Quick OC question.

Man of Honour
Joined
17 Feb 2003
Posts
29,640
Location
Chelmsford
Does increasing the frequency of the CPU increase the speed of the SM on the graphics card?

I didn't think it did but according to trend, 3DMark06 reports that it does.

CPU@ 2.7Ghz, SM2.0 = 4897
CPU@ 3.0GHz, SM2.0 = 4972
CPU@ 3.2GHz, SM2.0 = 5070

Spec:

Q6600
8800GTX
Asus P5KC
4GB Geil DDR PC6400
2 x 500GB Spinpoint HDD
 
Last edited:
You could well be increasing the PCI-E frequency, which would lead to increases i ngraphics card speed...

Alternatively, there is the theory of faster CPUs helping the graphics card to 'cut loose' as it were.
 
I see.

Interesting, I conducted the the same testr on the same system and overclock but using a 8800GTS (512MB)

@3.2GHz, SM 2.0 = 5405
@3.0GHz, SM 2.0 = 5366

This is more than the 8800GTX 712MB used on the OP.

:confused:
 
Last edited:
Indirectly, yes. It doesn't increase the shader speed, but it will give you more points.

By overclocking the CPU, you are reducing a bottleneck and allowing the graphics card to work faster.

In tests such as 3DMark, the graphics are quite CPU limited, even if a better graphics card will give more points. This is why the CPU overclock is important. If we're talking about 3Ghz-3.2Ghz, then you'll see a gain, if we're talking about 5Ghz-5.2Ghz then you're unlikely to see the same gains as the CPU bottleneck is practically removed.

AFAIK, the 9600GT is the only card that can have it's core speed overclocked if you overclock the PCI-E frequency, which is normally locked (or at least lockable) on most motherboards.
 
Last edited:
I would've expected so, but maybe not.

Were the two cards stock, or was the GTS a pre-OC'd card?

The only explanation is that as a very basic graphics test at 1280x1024 with no AA, the higher clocked G92 core can run faster. If you ran the same tests at your screen's maximum res, and/or at >4xAA, then I'm sure things would turn around.
 
Back
Top Bottom