Motherboard Problem?

[MrCoffee]

OK we have a h8DCE 2 socket opty board and the problem is that one of the sockets overheats.

There's nothing wrong with the cooler (active) as this has been swapped out with the other socket.

There's nothing wrong with the CPU as this has been swapped out with the other socket.

You boot the board and after 2mins or so you get the temperature alarm. If you look in BIOS as soon as you boot you can see CPU1 is at 38C and CPU2 starts out at about that but the temperature rises steadily untill the alarm goes off and we shut down the board.

The voltages reported in "system health check" are stable and equal for both sockets.

Any ideas people?

I should say the PSU cooked off prior and was replaced so *that* may have caused some damage... but I don't understand how the temperature of the CPU can rise continually at a constant voltage and with exactly the same cooling as the CPU next to it which is fine!!??

now your first thought maybe that there isn't a good contact with the heatsink.. There definately is, tried reseating it, replaceing it with the one from the other socket etc. The heatsink is solid as a rock on the chip, screwed in nice and tight.

 

[MrCoffee]

*Any* comments welcome, this computer has been a tale of woe for months. We don't want to RMA the board again, these things are a devil to get hold of in the UK.

 

[OatesyX]

any chance that the cpu is at fault, maybe swap them ino the other sockets, then you can rule out some stuff

 

[Tox1cThreat]

^ he did that.. try to read the whole post before you reply.


If its not the cooler or the CPU, it could be the northbridge, or a bad sensor/reader. try using a different program to monitor the temp (if you can, get one that logs the temps, so if the computer shuts down, you will have the latest temps.)

or if there is an external temperature sensor that you can apply to the motherboard (northbridge). Not quite sure how you would affix it, but stilll it might help. Also, search if anyone else has had those problems on your specific board.


Does the board tell you what sensor is reporting the problem?

 

[MrCoffee]

Not using a program, the computer doesn't have an OS installed, its in the process of being built.
We've been getting the temperatures from the BIOs, I presume any program would only be getting the same reading as the bios?
Anyway, even from cold it doesn't take the CPU in the second socket more than 2 minutes to hit 70C (at which point we switch off, not wanting to damage the CPU) so there isn't much scope for installing or running any programs.

I believe the sensor reporting is the one actually built into the CPU but i can check that out. In any case the BIOs reports the sensor name as CPU2.

It seems that that particular socket is causing the CPU to overheat, I don't understand how thats possible but its the only option thats left.

I have googled around the H8DCE but there is practically nothing out there on it, its a white box supermicro board and somewhat uncommon.

 

[markkleb]

This may sound dumb but did you touch the 2 heatsinks to see if you feel a difference?

If you have switched the cpu,s and the coolers and you still have a prob sounds like the temp probe is bad.

 

[MrCoffee]

I guess if its reading the temperature from the socket rather then the chip then the sensor may be bad.
come to think of it the first time we booted an went into bios the CPU2 sensor was reporting something silly like 25C.

Hmm, has anyone had any experiance of faulty temperature sensors?

I'm not convinced that the sensor is faulty though because as you say I did feel the heat sinks after a few minutes running and CPU1 was fairly cool and CPU2 was hot.

the heatsinks we're using are vapochill things, liquid filled with 3 verticle tubes in a fin array with a fan on the side.