This is certainly bad, but an even worse problem is that the person who reported it didn't "stress test" his system, as I bet many other quiet enthusiasts don't. If it cannot survive 100% CPU usage for long, that's a hardware issue, not software!
Not to mention, thermal throttling kicks in around 80C, so it was probably not even running at 2ghz anymore...
I have always been a proponent of worst-case testing by using programs from the CPUburn suite: BURNP6.EXE
for Pentium4 and BURNK7.EXE
for Athlon. This program pushes CPU power consumption beyond even manufacturer's "expected max" values, producing higher temperatures than any "normal" software can attain. In my experience, BURNP6 out-heats Prime95, Hot CPU Tester, and other programs by a long
My P4/2.26B @ 2.5ghz quiet system (system pic
; more pics throughout this thread
) attains a max of 66.0C CPU, no errors, no thermal throttling, after running BURNP6.EXE for a long time. At 2.55ghz it would actually encounter errors past 65C -- but every other test would pass with flying colors, even MemTest86! Now that I have a known rock-stable config, I don't even get to 60C with most normal 100%-CPU apps--the worst was 64C after a long UT2003 game with 20+ windows in the background. Idle temp is 37C.
My point is, why run at X GHz if you cannot actually use
100% of it, for good or evil? Rigorous stability testing in worst-case scenarios is mandatory
for ultra-quiet system builders--it's already mandatory for overclockers, anyway.