Hi folks, I'm having a very unusual problem with my "family" PC downstairs. Its an Athlon 3000XP, Asus A7V8X board, NV 6600GT, 1GB DDR2700, all fairly standard and a few years old. No O/C or anything on this system.

When I turn it on and go into the BIOS, its convinced the processor is running at 60C. By the time an OS has loaded, its upto 70C. Under a load the BIOS's thermal cutoff kicks in (Set to 100C - max setting, in bios) and the machine turns off.

Your probably thinking naff install, dodgy thermal paste or heatsink or something, but its none of these. Using a laser temperature gun, I've established the processor at idle is running at no more than 35-40C when browsing the web, and ~50C under a full load. My guess is its a dodgy sensor on the motherboard?

Is there any way to override the thermal shutdown in the bios? Or to force a different reading so it never cuts in? There is no option in the bios to disable it, and the max setting is 100C, and it still happens when set to this.

Its not really worth throwing money at a machine this old, but I'd like to keep it running.

The only way is to buy a new mobo (afaik), however i'd like to know how you used a laser thingy to check the CPU temp?

Core Temp is INSIDE the CPU core - the heatsink temp being 50 degrees colder than the CPU temp would imply a dodgy thermal paste or something...

Either that or a fubar'ed sensor - which needs a new mobo to fix.

If the board uses a thermal probe rather than reading the core temp for the CPU, you could try bending it flat so it doesn't touch the CPU. That should lower the reported temps a bit.

Quote from Jakg :

however i'd like to know how you used a laser thingy to check the CPU temp?

Watch this video and you'll see.

Er, not quite like in the video Smoking the processor isn't really my idea of fun. I take an average reading from several spots on the base of the heatsink, and if there's no heatspots its safe to assume the CPU core temp can't be much higher? Assuming good contact between core and heatsink, of course. Added to this, when you turn the computer on from stone cold, the BIOS already thinks its at 60C 10 secs after its turned on. At this point everything inside is still room temp.

I'm confident its not a dodgy install or bad thermal paste - I re-done it two weeks ago using the same method I have for years, and using Arctic Silver IIRC. Its not a problem I've ever had with any other systems I've built.

Thanks for the suggestion Bob, I'll take the processor out again today and see if I can move the sensor back a bit. Speedfan tells me its reading the motherboard sensor and not a CPU-Core one, unlike on my A64 system, so I'm confident its a motherboard sensor.

Try this program if you can get to it before it shuts down, and see what your temps are. It reads the temperature directly from the sensors built into the processor(s) itself. For me, it shows temps almost 10 degrees C colder than AI Booster, PC Probe AND the BIOS (I think...). It's a pretty neat little .exe utility.

I tried running the program, but it errored out, saying the sensor didn't exist in the processor. Which makes sense, Speedfan doesn't find one in the CPU either. Whereas Speedfan & the app you attached both read the temp off my A64 system. Nifty little app though

I flashed the bios to the newest one, its not helped. Don't have enough thermal paste left to re-do the heatsink which rules out playing with the sensor, so for now I've just underclocked it by dropping the FSB (2x)33Mhz. Its been running a couple of hours with no problem so it'll do for a short term fix.

So... I'm bumping this.... I have overheating problems with my CPU... Again....
So, it started when I got this PC from my dad, and nobody knows how to fix it.
The CPU is AMD Athlon X2 5000+ running at ~2,6GHz. It's on GIGABYTE GA-M61PME-S2.
The case is opened, and I installed 2 fans so it gets even more air, but the CPU just doesn't give a damn. And, yes, there is enough cooling paste on the CPU, heatsink is touching it correctly, it's stock, but works on my bro's CLOSED PC, it has a 8cm fan running on 3kRPM.

The problem starts in gaming, as you may assume. I can't even think of playing CoD4MW, and even Icy Tower is heating it enough to overheat and turn off. I'm playing LFS most of the time now, but same thing as Icy Tower and CoD4. After a while CPU temp. gets to 80°C, then to 100°C and dies on around 115°C.
I also noticed that my GPU (HD4650) is very hot when CPU is overheating which wasn't happening before. It gets up to ~65°C, but doesn't overheats, it's max temp is also around 115°C. Same goes with RAM, they get really hot, which wasn't the case before either.

The PC case is a bit small and all that is close one to another, and that could also be the problem, but is there any solution except throwing CPU/cooler through a window and buying a new one?

