I have recently setup a cluster of 5 dual processor PII 450 machines and
am now having them throughly tested. I have already identified one
processor as being "bad", in that while it is in the system, and is under
load, it will within a very short time cause the system to spontaneously
reboot. Now after some more testing we apparently have a second bad CPU.
The kernel is 2.0.31 SMP which runs quite reliably on identical systems,
reboots follow "bad" processor when installed in other, previously working,
machines. I'll try a 2.1 kernel later today and see if that helps.
Given that in only 10 CPU's, two are almost certainly bad, that's a pretty
good failure rate for CPU's that should have been tested already. All but
one of the CPU's we ordered came in one big CPU box w/o official Intel
"papers", so the possibility that one or more of them could have been
relabeled has occured to me (not that official Intel papers are any gaurantee).
What I need to know is can the 450's be reliably identified (if I recall
correctly they can), and where would be a good place to get the cpuid
program and instructions on what values a PII 450 should return?
- Steve
.------------------------------------------------. # * # # # # # #
| Steve Baker | Barely Working | # ## # # # # #
| [EMAIL PROTECTED] | System Administrator | # # # # # # # #
| Red-Hat Rulz! | Will work for hardware | # # # ## # # # #
`-- SYS-ADMIN FOR HIRE, HAVE UNIX, WILL TRAVEL --' #### # # # ## # #
