JezEngland
New member
Hello,
Guess what I got for Christmas......................
9800 GT
I got it running PhysX in a PciE 1.0 (x16) slot @x4. Its the third x16 slot on the Maximus 3 Formula motherboard. The other 2 are populated with GTX 480's, each at the @x8 PciE 2.0 setting, because it is a P55, Intel i5 750 build. The i5 is OC'd to 3.7Ghz.
I did tests with & without APEX PhysX, at maximum graphic's settings, at 1920x1080 screen resolution and recorded the average frame rate result on the 2nd run for each configuration with the Mafia 2's benchmark, with 9 different GPU setups configured in the Nvidia Control Pannel.
I just realised the Nvidia 3D settings in the control pannel are higher than default. the 480 may be producing a lower frame rate.
Here are the results:
2 GTX 480 SLI-APEX PhysX off: I didn't bother finishing this run, the fps started out over 100/150
2 GTX 480 SLI+9800 GT for APEX PhysX @ medium setting: 62
2 GTX 480 SLI+9800 GT for APEX PhysX @ high setting: 46
1 GTX 480-APEX PhysX off: 64
1 GTX 480+1 9800 GT for APEX PhysX @ medium setting: 50
1 GTX 480+9800 GT for APEX PhysX @ high setting: 43 - 9800GT max GPU usage: 74%
1 GTX 480+1 GTX 480 for PhysX @ high setting: 47 - PhysX dedicated GTX 480 max GPU usage: 69%
1 GTX 480 also doing APEX PhysX @ medium setting: 43
1 GTX 480 also doing APEX PhysX @ high setting: 37
This is my conclusion:
Some thoughts: The 9800 GT's core clock is 600Mhz. The GTX 480's core clock is 700 Mhz. The 480 doing the APEX PhysX setup's average frame rate(47) is between 1 and 4 frames faster than the 9800 GT's setup(43/46), with APEX PhysX @ high setting. Core clock frequency must be(probably) a crucial factor for driving PhysX. The 9800GT could have more input for this benchmark because GPU usage results maxed at 74%, and the GTX 480 dedicated to PhysX max usage was 69%. I rekon a heavily overclocked 9800GT core frequency can put a better result, better than the stock 480? mm maby.
Nvidia drivers and game code updates will erm, always/sometimes(sometimes they make mistakes) improve results. The logic of scaling(or watever it's called) is crippled by Mafia 2's APEX PhysX. The decrease in performanc with APEX PhysX working is big. I think Mafia 2 runs PhysX when APEX PhysX is off, ? playing without APEX the long jackets movement looked the same
i dunno
PCIE 1.0 @ x4 could be making a small difference though I'm not convinced
In Metro 2033 the 9800 GT's usage averaged at 10% and max was 22% and overall performance is better.
So the mk1 VW Golf is faster than some of the updated Golf GT's
Thanks
Guess what I got for Christmas......................

I got it running PhysX in a PciE 1.0 (x16) slot @x4. Its the third x16 slot on the Maximus 3 Formula motherboard. The other 2 are populated with GTX 480's, each at the @x8 PciE 2.0 setting, because it is a P55, Intel i5 750 build. The i5 is OC'd to 3.7Ghz.
I did tests with & without APEX PhysX, at maximum graphic's settings, at 1920x1080 screen resolution and recorded the average frame rate result on the 2nd run for each configuration with the Mafia 2's benchmark, with 9 different GPU setups configured in the Nvidia Control Pannel.
I just realised the Nvidia 3D settings in the control pannel are higher than default. the 480 may be producing a lower frame rate.
Here are the results:
2 GTX 480 SLI-APEX PhysX off: I didn't bother finishing this run, the fps started out over 100/150
2 GTX 480 SLI+9800 GT for APEX PhysX @ medium setting: 62
2 GTX 480 SLI+9800 GT for APEX PhysX @ high setting: 46
1 GTX 480-APEX PhysX off: 64
1 GTX 480+1 9800 GT for APEX PhysX @ medium setting: 50
1 GTX 480+9800 GT for APEX PhysX @ high setting: 43 - 9800GT max GPU usage: 74%
1 GTX 480+1 GTX 480 for PhysX @ high setting: 47 - PhysX dedicated GTX 480 max GPU usage: 69%
1 GTX 480 also doing APEX PhysX @ medium setting: 43
1 GTX 480 also doing APEX PhysX @ high setting: 37
This is my conclusion:




Some thoughts: The 9800 GT's core clock is 600Mhz. The GTX 480's core clock is 700 Mhz. The 480 doing the APEX PhysX setup's average frame rate(47) is between 1 and 4 frames faster than the 9800 GT's setup(43/46), with APEX PhysX @ high setting. Core clock frequency must be(probably) a crucial factor for driving PhysX. The 9800GT could have more input for this benchmark because GPU usage results maxed at 74%, and the GTX 480 dedicated to PhysX max usage was 69%. I rekon a heavily overclocked 9800GT core frequency can put a better result, better than the stock 480? mm maby.
Nvidia drivers and game code updates will erm, always/sometimes(sometimes they make mistakes) improve results. The logic of scaling(or watever it's called) is crippled by Mafia 2's APEX PhysX. The decrease in performanc with APEX PhysX working is big. I think Mafia 2 runs PhysX when APEX PhysX is off, ? playing without APEX the long jackets movement looked the same


PCIE 1.0 @ x4 could be making a small difference though I'm not convinced
In Metro 2033 the 9800 GT's usage averaged at 10% and max was 22% and overall performance is better.
So the mk1 VW Golf is faster than some of the updated Golf GT's

Thanks