Celeron 566 Overclocking Problem

[Ruger]

Hi. I've just purchased a new, boxed version of the Celeron 566 with the latest cCo stepping, 1.7 default vcore. My system includes an ASUS PB3-F motherboard with BIOS revision 1006 (Coppermine support), and a Startech FC-PGA slocket.

I'm attempting to run this Celery @ 850mhz clock, (100mhz FSB). Computer will boot into Windows just fine and perform all generic tasks, run DVD movies, ect. without a hitch. The Celery passes the CPUTest program after a thirty minute load, and will even run the Unreal Flyby mode with no problems.

When running Falcon 4.0, however, I get random lockdowns and display corruptions. Game runs fine at normal clock.

Since I'm running at 100mhz FSB, I know the PCI bus is running at normal speeds. I'm using PC100 RAM as well, so I can pretty much trace the lockdowns to the new Celeron. I am also using the ASUS cpu temp monitoring software (not sure how accurate it is with the Startech Sloket), but with the Celery normally clocked I get baseline indicated CPU temps of around 30c at idle and 35-40c under load. Overclocked I'm getting to 50c under load. I'm only using the retail heatsink with silicon based thermal grease and running with my case open.

Is there any way to determine if my problem is directly heat related or rather simply a CPU instablility issue?

I have tried using a portable fan next to my open case, and this did buy me a temp reduction all around with both the CPU and motherboard, but I'm still getting the lockdowns with Falcon4.

Thank you for any advice you can offer. Great board forum here gentlemen!

[Ruger]

As an addendum to my above post, I have also tried vcore's of 1.75, .1.80, and 1.85, all with no effect on the lockdowns when running Falcon4. Thanks again.

[Tim-]

50c is a little high- My Celermine doesn't like to be too stable when it's that warm. A new cooler is definitely called for. It may be an issue of the chip's ability to cool itself fast enough by transferring heat into the heatsink. This can happen if the thermal compound is applied a little too thick also. You may also try bumping the I/O voltage up to 3.4 or 3.5v to give the video card a little help.

[DocClock aka MadClocker]

Arctic Silver compound reduced my temps by a couple of degrees. Really this stuff did what was advertized, I averaged a 6/7 deg drop just by using arctic silver. but if you buy it, follow the directions especialy when it comes to removing previous compounds and cleaning the surfaces. This stuff could make the difference in your stability.
One other thought, this could be the lack of cache rearing it's ugly head..as speed increases could the need for cache become more important, becoming the bottleneck?
Hmmm

[Ruger]

Thanks for the reply. I am definitely going to install a high performance heatsink/fan combo at this point. One thing I wanted to follow up with you about increasing the voltage supply to my PCI video card - since I'm overclocking at 100mhz FSB, my video should be at the standard 33.3mhz PCI bus speed. Why would I need to increase the voltage for more stability? Thanks again.

[Ruger]

DocClock aka MadClocker (Jan 03, 2001 03:39 a.m.):
Arctic Silver compound reduced my temps by a couple of degrees. Really this stuff did what was advertized, I averaged a 6/7 deg drop just by using arctic silver. but if you buy it, follow the directions especialy when it comes to removing previous compounds and cleaning the
urfaces. This stuff could make the difference in your stability.
One other thought, this could be the lack of cache rearing it's ugly head..as speed increases could the need for cache become more important, becoming the bottleneck?
Hmmm

Thanks for the reply. The Celery's do have 128k of internal cache running at the processor core speed, while the PIII Coppermines of course have 256K. How could overclocking the Celerys lead to problems with cache? Thanks again.

[forsure]

50c is definitely too hot. I have my 566 oc'ed to 851 (100x8.5) and temp stays in the 32 area. I use a golden orb fan and silicone compound.

[Ruger]

Thanks. I think I will indeed try the orb out with some silver oxide goo.