Henrik
New member
Hey guys!
I recently got a little upgrade from my GTX280 to a GTX580. So I was wondering what i could do with the 280 that's was just collecting dust...
Why not use it as a dedicated PhysX card i thought. So I installed it in my system wondering what kind of impact it would have on my system. These days the PhysX calculation on the graphic cards are so good that I thought i would get very little if any FPS gains, but what the heck i gave it a shot.
System Specs:
QX9650
XFX 790i FSB @ 1700Mhz
4gig of Corsair RAM @ 1700Mhz
EVGA SC GTX580 (main GPU)
GTX280 (PhysX card)
2x Corsair F60 in RAID 0
I tested three scenarios:
1. PhysX calculated on the GTX580.
2. PhysX (and PhysX only) calculated on the GTX280.
3. PhysX calculated on the CPU.
I started of with 3DMark Vantige with preset on performance. I "disabled" Texture fill, Color fill, POM and Perlin noise from the feature test (as I felt they where irrelevant for PhysX testing) and only left GPU cloth and GPU Particles on in the feature selection.
Remember the GTX580 is allways the GPU doing the main calculations.
First out is System at stock speeds (CPU @ 3.0Ghz, GPU @ 797Mhz)
Ass you can see when PhysX is set to the CPU, the P-score and CPU-score take a huge impact..
The difference between PhysX on the GTX580 vs the GTX280 is not too bad but the winner here is with PhysX set to the 580.
Next I overclocked the GTX580 to 860core 2100mem and ran the same test again to see if there would be any gains. (CPU @ 3.0Ghz)
Same story here. But we see a increase in P-score when PhysX is set on the 580 but not when set on the 280...
Are we seeing the 280 bottlenecking the 580??(I don't think so) I think we might have a CPU limitation here!
Now we have the GTX580 at the same speed as before but the CPU is now overclocked to 4,0Ghz.
This is where it starts getting interesting! We see all the scores taking a significant jump up. But most interesting is that the P-score when PhysX is set on the 280 increased almost 4000 from previous test. Getting really close to the P-score of the 580 alone. I think we can assume that running a dedicated PhysX card takes its toll on the CPU..
Here is a cart to sum up all the above results. I will let you make your one conclusions.
Maffia 2 testing with PhysX
Lets see what impact these configurations have on gaming. What better game to test PhysX on then Maffia2??
I tested three scenarios:
1. PhysX calculated on the GTX580.
2. PhysX (and PhysX only) calculated on the GTX280.
3. PhysX calculated on the CPU.
Remember the GTX580 is allways the GPU doing the main calculations.
Benchmark settings:
Video Mode: 1680x1050
Antialiasing: On
Anisotropic Filtering: 16x
Shadow Quality: High
Ambiant Occlusion: On
Geometry Detail: High
APEX PhysX: High
The difference between PhysX set to the 280 and 580 is consistant with about 4-5FPS difference.
Its only when the CPU is overclocked to 4.0Ghz that the gains of overclocking my GTX580 and running the GTX280 as dedicated PhysX card is significant.
To summarize it all I guess that running a dedicated PhysX card will gain you a little fps increase assuming you have the CPU power!
In games that have none or very little PhysX I would definitely recommend running PhysX on your main card as it could drag your FPS down..
Pleas don't see this as some kind of showoff of my system in any way. I was just curious myself and wanted to test the difference in performance. But i thought i could share it with you guys if there is someone that's thinking of putting in a dedicated PhysX card.
I recently got a little upgrade from my GTX280 to a GTX580. So I was wondering what i could do with the 280 that's was just collecting dust...
Why not use it as a dedicated PhysX card i thought. So I installed it in my system wondering what kind of impact it would have on my system. These days the PhysX calculation on the graphic cards are so good that I thought i would get very little if any FPS gains, but what the heck i gave it a shot.
System Specs:
QX9650
XFX 790i FSB @ 1700Mhz
4gig of Corsair RAM @ 1700Mhz
EVGA SC GTX580 (main GPU)
GTX280 (PhysX card)
2x Corsair F60 in RAID 0
I tested three scenarios:
1. PhysX calculated on the GTX580.
2. PhysX (and PhysX only) calculated on the GTX280.
3. PhysX calculated on the CPU.
I started of with 3DMark Vantige with preset on performance. I "disabled" Texture fill, Color fill, POM and Perlin noise from the feature test (as I felt they where irrelevant for PhysX testing) and only left GPU cloth and GPU Particles on in the feature selection.
Remember the GTX580 is allways the GPU doing the main calculations.
First out is System at stock speeds (CPU @ 3.0Ghz, GPU @ 797Mhz)
Ass you can see when PhysX is set to the CPU, the P-score and CPU-score take a huge impact..
The difference between PhysX on the GTX580 vs the GTX280 is not too bad but the winner here is with PhysX set to the 580.
Next I overclocked the GTX580 to 860core 2100mem and ran the same test again to see if there would be any gains. (CPU @ 3.0Ghz)
Same story here. But we see a increase in P-score when PhysX is set on the 580 but not when set on the 280...
Are we seeing the 280 bottlenecking the 580??(I don't think so) I think we might have a CPU limitation here!
Now we have the GTX580 at the same speed as before but the CPU is now overclocked to 4,0Ghz.
This is where it starts getting interesting! We see all the scores taking a significant jump up. But most interesting is that the P-score when PhysX is set on the 280 increased almost 4000 from previous test. Getting really close to the P-score of the 580 alone. I think we can assume that running a dedicated PhysX card takes its toll on the CPU..
Here is a cart to sum up all the above results. I will let you make your one conclusions.
Maffia 2 testing with PhysX
Lets see what impact these configurations have on gaming. What better game to test PhysX on then Maffia2??
I tested three scenarios:
1. PhysX calculated on the GTX580.
2. PhysX (and PhysX only) calculated on the GTX280.
3. PhysX calculated on the CPU.
Remember the GTX580 is allways the GPU doing the main calculations.
Benchmark settings:
Video Mode: 1680x1050
Antialiasing: On
Anisotropic Filtering: 16x
Shadow Quality: High
Ambiant Occlusion: On
Geometry Detail: High
APEX PhysX: High
The difference between PhysX set to the 280 and 580 is consistant with about 4-5FPS difference.
Its only when the CPU is overclocked to 4.0Ghz that the gains of overclocking my GTX580 and running the GTX280 as dedicated PhysX card is significant.
To summarize it all I guess that running a dedicated PhysX card will gain you a little fps increase assuming you have the CPU power!
In games that have none or very little PhysX I would definitely recommend running PhysX on your main card as it could drag your FPS down..
Pleas don't see this as some kind of showoff of my system in any way. I was just curious myself and wanted to test the difference in performance. But i thought i could share it with you guys if there is someone that's thinking of putting in a dedicated PhysX card.
