FSB versus GPU core increases

[maluckey]

O.K. I'm looking for opinions or answers to something. After two weeks of toying around with my system, I see a pattern. The pattern is that as the FSB rises, all is well, then it stutters, then resumes AOK at one or two mhz faster (could this be resonant frequency interference, or harmonics?). Then, as I raise the FSB more, performance (FPS in games) begins to fall off, despite the bandwith increase with memory, and the CPU frequency getting faster. If I drop the GPU speed, then I can regain the lost FPS, and in most cases, gain more than I started with. Video memory is less susceptible to this, and is almost set and forget The only thought that I have on this, is AGP bus saturation, but thought that it would saturate, and not increase at all beyond a point. The system hangs around the. There is no PCI lock on this board, and 155 would be the limit to stay in range I guess. I'm not very up on this stuff, and learn mostly by breaking things, then fixing them, and in this one, I'm stumped. The system rocks, and gets great FPS in UT2003. I'm just curious as to the reasons for all this.

The system in question is:

ABIT KR7A-133 (KT266a) - tested to 169mhz FSB, probably will go higher, many have
Gaiward Ultra650 XP 128mb Golden Sample - tested stable to 324mhz/599mhz
Corsair XMS pc3200 - stable to....hell, I can't crash it yet.
OS=XP Pro SP-1

 

[StraightPipe]

I'm not an over clocker, but i've read quite abit.
I'd imagine that as you bump the FSB speed it becomes less stable, and as FSB gets less stable there is less of a performance increase.

at some point there is a maximum yield due to your OC. thats why benchmarking is so important.

 

[maluckey]

True, the stability decreases as the stress rises, and that's why you have to raise the voltage, etc. But that doesn't explain the inverse nature of the GPU results, unless it has some built in thermal protection/slowdown.

I am like you in the respect that I read .........lots. I guess that's why this problem vexes me. Not one article that I've read, fully explains this phenomenon. I've seen it mentioned, but never actually experienced it, till now.

Thanks for the reply! Mark