Re: [AMBER] md simulation error

From: David A Case <david.case.rutgers.edu>
Date: Sun, 29 Nov 2020 13:19:59 -0500

On Sat, Nov 28, 2020, Kehinde Idowu wrote:

>The job fail at the beginning and no output was produced. I ran it with
>CPU and it works.
>
>>Please, I am running an md simulation on GPU but the job is
>>terminated giving this error message 'Post job file processing error;
>>job3194139.sched01 on host gpu4002' .

If you got no output at all, then I suspect something must be wrong with
the scheduling system you are using. However, it is possible that you
and I have different ideas of what the phrase "no ouptut was produced"
means. Can you run the jobs outside of the scheduling system, and/or
verify that that you indeed have no mdout file?

>
>>I have checked my md_error file, this is the message I am
>>getting 'Note: The following floating-point exceptions are
>>signalling: IEEE_UNDERFLOW_FLAG IEEE_DENORMAL'.

It's not clear to me how you got a "md_error" file if you received no
output. Maybe it would help to give the exact commands you are using to
run the jobs.

It is also quite common to have to carry out minimization, and
somethings even the intial steps of equilibration, on a CPU. This can
be especially true if your initial structure is a bad one (with lots of
clashes) or if the initial density is far from the one you will finally
get (usually 1 gm/cc.)

It's very hard to tell which of these many possibilities (or none of
them) might apply to your situation.

....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Nov 29 2020 - 10:30:02 PST
Custom Search