Systems used:- Motherboards - Asus P5N32-E SLI,P5N-E SLI Processors Used - E6600 and E6550 stock Cards used - 1x XFX 8600GT,2x XFX 8800GTS 640MB,1x XFX 8800GT Power Supplies Used - Corsair HX620W and CMEP 600 Ram - 2GB DDR II 667Mhz Kingston Software used - MKZ demo,Nurien Tech Demo,GRAW 2 Ver 1.2 Settings Used - 1600*1200 PhysX High,Graphics Settings Highest. Benchmark Result For 8600GT(PhysX) 8800GTS(Video) - Frames: 1797 - Time: 50855ms - Avg: 35.335 - Min: 12 - Max: 48 Benchmark Result For 8800GT(PhysX) 8800GTS(Video) - Frames: 1464 - Time: 45025ms - Avg: 32.515 - Min: 11 - Max: 44 Benchmark Result For 8800GTS(Single) - Frames: 2962 - Time: 88927ms - Avg: 33.308 - Min: 11 - Max: 52 Benchmark Result For 8800GT(Single) - Not Available Benchmark Result For 8600GT(PhysX) 8800GTS(Video) - Benchmark Result For 8800GT(PhysX) 8800GTS(Video) - Benchmark Result For 8800GTS(Single) - Benchmark Result For 8800GT(Single) - Benchmark Result For 8600GT(PhysX) 8800GTS(Video) - Benchmark Result For 8800GT(PhysX) 8800GTS(Video) - Benchmark Result For 8800GTS(Single) - Benchmark Result For 8800GT(Single) - The 8600GT performed very poorly in the MKZ test.The scaling wasn't much different for Nurien tech demo.
Thanks for taking the time to do this, some very interesting results, especially the GRAW2 results where using the 8800gt for PhysX is slower than the 8600gt. For the best part though, it seems the 8600gt for PhysX isn't worth using over a single 8800gts, but maybe in games which are really heavy on the gpu that might change.
Thanks for the input.The performance i saw with 8600GT physx on MKZ demo was very poor.So the newer games like bionic commando which are going to use PhysX technology might run slower with 8600GT being used as PhysX.
I have xp-sp3 and geforce 177.83 w/ physx 8.09.04 and I got: avg-36.14* max-49.78 min-28.02 I have vista-sp1 and geforce 177.83 w/ physx 8.08.01 and I got: avg-32.33* max-51.148 min-22.39 *- without antialising but... are you running xp or vista? do you have antialising?
kinda hard to understand this, i used 3 cards as a physx card (8400gs, 8600gt, 9600gt) and the last result(9600gt) was better than the others, but here in your report i see that using an 8800gt is lower than using an 8600gt so :s for me using my 9800gtx+ as a main vga an 9600gt as a physx dedicated is great.. i thing all is about differents posibles ways to setup this.. (cpu, motherboards ram, OS )
I used XP 32-Bit.Well the 8600GT performed well in only the GRAW2 benchmark.The benchmark involved me exploding fuel tanks. And yes the 8800GT or 9600GT performs way better than an 8600GT in PhysX.
GRAW2 does not support normal AA,instead it uses blurring techniques to Blur the zagged edges which destroys the crispness of the image.As for other benchmarks,they were tested at default settings.
Thanks for this reports...it proves what I was thinking...I made a good choice by keeping my 8800GT for physx vs going with a cheaper card.
is it me or is physx is broking under vista i get same scores when i enable 8800gt as physxgpu n disabling physx n run on cpu. n score goes up when i enable gtx280 as physx n main gpu.or is the driver been using 180.42 on both xp 32 n vista 64... n xp works with enabling 8800gt as physx gpu . both test where run with cpu o.c to 4gz n i used MKZBenchmark_lowResolution (1280x1024) af x8 aa x4 bilinear clamp high performance vsync force off both os's
Well the newer drivers have a bug in vista even though 8800GT is selected PhysX is not enabled on 8800GT.To find a way around this problem simply goto display properties after selecting customization from right clicked context menu and forcibly extend the desktop to 8800GT then re-enable physx on 8800GT.Hope this helps.
I think newer driver .48 just came out fix the issue in vista64 ,but xp still better lol. XP MAX FPS WAS IN 90'S VISTA I THINK WAS IN 80'S NOW
I don"t believe the Vista x64 bug is fixed with the 180.48 drivers. I downloaded the new drivere and ran the same two test above (MZK and Nurien demos) with my system and if I did not connect and enable the 9600GT (PPU) as a second monitor when using my 280 GTX as the GPU the scores were very low. In the MZK bench it would render in software mode if i did not connect and enable a second monitor. ( for both the above circumstances the 9600GTwas selected as the PPU in the NVIDIA Control panel). I still get better results in all the Physx tests if I let the 280GTX handle doing both GPU and PPU by it self.