Jeff Ross wrote: ... > As a followup, here's what I have done to try to alleviate this: > > I bought and installed the plastic air shroud using the passive > heatsinks that came with the motherboard. System still overheats and > shuts down within a couple of minutes. > > I bought 2 AMD brand active heatsinks, specific to this processor, and > installed them. That meant I had to ditch the plastic air shroud, but > the motherboard manual says that active heatsinks are suggested for 2U > chassis and the air shroud was only $10. I also used new heat sink > compound when I put everything together. > > System seems to run okay at idle. but make it work a little--like > compiling a kernel or tar-ing up a big file and the temp indicator comes > on and sysctl reports > temps (on both the kate and lm sensors) finally exceeding 100 degrees C > on one processor, with the other is not that far behind at over 80 deg C. > > At that point the system shuts down. > > I'm at a loss as what to try next. If I've read the AMD specs correctly > these processors should not exceed 71 deg C but I see temps near that at > inear dle. > > Did I just get a lemon motherboard/CPU combo? I still have a couple of > days on my 30 day exchange if this is the case.
something is wrong. Any good computer, surely any server, should be able to run at 100% proc load indefinitely, regardless of the OS. Some laptops will have issues with this test, maybe some junky home-oriented machines might, but I don't think I've actually seen a desktop with such an issue (lots of people run distributed apps :). Servers should be able to do 100% 24x7xlong-freaking-time, and if it fails, it needs to be fixed. Keep in mind, the tiniest piece of debris in between the CPU and heat sink can and will cause serious heat flow problems. And some heat sinks in many machines are very hard to verify that they are flat on the CPU (or that they touch the CPU at all). Nick.
