OT: System stable at 95MHz, unstable at 100MHz
Last month I upgraded my system with a bunch o' new parts. Here's what I
got:
ASUS P5A Super-socket-7 motherboard
AMD K6-2 400MHz processor
128MB PC-100 RAM
IBM 4.5GB SCSI-2 drive (2.5GB for Linux, 2GB for Win)
I moved over a lot of my old parts, like:
Diamond Stealth 64 (S3 Trio64 chip) video card
NCR53c825 SCSI controller
425MB IDE drive as hda (all for Win98) (hey, my wife wants it)
4x CDROM as hdc
Anyway, I keep getting random sig11's and other problems when I push the
system hard. Generally, if I repeatedly compile a kernel, sooner or later
I get hosed. Either a sig11 (seg fault), sig4 (illegal instruction), or
corruption in the RAM cache (one-character corruption, e.g. DEVICE
becoming DtVICE, that goes away when I reboot), and so forth.
First, I got a better CPU fan, and used thermal grease, but it didn't
help. I slowed the SCSI bus from 10MHz to 5MHz - still no good. Then, I
backed off on the CPU speed, to 350MHz (3.5x100MHz). Still failed. Then I
dropped the bus speed to 95MHz - finally, things seemed to work.
Okay, I thought, let's try replacing parts 'til we find the problem. The
place I bought the stuff has been very nice about things. I swapped out
the memory for a different DIMM, and I couldn't even finish one compile.
Hmmm. I swapped the motherboard - better, but still a problem. Last night
I swapped the CPU - still not stable at 100MHz. At this point, the only
part that I haven't swapped out is the case.
For a while it looked like the CPU temperature was correlated with the
problem, but I'm not so sure now. I've also tried setting the bus to
100MHz and slowing down the memory timings in the BIOS. It hasn't had any
effect. At 100MHz, I can't compile more than 1 kernel in a row, and at
95MHz I had a script run overnight - 71 kernels compiled with no problems.
At this point, I don't have any good ideas. I swear, I am *not*
overclocking - I'm running everything at the rated speeds... or below,
now. The motherboard has voltage and temperature sensors and both
motherboards that I've tried have shown acceptable readings.
I only have three theories left. One, I'm getting radio interference at
100MHz from somewhere. I'm going to try overclocking to 105MHz tonight,
just for the sheer heck of it. (It'd be something if that worked, wouldn't
it? :-/ ) Two, it's the case && || power supply. If all else fails, I'll
try swapping *that* Wednesday. Three, the place I bought from got a bad
batch of one part or another, and I just keep getting bad
memory/motherboards/CPUs.
Does anyone have any other suggestions, ideas, or wild flights of fancy?
When it works, it's a *great* system - the kernel compiles in 3 minutes.
But if I can't trust it, it's no good to me. And I paid for 100MHz, dang
it - I don't want to run it at 95.
Sincerely,
Ray Ingles (248) 377-7735 ray.ingles@fanucrobotics.com
"Transported to a surreal landscape, a young girl kills the first woman
she meets and then teams up with three complete strangers to kill again."
- TV listing for the Wizard of Oz in the Marin Independent Journal
Reply to: