I forgot about replying to this, or anything else. Sleep deprivation does wonders.
BenWang: I've heard talk of electromigration, but never heard of a case of it. My 300A Celeron is in the hands of
Starfalcon, and though I haven't seen any posting from him in a fair amount of time (and it appears that even though he's a forum moderator he hasn't posted anything since December 2006), I wouldn't be surprised if it was still working fine at that speed. I built it in the spring/summer of 1998, ran it at 450 MHz every day of its life, and it never indicated a problem. I'm not sure how long it's been since
Starfalcon has had it, but maybe since 2004 or so.
Anyway, I'm not overvolting much at all, and while the OC is over 40%, it's still well within tolerances for the processors since they were selling faster ones than this one is running at the time I bought mine. I have to clean out my system again but I lost my can of compressed air somewhere and keep forgetting to pick up a new one. After I do that, and re-apply the thermal goop, I expect it to behave better.
notfred: No wonder I couldn't find tftp.
I have no idea how to find out about available software for Linux distros, because most of that online networking stuff is newer than what existed when I first played with Linux.
I've not read about it, but I never saw anything that looked like it would help in the output of dmesg. Of course, I have yet to run it when the folding client has crashed. The odd thing, though, is that the entire virtual machine just froze up the last 2 times, as if I'd pushed the suspend button on the console. Just pushing the restart button after lowering the clock speed got everything working fine again. I've been running the Linux SMP client for I believe today makes 6 months, and never had a problem until recently. I mean, yeah, I've OC'd too far before, but never had these recent minor OC problems. And, every time in the past the client has crashed but not the Linux VM, and when it crashed the client eventually killed the WU and downloaded a new one and started it from 0. This time things are worse in the fact that it's failing more easily, but better in that I only lose a bit of time but keep the progress in the WU... and some of that loss is made up for by the dual Windows clients taking up the free CPU cycles.
For now, I'm putting the problems down to the machine needing a "refresh" to clear out dust and get better heatsink contact.