• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

DirectCompute benchmark

Actually there may be a way to find out sort of...

Start up 2 copies or more of the program and click start on each as quickly as possible.


EDIT: Infact this is far better for comparision as it will play to the strengths and weaknesses of the respective APIs.
 
Last edited:
Rroff said:
Actually there may be a way to find out sort of...

Start up 2 copies or more of the program and click start on each as quickly as possible.


EDIT: Infact this is far better for comparision as it will play to the strengths and weaknesses of the respective APIs.
Click to expand...

will give that a try shortly;)
 
If you can manage 4x 30+K scores then I will be impressed :D

4xDXDC.jpg
 
Last edited:
Looks like ATI have some work to do on their directcompute support then - it would appear they are forcing thread opptimization which is why they kicked ass in the wave demo I guess too.


At a guess they are still running the drivers configured for promotional useage (showing off directcompute performance in simple test apps) and probably haven't spent much time on threaded support.
 
Rroff said:
Looks like ATI have some work to do on their directcompute support then - it would appear they are forcing thread opptimization which is why they kicked ass in the wave demo I guess too.


At a guess they are still running the drivers configured for promotional useage (showing off directcompute performance in simple test apps) and probably haven't spent much time on threaded support.
Click to expand...

maybe thats what all the performance hype is about in 9.12's drivers:confused:
 
Who knows - I'd have expected them to get around 58-59K at stock tho... its really odd the 5770 aren't far behind the 5870 - would be interesting to see what scores they get when running multiple copies.
 
Just to be annoying... I disabled SLI and get the same scores - so its only running on the primary card anyhow :S
 
I don't think SLI needs to be enabled for DirectCompute stuff, it still uses both cards either way i think (Both GPU temperatures increase at the same rate, unlike the stability test benchmark in the other thread). Same as when using F@H you don't need SLI enabled iirc.

Unfortunatly GPU-Z doesn't show GPU usage for my cards to confirm this tho.
 
Ah could be - it only shows one directcompute interface even with SLI disabled... would have to disable one card in device manager to tell for sure.
 
Very strange... think until either the programmer releases more info or ATI update their drivers this isn't a very conclusive test.
 
Rroff said:
Very strange... think until either the programmer releases more info or ATI update their drivers this isn't a very conclusive test.
Click to expand...

I think maybe we should think about trying to package some OpenCL demos from Nvidia, AMD, Apple, whoever really, into a sort of rough-and-ready benchmarking tool. At least that way we could see how the cards compare under a wider range of loads.
 
You must log in or register to reply here.
Back
Top Bottom