366@550 failure - What am I doing wrong?

[Rmcky]

I think your problem lies in your application of thermal paste. When you press it together and twist it slightly, then break it apart, you should have total coveage with a crater patern on both pieces. If you see any flat smooth areas, apply a little more paste to them and smooth out the rest, then try again. when it's right, put it back together, clip it on, and leave it alone. It sounds like you are being a little too conservative with the grease. You don't want too much, but atleast total coverage. Hope this gets you over the hump.
Rick

[daveleau]

Exactly, might be a little too much paste. also did he warranty it at 2.0V? If not you can go up to 2.3V on these chips safely. That would help stability. Use RAIN if you aren't running RC5 or Seti@home. That will help as well. If all else fails, you must be holding your mouth wrong... http://www.sysopt.com/forum/wink.gif
Good luck
Dave

[mrdisco]

Hi all

I bought a pretested Celeron 366@550 and decided to try my hand at an overclocked setup. I got an MSI 6905 socket adapter and a GlobalWin VFP-32 fan/heatsink for it. I had to go with that size heatsink, since anything bigger would overlap into my DIMM slots. I'm using thermal paste to go between the CPU and heatsink (just enough to make them stick together but not slide around -- it makes little leaf-like patterns when I pull them apart).

My motherboard is an Asus P2B with the latest BIOS revision (1011). It's the kind with the jumpers for FSB and multiplier.

Anyway, I put it all together and turned it on. Win98 would not boot. Linux booted but it had trouble finishing CPU-intensive tasks like building a kernel.

I took it back apart and looked to see what I could do. I upped the voltage to 2.1 and tried it again. This time, everything booted, but Q3A would not last for more than a few minutes. I was also still getting weird errors on both OSes. So, I decided to bring everything back to factory spec and ask what you guys think...

The guy who sold me the chip seemed to be the reputable type, so I don't think I got a bum chip. I'm wondering if...

- I'm may not be doing the cooling right. It's my first time ever messing with thermal paste, so I may be doing it all wrong. I just got a temp probe for my motherboard, so next time I try to overclock I'll try to get some temperatures to report. I'll be sticking the probe to the underside of the heatsink near the top of the CPU pins.

- I may need a new motherboard. Is the Asus P2B a good overclocker? It's one of the earliest BX boards out there. Maybe I should switch it for a newer Abit board?

- There may be something else wron entirely. Any hints you guys can give would be appreciated.

Thanks

[mrdisco]

Thanks for the hints. I'll try them out in a bit.

My motherboard won't let me bring the voltage up past 2.1. Any time I bring it to 2.2, it doesn't post. Maybe there's a way to get around that? I guess it probably won't help if my cooling is still bad...

I'll try reapplying the thermal paste and let you guys know what happens.

[mrdisco]

OK, here's the update...

I reapplied the thermal paste, as suggested. Unfortunately my results only improved slightly. The system booted at 2.0, but just barely, i.e. it didn't last very long. At 2.1 it did pretty well, but not perfect.

I did, however, manage to get my motherboard to accept 2.2v, which is where I'm running now. After a couple hours of Q3A, a complete Linux kernel rebuild, and some SETI@home (still running) this thing is still rock solid. The final test will be whether it'll let me watch my Matrix DVD in its entirety using a software decoder. So far, it looks good, though. I'll let SETI@home run overnight and check on it in the morning.

I still wish I could get this chip to run at 2.0, but I have a feeling it won't happen with my current setup.

Oh yeah, I couldn't get the temp probe to stick, so I still don't have any temp readings. So far it's been running for 3 hours, though, and I don't smell anything, so I guess temp is OK. http://www.sysopt.com/forum/smile.gif The little crappy piece of tape that comes with it is kinda useless. Next time I open the case, I'll try shoving the probe under the heatsink mounting clip and try to get readings from there...