LinuxQuestions.org - hardware diagnostic utilities

Hey all-

So skipping directly to the question,

are there any diagnostic utilities for diagnosing problems with a processor/motherboard? I'm looking for something like memtest86, but for other hardware components.

Here's my motivation:

I'm one of us hobbieists that picks up parts for cheap when I find them. I then try to throw together a number of working systems. So, I have a lot of spare parts just lying around. Some of them work, and some of them don't. Problem is, I don't know which ones work and which ones don't, so testing them can be a pain since I don't really have a control situation.

I'm working on a system right now with an ABIT VP6 mobo and a single PIII 800(133)MHz processor. I can get the system to boot normally into Slackware 9.0 and everything will be peachy. However, when I do something processor intensive I experience a system freeze after anywhere from a few seconds to a few minutes (we're talking HARD frezze here, no options but to hard reset the system). If I leave the system alone without doing something processor intensive, it will run for days with no problems.

I've been trouble shooting for a while now. Initially I thought that it might have been a problem with my graphics adapter because I would always experience this system freeze after installing the latest nVidia graphics driver (Video card: Diamond Viper V770, nVidia TNT2 chipset) when X displayed the nVidia spash screen on startup. However, after disabling the splash screen in my X86Config file the system no longer froze upon starting X.

Next, I thought that X might be having a problem with something, but I was able to reproduce the system freeze in a console (X was completely shutdown, I'm not that silly!)
The next thing on my list to check was the kernel. I built a very minimalistic 2.4.20 kernel for the box. The freezes still persisted however (I do find it to be somewhat odd though that compiling a new kernel didn't cause a system crash, doesn't the system use virtually all of the CPU time when compiling?). So, although it's difficult to *completely* rule out problems with a custom kernel, I decided that it was pretty unlikely at this point.

Now I'm down to the hardware components. After removing everything but one stick of ram (I've actually tried three different ones, two 128M PC100 sticks, and one 512M PC133 stick. I purchased these from retailers a long time ago, but since they've been sitting for a while I've become suspissious of their integrety), the processor, the video card, a hard drive and a cdrom drive (the latter two being needed for various booting options) the freezes still persist.

So, at this point I'm have a strong suspission that it's the memory, the processor, or the motherboard that's not functioning (have I missed anything?). So I've been trying to figure out which it is.

The first stop was Memtest86 (v3.1 booted from a CD-ROM). I've tried all three sticks of ram, and memtest86 actually crashed on all three! The crashes occured during test 6 in all cases. Memtest86 has, however, provided me with the first actual clue of my entire search. It tells me that there is an unhandeled interrupt generated, although it doesn't tell me where it was generated.

Knowing this, my first thought was if a heat problem. Perhaps the PIII was generating an interrupt that caused it to halt if it was starting to overheat. I had a heatsink and fan on the CPU, but I hadn't used any thermal compound. So I got a hold of some Arctic Silver 3, and applied that. Guess what? Still get the system crashes.

I find something suspisious about how memtest86 crashes in the same place each time (literally, the WallTime is consistant. i.e. everytime I run the test on the 512M stick it crashes at the same WallTime, and it crashes at the same WallTime for both of the 128M sticks, obviously the time is different between memory sizes, but it is consistant among memories of the same size). My gut instinct tells me that this more then likely means that the memory is okay, or at least there is a problem other then the memory.

So, I'm starting to run out of ideas. I don't happen to have any other processors or mobos that I can use to test the ones in question. Hence I'm looking for some kind of software diagnostic program like memtest86 that I can use to try and narrow down the problem. I realize that there's probably nothing for motherboards because there are so many of them that it would be nearly impossible to write a program to test them genarically. However, I do have hope that one might exist for processors, since there is a limited number of them. Anyone have any ideas either about diagnostic programs, things I can try, or issues I may have overlooked?

Thanks,
--Chad