bench OpenCL (on OS X) use Luxmark - an real number cruncher for gpus using gpu with OpenCL code for raytraycing.
Even more speed diffs between slow and fast gpus.

Luxmark (OpenCL, OS X)