EDIT: Also, I noticed that AUX/PSU temp. in PC Wizard 2010 gets a bit wild. It's around 40 when idle, but as soon as I turn anything (even Chrome) on it gets to around 90°C.

If the CPU temp really is that high then the heatsink is not touching it correctly, or maybe there is too much thermal paste. Does the heatsink heat up? Can you touch it and see? Careful not to burn yourself. Maybe try underclocking the CPU, or undervolt it. Actually, you could try setting the BIOS to default (factory) settings.

Quote from shiny_red_cobra :

If the CPU temp really is that high then the heatsink is not touching it correctly, or maybe there is too much thermal paste. Does the heatsink heat up? Can you touch it and see? Careful not to burn yourself. Maybe try underclocking the CPU, or undervolt it. Actually, you could try setting the BIOS to default (factory) settings.

BIOS already is on default setting. It is touching it correctly, and it gets so warm that I can't my finger on it for longer than 2-3 secs. There isn't too much paste, there's a thin layer just so you can't see the numbers on it. Once I put all the paste I got in a small package and the result was much better than with thin layer, but nvm... Also, CPU already is underclocked as I'm using the "Minimal Power Management". It's working on 1004,57MHz when idle and the voltage is 1,104V and under load it's on ~2600MHz and on 1,3xxV.

So, nobody else?

bigger case? different HSF?

i've always said that amd's run hot. i have a delta 8000rpm fan on my athlon xp2000+. </intel fanboy>

No it sounds like the motherboard might be providing too much power for some reason, maybe the voltage regulators are failing or something. I would try that CPU in a different motherboard and see if it's the same problem.

115°C is DAMN hot for any CPU. I'm afraid that I'll just repeat what cobra said, but your heatsink is most likely improperly seated. The best thing to do now would be removing the heatsink and reattaching it again. While you're at it, you should give the heatsink a good clean and remove all the thermal grease left on the CPU and the heatsink. Simple adding of a new grease is often a bad way to improve cooling. When you apply the grease, make sure you apply it both on the CPU and the heatsink, not just on the CPU. Also make sure there is a uniform evenly distributed layer of it, possibly as smooth as you can get it.

Sounds like a voltage overload to me. Would explain more heat to everything even with all that cooling. have you tried checking voltages at all (externally, not using software etc) when its running?

If nothing works, buy an air conditioner for your PC.

So, me is answering.

Cobra @ The voltage is 1,312V, which should be normal... lol If I get some (lots) of money I'll make sure to do that.

MadCatX @ Well, I know it's too damn hot. I already reattached the HS hundreds of times, and cleaned both CPU and HS, reapplied the paste on both, but nothing, and HS is seated good, 'cos I can feel it slide on paste on CPU while securing it in place. And yes, it is distributed evenly.

robt @ Probably the only thing possible now. I'm momentaly on "Minimal Power Management" which lowers the voltage to 1,104V and the temps don't go over 35°C, but the speed is also lowered to 1004,57 MHz. And if I run anything, even Icy Tower (screenie of it few posts up) the voltage gets back to 1,312, the speed gets back to 2600 and the temps reach the skies.

And how could I lower the voltage only a bit so I could keep the max speed so I can check would it overheat then?

So, I got sick of it today, and I removed the HS. I was like WTF when I saw it... The cooling paste "disappeared". Instead on the CPU it was all over the HS, so I removed all that wasn't on the middle of HS and put back CPU, the temps seem OK for now, and I'll let you know about final result ASAP.

Sorry for triplepost, but.... **** YEAH! It's working!

LOL I told you it wasn't making contact properly!

Quote from shiny_red_cobra :

LOL I told you it wasn't making contact properly!

It was making a proper contact, but the paste somehow went from CPU to HS, and I have no idea how... When I put it back on the CPU all was good.

Good thing don't last for long time, the overheating strikes again!

Quote from matijapkc :

I can feel it slide on paste on CPU while securing it in place. And yes, it is distributed to the end of the cpu and HS contact point..

there's your problem. overly cautious. installing a HS should be done in one smooth move to reduce the chance of over spreading the thermal paste.

Quote from dadge :

there's your problem. overly cautious. installing a HS should be done in one smooth move to reduce the chance of over spreading the thermal paste.

It worked today. I was playing for 2 hours... No overheating. I turned PC off. Later the problem was back again...

change the heatsync.