Project: Find the bottleneck?

[lite12]

I'm just having fun with overclocking tests. Keep in mind I'm not as knowledgable as the majority of you out there, so "patience is priceless".

I wanted to give the system a stress-test on the FSB and see how well it would perform against a low FSB and high multiplier.

Using a Asus P4P800 board coupled with a P4 3.06Ghz (23x133) @ 3480Ghz (12x290), it booted and ran all CPU benchmarks flawlessly. However when it came time to run the performance tests with video (anything with 3DMark or games as a stress test) the benchmark program would predictably quit. Usually crashing to the desktop without any error message.

I'm using 512Mb kingston for memory, and although a Vcore of 1.2 volts seems to run fine I've also tested with 1.55 and 1.65 with the same crashing-to-desktop pattern.

My video card is a temporary one (Geforce 4 Ti 440) until the Radeon arrives in the mail. This system will eventually use a Radeon 9600XT or Radeon 9800 Pro A.I.W.

Any thoughts or recommendations?

Will this error likely persist with the newer video card?

Thanks,

-Lite

[batboy]

Unless you have an engineer sample, Intel multipliers are locked. What mobo? I doubt you'll ever be stable at a decent overclocked speed while using 1.2v.

[lite12]

Aside from crashing to desktop with video-intensive programs, there doesn't seem to be instability. The system boots and CPU benchmark programs work well whether I set it at 1.2volts or 1.8volts.

It's an Engineering Sample chip with an Asus P4P800S.

-Lite

[batboy]

Crashing to desktop is a sign of instability. I would not exceed 1.65v. Ok, 290 FSB is sometimes tough to do on some mobos. I still don't know what mobo. At those high speeds, you need good northbridge cooling. Maybe try 13X267 FSB and see what happens? That will be about the same top speed. If it does the same thing, then your CPU is probably maxed out. If that's stable, then your system don't like high FSB. Don't know what kind of RAM you have, but maybe try a lower memory ratio to eliminate the RAM as being a bottleneck.

[coil]

I used to crash to desktop when running 3dmark, it was my memory causing it so i increased the voltage till it was ok.

[Albuquerque]

I had a 2.4C that would hit 284FSB (3.4ghz) reliably, but even 1mhz FSB faster would make it unstable no matter the voltage. Turned out to be the board; it just didn't like anything higher. I'm using that same board now in the system in my signature, it's currently at home testing at 253FSB / 3.8ghz without any problems.

[lite12]

Thanks for the replies!
I did a number of things (pretty much everything recommend in this thread), in addition to installing the new Radeon card in place of the Geforce. Everything looks good and performs well on the 3dmarks. = D

-Lite