Does CPU OC effect GPU Performance

[abvolt]

How much does CPU OC actually effect the performance of our GPUs. Some people say and I have read this on different boards that if you run your CPU in stock speed vs OCed, it will effect your FPS in a major way while gaming which seems to be true but I've noticed that when benching my OCed GPU with my CPU OCed there is very little if any difference only while benching the GPU.

I would think that a OCed CPU with an OCed GPU is best for benching and even gaming. Is it best to downclock your CPU when benching your GPU.

[shakaw]

In my experience CPU OC does not affect the GPU (same graphics score with OC'ed and non-OC'ed CPU), but it can (and does) affect your FPS in games and benchmarks.
In my case, OC'ing the CPU got me about a 1000 extra in valley. There is also quite a large score increase in 3dmark Firestrike due to CPU overclock because of the physics test that's run on the CPU.

[Zka17]

I think, it is application dependent...

[Arne_Saknussemm]

For some games and some benchmarks the CPU can be a limiting factor, a bottleneck so OCing it of course will reduce the bottleneck and give a performance increase.

Many benchmarks include a CPU score so the higher your CPU OC the better your score. But if you look at the graphics score part only (assuming no bottleneck) there is virtually no effect from OCing your CPU. If you run 3Dmark stuff at stock or 5.0 ghz your CPU scores ( and hence your overall scores) go up but your GPU scores stay the same for any given GPU OC.

Of course there are OCs on the CPU made with straps that directly affect GPU BCLK which might increase or reduce GPU performance....

[HiVizMan]

List the benchmark and the games and I will tell you which is CPU dependant and which is VGA dependant.

Let me give you an example.

Multiple VGA cards benching 3DMark Vantage with the top end VGA is all about the CPU not the VGA.

[abvolt]

When I first oc'ed my cpu I noticed that it made a big difference in score with cinbench & heaven but with 3d mark not very much. Im more interested in benching my vga & thought if the cpu is also oc'ed I would see big increases in scores not so I guess sli is the best way for big gains.

[HiVizMan]

That is not really true. It all depends on what level you are benching at.

90% of all VGA benchmarks are now CPU bound, CPU core speed is more important than any VGA clock these days. There are a couple that this is not the case but ....

[abvolt]

HiVizMan wrote:
That is not really true. It all depends on what level you are benching at.

90% of all VGA benchmarks are now CPU bound, CPU core speed is more important than any VGA clock these days. There are a couple that this is not the case but ....

Would it also be wise to oc the vga ? sometime when I do things get unstable fast that sweet spot seems to elude me more often than not.

[HiVizMan]

When you OC multiple things at once there is an exponential increase in the possible things that can go wrong. That is why you experience what you do.

I always follow a set formula when OCing a new system. Memory first, then CPU and finally the VGA. And with VGA I find the memory max first, then the core max and then drop them both by 100Mhz when combining their OC.

[chrsplmr]

HiVizMan wrote:
When you OC multiple things at once there is an exponential increase in the possible things that can go wrong. That is why you experience what you do.

I always follow a set formula when OCing a new system. Memory first, then CPU and finally the VGA. And with VGA I find the memory max first, then the core max and then drop them both by 100Mhz when combining their OC.

This is why I read this forum diligently - truisms of Gold. This one's a keeper. Short, sweet,neat. Thanks Viz. .c.