Wrong compilation artifacts on ES40s?

From: Sanjeev B.S. <sanjeev_at_mbu.iisc.ernet.in>
Date: Sun 24 Jun 2001 22:43:11 +0400 (RET)

Hello,
        I am trying to make ES40s work for amber6. I did benchmarking
tests aganist a variety of platforms (linux, linux cluster, IBM SP3, etc.)
and found that after running the benchmark test, the energy values were
identical in all. Except on ES40s! I am attaching the MACHINE file and the
mdout difference files. The first one in difference is the correct(?) one
and the second one is from ES40s. (Incidentally even the number of
processors change the output!)

        Can anyone suggest how to get over this problem please? I feel the
numbers are junk. But cannot pinpoint anything.

Thanks in advance,
-Sanjeev

PS:
Command given for sander:
$ /usr/bin/dmpirun -np 2 ../../exe/sander -O -i mdin -c md12.x -o mdout

Command given for difference: (energies are written at steps 1,25,75,100
and other details were removed from mdout except energies when 'diff' is
run)
$ diff mdout.std mdout.es > std_es40.diff

Attachment Converted: "c:\eudora\attach\MACHINE.bss"

Attachment Converted: "c:\eudora\attach\std_es40.diff"
Received on Sun Jun 24 2001 - 11:43:11 PDT
Custom Search