Multiprocessor temperature differences?

Status
Not open for further replies.

Mictlantecuhtli

Posts: 4,049   +13
I find this quite odd: I'm using a dual-P3 machine, both CPUs running at same speed, same voltage and have same kind of heatsinks.

However, one gets considerably hotter than the other, even though both heatsinks are properly in place and have sufficient amount of thermal paste. The following was after running cpuburn on both CPUs:

hot_cpus.png


These temperatures are achieved when the setup looks like this:

2cpu_case.jpg


It also looks like either the thermal paste or something else is getting a bit hot:

cpus.jpg


heatsinks.jpg


Nevertheless, 25C difference in this setup seems quite odd. Do any of you have any explanations?
 
Which one was the hot one, top or bottom? No fans? Does that video card produce a lot of heat? Both of those temps are definatly way to high. Can you switch the cpus around and see if the same one overheats? Is all the stuff you are using brand new?
 
The one on the left. Videocard doesn't produce much heat. I've switched CPUs and heatsinks, still the one on the left in the picture heats up more. I've been using this setup for about 1.5 years now.

Of course the temperatures were high, I was testing if the cooling is adequate for running the computer without a fan (it apparently is, as the system didn't crash or slow down).
 
Try running a single CPU in each socket and see if the temps change. If you've already tried different heatsink/fan combinations, then I would suspect that either the socket on the left is either suppying a higher voltage to the left chip socket (the voltage reported could be wrong), and/or one of the sockets is getting better cooling than the other. Your picture does seem to show that one heasink has its fins aligned with the airflow, while the other is at right angles to the airflow. Also, one of the heatsinks seems to be in a position where it is blowing hot air onto the other's socket. Try running with the case off and checking if the temp difference is still the same.
 
Well, I've swapped CPUs in their sockets, and during this test heatsinks were not getting airflow at all. The fan on the PSU is pointing outwards, rotating at about 1400 rpm, and it wasn't affecting the temperatures when I swapped the CPUs. The side was open like the picture shows (the computer is on in that picture showing the mainboard). I'll try uniprocessor setup soon.
 
One thing I can see is that when the case is sitting upright, the chip in the lower socket has its heasink fins aligned vertically, whereas the upper chip has its heatsink fins aligned horizontally. That would have a significant impact in cooling, because the vertical fins would get better cooling through convection (which appears to send hot air to the upper socket making matters worse for the second chip).
 
Hehehehe, this gets even funnier!

When using only one CPU, temperatures rise even more than in dual-CPU mode.

cpuburn.jpg


If you can't take the heat, get out of the kitchen..

Well another CPU still runs cooler than the other, it goes only to about 103C.

I couldn't see any differences in heatspreaders either, both are firmly in place.
 
No, they are not, I just tested that the CPUs shut down automatically when the core temperature exceeds 135C. Looks like I'm safe for now.
 
Status
Not open for further replies.
Back