Zach Fine <czyz@u.washington.edu> wrote
>Is it likely that my problems with stability are due to the Cyrix
>processor? The reason I'm inclined to think that it might be the cpu
>is that there was someone on the Windowmaker list who's experienced
>the same wacky problems as I have (fatal signal 11, computer
>instability) and also runs this Cyrix cpu.

There have been some models of the Intel clones with design flaws but in
general such problems seem to be associated with marginal hardware, most
commonly RAM (from what I've seen on the NetBSD list and some of the linux
FAQs).

FWIW the two times I've seen hardware problems were:

1) a few years ago (eg NetBSD 0.8) when PC clone motherboards weren't often
stressed by 32bit protected-mode OS's, I had a motherboard that failed
during the boot unless both 486-internal & external caches were disabled.
It got exchanged.

2) more recently I ran disk-testing utility (adaptec threadmark 2) under
NT4 on a few identical PCs we received, some destined for NT some for
NetBSD. Apart from saying that a UDMA-capable NT driver from VIA gave 20%
greater disk throughput with significantly less CPU overhead, threadmark
revealed one AMD K6/200 which didn't like running at 225MHz in these
VIA-based boards. Normally the K6s are considered to handle overclocking
very well. These chips ran almost cold (maybe 45C) so it wasn't thermal
stress, and confirmed it was just this chip when I swapped CPUs between
several motherboards. This chip produced gcc signals when compiling a
kernel under NetBSD 1.3.1 - until I switched the clock back to 200MHz.
Curiously the first problem I saw was when trying to install NetBSD, the
sysinstall failed early on with msgs indicating read errors. Things worked
after I swapped to another floppy drive, but found the original floppy
drive worked perfectly as long as CPU was at 200MHz.

cheers,
Danny Thomas