heat causing my instability?

[djXternal]

Hey guys I have an 8rgm3i board with an athlon xp-m and ddr333 ram, for awhile I had the board running at 168Mhz FSB and the memory the same, I kno this is suppossed to be past the limits of both the memory and chipset, but this was stable for a long while.... Recently I have been getting alot of instability issues while it has been folding, the cpu is on a cheap but not stock hsf and the chipset has the stock epox heatsink on it with a small fan sitting on top... but the temp in my room has been getting warmer lately because of the weather here...

I was wondering if the increased heat could be causing my instability problems, or should I be looking at something else?

[Wolf2000me]

Yeah, it can. Though normally at your clockspeed it shouldn't be a problem. What are your temperatures under full load? Mobiles, as a general rule, seem to react to higher temperatures in a bad way. Mine gets unstable over 55°C at sig frequency.

[djXternal]

my cpu temp reaches a high of 32-33 normally, but the sink on the chipset can get really hot...

one thing on my temps, the case temp and cpu temp didnt seem right when i first got the board, the mobo reads that the cpu runs around 70 and the case 30, but seemed to me that something may have been mixed up.... but ive never been able to find any conclusive info on where the second sensor is located

[Wolf2000me]

Try USDM for monitoring. That should do the trick. Maybe your northbridge is indeed getting too hot. That can of course cause instability as well.
Though it is an nforce2 motherboard so it should be able to pull it off quite easily. What kind of instability are you looking at and when does it happen?

[djXternal]

I use MBM right now, i just had to switch the sensors it reads from... i get an error in F@H saying the simulation got an unexpected calculation or something of the sort, and the system reboots as weel, but it does not reboot everytime, only most of the time... it has been running stable for the past few hours at 158Mhz

[Aedan]

A temperature of 32C on a CPU would be very low for aircooling, and you're claiming that you have a heatsink that's not that amazing. I'd suggest that 32C is your case temperature, and the 70C may be your CPU!

[djXternal]

but that temp seems very extreme for an xp-m, doesnt it?

[djXternal]

looks like you were correct about the temps, I just did the pin mod to drop the vcore to 1.45v and dropped the fsb to 133Mhz and that severly dropped what i thought was the non-cpu temp, dropped from low 70s to low 60s, at what temp do the xp-m generally become unstable?

[robbie]

Originally Posted by djXternal

dropped from low 70s to low 60s, at what temp do the xp-m generally become unstable?

Ummm... I'm thinking somewere in the 70's!!!

[jnansley]

Think you found your problem

[T. N. Chua]

The easy way to find whether is the temperature causing the problem is to open the case and use a big table fan brow cool air into the system. This is the way I slove my system before.

[Aedan]

How did you attach the heatsink to the CPU? Using the thermal pad that came with the heatsink?

[dsio]

The XP-Ms are designed to handle very high temperatures compared to regular XPs due to their notebook usage, but that is an overclocked computer.

I get the feeling you are over volting this chip. What is your current VCore, and are you sure you need it that high?

[djXternal]

yeas i dropped the vcore down using the pin mod, and USDM reads it at 1.5v now, the pnimod was suppossed to drop it to 1.45, so that helped stability alot....

I also think MBM was reading the sensors wrong, after installing USDM the cpu reading came out to be mid 40s to low 50s

and yes I am using the thermal pad/paste that came already on the HSF

[dsio]

Hmm. 40-50C is fine. And the voltage is ok.

Are you able to isolate some hardware by swapping in spares? See if different ram improves things, then see a different mobo makes it stable.

[djXternal]

actually stability is great now, i think the vcore was runnign high causing unstable temps, its been running all morning at 173x13.0, higher than ive ever been able to get it beofre, and i still dont think its to its limit

[Wolf2000me]

Hey,

It probably isn't relatively near its limit as it is an xp-m. I haven't seen any xp-m yet that didn't do 2.4ghz at 1.8v vcore. Though with that kind of temperatures with a bit of higher voltage you're not going to get very far, as you noticed yourself. I suggest getting some better aircooling. The newer Zalman solutions don't produce much noise and perform very well. I'd say from the 7000cu and up are good ones. I myself have been using the Thermalright SP97 for some time and it got me to 1.875v without overheating problems though i suspect that one is rather hard to come by. You may have better luck with the SI97 which performs almost equally.
As I already said, the higher you go in frequency the lower your max stable temperature you will get. Though i really think with a bit of better cooling you'll get much further.

[djXternal]

the pc will actually be watercooled coming soon, and the limit im talking about is actually the chipset (or the ram im not sure which it is), ive heard from people that it cant get past 166Mhz, but seems 174Mhz is about its limit, im going to try raising the multi a little more later today now that I have temps better under control...

Running Prime95 on it right now at 174x13.0 (2262Mhz)

[dsio]

Before spending $40 on a new cooler, take your existing cooler off, scrape all of the crap (thermal pad, thermal goop, jelly jam) off the bottom of it with a sharp plastic instrument to protect the surface.

spend $3 on a tube of Artic Silver 5 and use it.

Put a dollop onto the coolers base, and rub it in with a plastic glove over your hand, or just some plastic something, so as not to contaminate it with oils and skin. Keep rubbing more and more into it until you have a nice even surface, but dont be sloppy. At the end of rubbing it in there should be no more than a tenth of 1 mm thickness. It should be a haze only.

Next, rub some AS5 into the CPU die. Get it to all the edges, and a nice covering that is even, flat and consistent, but not thick.

It should be a haze on the surface. So if there was a picture on the surface, you could still see it through the AS5. Mate the surfaces tightly, and make sure the clamp holding the HSF down is tight.

[djXternal]

the watercooling should be happening in the next few weeks, and ill need some artic silver for that as weel, so i might buy that early