[Bioclusters] Opteron Perl64 segfault issues

[Nathan O. Siemers]

Hello Joe,

Joseph Landman wrote:
> Hi Nathan:
> 
>   I took the liberty of modifying your Perl script to instrument it for
> process monitoring (memory size), and time stamp the iterations.  If you
> have a memory leak, you will notice the process getting monotonically
> bigger as a function of time.

Have not seen evidence of memory leakage - the process quickly jumps to 
a given amount of memory ( ~ 1.4 G on opteron/64, under 1G on IA32 ) and 
stays there (similar to what you report).

>   Have you run memory tests on this machine?  I highly recommend
> memtest86 (http://www.memtest86.com) though I do not know if it runs on
> an Opteron.  Let memtest grind over a weekend.  If it catches faults,
> you might have a possible culprit.

memtest86 does not compile on the opteron, giving an error during an as 
assembly.

Thanks for your help - I'll let poeple know what I find.  Hopefully it's 
just a ram chip, not an underlying instability with the kernel or perl.

nathan