glindahl at hpti.com
Thu Nov 9 19:13:20 PST 2000
> We have a UP2000 that is causing us endless problems. It locks up
> whenever a big memory job is run. It does this for any of two or three
> different jobs, often very quickly. We've pretty much completely
> replaced the memory itself and the problem persists. We suspect
> hardware, but which hardware, and how to be sure?
Try running a memory test. I've run UP2000s with 1/2 gbyte on all kinds of
long-running intensive calculations with no problems.
> Are there known SMP deadlock problems or memory bus problems
> that are fairly viewed as software?
I did manage to occasionally deadlock DS20Es; they print a message every
20-30 seconds on the console.
More information about the Beowulf