kernel stability testing

cpu stability testing

(edited: damn I typed the wrong thread name.. it's "cpu stability testing" not "kernel..")

it seems to me that compiling kernel is a lot faster method of testing your cpu than running prime, since it's 99% going to crash on any bad o/c
not trying to talk anyone out of participating in a project that uses prime95, but for people that don't (and download the program just for cpu testing), this is a lot faster
just write a script that does make clean bzImage 3-4 times and run glxgears in the same time, and wait for segmentation faults or freezes

The only problem that I can see with that is that prime95 would have thread after thread of difficult mathematical equations, whereas a compile would have easy parts and hard parts. A compile sometimes has to wait for the harddisk to read/write files, or has downtimes during the compile, where it is only scanning your harddrive, or environment.

Prime95 would put a harder, more constant load on the CPU. Prime95 can get my CPU temps to go higher than any other program, and by that measurement, I'd say it's a better load tester.

have you tried it *combined with glxgears* ? it is a certain cpu eater
anyway, this is only a suggestion

One of the nice things about Prime95 is that it checks all data, so if you pass its test, you know your CPU is generating no errors. It seems to me that, during kernel compilation, small errors could occur without causing seg faults, crashes, or the like. Perhaps they would only show themselves at run time, or maybe not at all.

Also, while I'm by no means an expert in the field, I don't think glxgears stresses the CPU in the same manner as Prime95. I've tried both glxgears and Prime95, and did notice that my CPU temperatues were about 1.5 C higher with Prime95.

ok... btw just noticed that although glxgears & kernel go smoothly, mozilla crashes on a bad overclock...
so.. prime95 everybody