OK guys, can any one give me pointers to my problem.

Whenever my Athlon 1.4GHz (Biostar M7VKD, VIA chipset, Radeon 7200) goes above about 44 deg C my PC hangs....

My idle temperature is down at 26deg C but when running full load for any length of time the temperature creeps up to 40's, that's when the trouble starts. If it stays up there for any length of time it freezes (it can stay at 40 - 41 for ages with no trouble but when hitting 42 - 45 deg C :dead: )

I have lots of case fans, I've used thermal compound, I've got a reasonable CPU fan - which means when I take the CPU off full-load the temperature goes back down very quickly.

I have changed my cable modem from USB to ethernet, I have changed my graphics from PCI Voodoo 3000 to the AGP ATI Radeon 7200.

Could it be a power supply problem, or SDRAM (PC133 CAS3 running on "normal" timing in the bios) or -god forbid - the CPU???

Any ideas?

Well it could be, try going into bios and seeing what the temp reads there, also take your fan off your CPU and look at it. If it is getting to hot, you may be able to start seeing scorch marks indicating that its the CPU or the HSF. Also check fan speed, a dusty, dirty, clogged fan will run slower but to the naked eye look fine. Make sure the HSF is running at or near its rated speed. How many case fans, are they blowing in the right direction ?? Front ones in and back (rear) ones out. If you changed a power supply, you may have one that blows the hot air directally on to the HSF giving you higher temps . Hope this help.

40C isn't that high is it? What were you doing to cause the lockups. It might just seem like it's the increase in temperature that causes it but you are also stressing other factors in your PC when it's active. My best suggestion is to turn down all the settings in the bios and play around with the PC for a few days to at least rule out a bios-setting problem.

40c shouldnt cause a freeze up..There is a problem else where in the system.

When did the problem start?

How can 40c cause a problem when my system runs at 53c without any prob. Just curious ??

I made a mistake boeing..I edited my post to reflect what I wanted to say..My system idles at 51 and it runs perfectly. Thats why I was saying that there is probably a problem other than the temperature.

WHEWWW, thats a relief, I just checked again and thats about where mine is. I appriciate the heads-up !!

davidjmorris,
How are you taking the temp. (oral or the other way? )?
ARE you SURE it's accurate? Some boards are a little less than perfect Since the system reliably dies at a set temp.
You can troubleshoot it, fairly easly (case open, running full bore, keep the monitoring program on top, at your own risk of course)
first power supply~ multimeter or do the fans slow down/change just before overheat?
you can use "canned air" to add a little extra cooling to the chipset/other components/memory one at a time just before you expect shutdown to occur (I use a rechargable can, called "Re-Air" for this, cause it's cheaper, no nasty fumes, andavoids the danger of cooling shock, that happens with fluro carbon store bought brands)
Does that help it stay up longer?
It still could be a CPU problem/ or chipset/or a bad trace/component on the MB
This all takes time , so how much you play with it depends on what it worth to you>
Good luck

I'm not saying that davidjmorris is a complete newbie or anything, but it would be better to insist that the computer be shut down before doing so

And shut your computer down before you take the CPU fan off !

Thanks Guys.

Yes, 45 deg C. is stupid!!!
Most of the things suggested have been tried at some stage.

I've had the HSF/Fan off and cleaned it up using IPA, plus replaced the "pink" compound that came with it (now using white HS compound from servisol). I'm fairly sure that it's a good contact (followed all the usual guidelines - i.e thin layer, only on the die etc.)

The temperature as recorded by MBM5 and the bios is the same. I have also fixed a temperature "label" on to the heatsink (this is one of those label with different dots on it - they change colour when the temperature goes though the specified range). 45 deg C was mid-range for the label. BTW I did this for just the reason mentioned by you guys - concerns over whether the MB temp sensor is accurate. Bearing in mind that the AMD does not have an internal sensor but relies on a probe set in the PGA cavity below the CPU (about 1 -> 2mm, I'd say), this will probably make the measured temps 10 - 15 deg C (guestimate) lower than the actual die temp.

Also, I have compared temps with a friend and notice that his idle/loaded temps are similar to mine - but his system doesn't freeze. I would try swapping components in his system but unfortunately he does not live close by

The problem started ever since I built the system - the memory I had in my previous system but it was a lowerly AMD K6-2 233 and so it would not have been working anywhere near it's capacity, the case/PSU were new, as was the CPU.

I have a rear extractor sucking from over the CPU, I have a "PCI slot" extractor and a harddrive cooler. I have ensured that the air from the holes in the front is un-obstructed (and that there are plenty of holes). I know I could put a fan here that would force air *into* the case but since I am not overclocking and I should not need to!!! There is also the PSU fan (obviously ).

I know that PSUs are "supposed" to be regulated to +/-5% on the +5V and +3V rail. I also know that some manufacturers cut corners on the design and have the 5V & 3V sharing the same secondary transformer winding and regulator. Hence a load on the 5V affects the 3V.

I have noticed using MBM5 that the voltages swing around a bit - maybe this is indicative of poor regulation or noise. Unfortunately I do not own an oscilloscope.

Time for some more data gathering and experimenting. I shall go back into the bios and turn the RAM to "slow" rather than normal. But in the end I may have to start replacing things.

Cheers.

I would read the AMD thermal white paper if I were you, it is better to have a fan blowing air IN than just sucking out. Better yet is air blowing in and sucking out. It is in their white paper. With hard drive coolers and other stuff you really aren't cooling the inside of your case. You are just creating turbulance. Also check out the list of approved power supplies. I have seen plenty of "weird" stuff with a computer with a bad power supply. We are not saying you are stupid, we just need alot of info and people are trying to give you advice from all sides from their personnal experiance.

Here is the page on AMD.com.

AMD configuration page

Thanks for the pointers guys.

wasn;t implying that you thought I was stupid just that I was agreeing with you that the temperature was stupid...

Anyway, your suggestions gave me food for thought - such that I believe I ahve solved the problem.

The problem may still have been a thermal issue but I think this may have been a red herring

Anyway, the low idle temperature (26 deg C) on my system were acheieved with the help of WPCRSET, which I used to manually set the "idle" bit in the AMD Athlon register (reg 52, I think). When I disable this vxd the system idles at about 40 deg C BUT the full load is still at only 44 - 45 deg C. AND the system does not hang!!!

Maybe this is to do with the temperature hysteresis - I don't know but whatever! my system doesn't seem to hang now.

Cheers guys. :grinthumb

More than glad to throw our $.2 in to the frey, if you have any more questions just ask. As you know, we like to answer (and ask too)

Your board uses the VIA KT133A chipset and you are running a
1.4 Athy...... you can't use HLT to "cool" the CPU - I had the same
problem.. AMD does not recommend it - at all. In fact it's the reason
why the ABIT KT7A range of boards never made it onto the AMD
"approved" list. AMD has since removed this function from their
BIOS upgrades for all these boards