considering that the G3 ran cooler than most CPUs I have seen, and the other fact that I think the G3 chip itself has a heatspreader on it (not quite sure on that) which the pad/stuff rests on, it may be disapiting a fair bit of heat through that. Though full contact may be preferred. I will take a full look at my Lombard tomorrow and tell you what you should see. If you don't see what I describe, it could possibly be an issue.
Time-based problems are usually one of two things:
RAM or Heat
Over time he will eventually use some or all of his RAM. If there is a Memory leak, it could possibly fill all the ram up to the point where it would hit a bad area and freeze. It's possible and has happened to a customer's laptop that after working with it (light duties) he would bluescreen. it was usually 20-25 minutes after he started it up. Running a test found bad RAM in the higher memory banks causing issues, and a quick replacement of the memory module fixed it. Granted at the time he was running Windows 2k Pro and 512MB RAM, so with 2k and light duties it is kind of hard to hit the 512MB RAM unless you either opened a lot apps, or had a memory leak in one of your programs (esentially eating up all the ram till it hit that bad area)
Heat is the second issue. I have (believe it or not) run into a similar issue with heat. I ran a Celeron 366Mhz Socket-370 chip on an eMachines board *WITHOUT* the heatsink. If you want to see what it looked like, hit AppleFritter for the XP on the slowest machine thread. After doing the install (which it ran fine with the entire chip exposed, no air going across the surface) it reached the windows login. Something it did when it reached there pushed the chip way above what it could do even without issues before, because the chip overheated so quick it would freeze without BSoD'ing which was rather remarkable for the chip. I still believe I have that chip somewhere in my stash.
Like with every CPU, some have a higher heat tolerance than others, the same way some chips overclock better. Some just do better at withstanding extreme temps/overvoltage. Which could partially be the heat issue. If the chip in his laptop is not coming in contact with anything, it *could* theorhetically run for a while even at max with all the heat it puts out, *If* the surface of the chip dissipates the heat at enough rate without help to keep the chip from having errors. However much it would dissipate, if the chip heats up quicker (or very slowly over what the CPU can correct itself through error correction) it could overheat and freeze.
I seem to recall seeing that in a CPU with a bad fan. with the fan bad it would turn very slowly, like maybe 100rpm (it was a 700-1500RPM fan) and the system would run, but eventually freeze, but if the fan stopped it would freeze within 2-4 seconds. Letting a CPU heat up slowly seems to do better than having a blast all at one second. I suppose it can be true of anything, but it also goes along the lines of hardening a CPU during overclocking (Burn-in?)
Hard to explain it, but if you had seen the temp rise I saw with a thermometer, I saw the CPU hit 90º C and freeze, but if I slowly let it build up, I saw it go to almost 110ºC before freezing.
(not time related, but task related)
