AMBER Archive (2007)

Subject: Re: AMBER: Test fails in Parallel version Amber9

From: gong wb (bnmrcamber_at_gmail.com)
Date: Mon Oct 22 2007 - 01:52:48 CDT


Hi Ross,
    Thanks a lot for your reply. Following your suggestions, We have
made some testes. The testes are only focus on rem_gb_4rep and
rem_hybird, we used 4, 8, 12 and 16 process to run the test in
rem_gb_4rep and 2, 4, 6, 8, 10, 12, 14, 16 process to run the test in
rem_hybrid. To our supprise, the results showed some kind of
fluctuation. In rem_gb_4rep, only the test using 4 processes gave
failure, others were all pass. And in rem_hybird, testes using 2, 10,
12 and 14 processes gave failure and the others were pass. Besides,
the failure givend by 12 processes and 14 processes is indentical.
    Thus, we want to know is it normal, or there is something wrong in
rem or our Lam_mpi?
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu