Re: [AMBER] tests with different # of threads

From: Vaibhav Dixit <>
Date: Thu, 4 Apr 2019 20:41:38 +0530

Dear Prof. D. Case,
Details of the failure are given below.
My another question is, while I troubleshoot these errors (with your help)
can I rely on this local-installation to generate tutorial/research-project
input files since am submitting jobs on our HPC-cluster.
Are there AmberTools specific tests which I should check for failures?
(I checked logs on HPC it has only 1 test with error and 2 failed
The constant-pH tutorial output is showing some differences. I will rerun
that job on the cluster.

For my local installation.
There total 36 mentions of dhfr in log file located in
amber18/logs//test_amber_parallel directory.
Earlier failures are due to using "np -4", I hope they will get test/fixed
with "np -2".
In addition to dhfr, other scripts are also failing but not complaining
about use of 4 processors.
*Please suggest how should I proceed, start fresh installation or fix error
1-by-1 (doesn't seem a practical option since there are >100 errors to

  ./Run.cytosine: Program error
Makefile:1120: recipe for target 'test.parallel.pmemd.basic' failed
make[2]: [test.parallel.pmemd.basic] Error 1 (ignored)
export TESTsander='/home/pharmacy/amber18/bin/pmemd.MPI'; cd dhfr &&
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
[unset]: write_line error; fd=-1 buf=:cmd=abort exitcode=1

*I get the following for mpif90 -show command:*
mpif90 -show
gfortran -Wl,-Bsymbolic-functions -Wl,-z,relro -I/usr/include/mpich
-I/usr/include/mpich -L/usr/lib/x86_64-linux-gnu -lmpichfort -lmpich

Thanks a lot for valuable and timely suggestions.

On Wed, Apr 3, 2019 at 6:00 PM David A Case <> wrote:

> On Wed, Apr 03, 2019, Vaibhav Dixit wrote:
> >Is it possible to quickly identify, run another make test for 2 thread
> jobs
> >without rerunning the 4 thread ones?
> If you set DO_PARALLEL='mpirun -np 2' you will just do the 2 thread
> tests; if you set DO_PARALLEL='mpirun -np 4', you will just do the 4
> thread runs.
> However, with the enormous number of failures you are getting, something
> is seriously wrong. Concentrate on just the first test failure you find
> (is it in something like dhfr?) See if you can find something that
> helps there. Re-running the whole test suite, to get hundreds of
> failures, doesn't seem like a good way to debug this problem.
> Did you ever report what you get from "mpif90 -show"?
> ...dac
> _______________________________________________
> AMBER mailing list

Dr. Vaibhav A. Dixit,
Assistant Professor,
Department of Pharmacy,
Birla Institute of Technology and Sciences Pilani (BITS-Pilani),
VidyaVihar Campus, street number 41, Pilani, Rajasthan 333031
Phone No. +91 1596 255652, Mob. No. +91-7709129400,
P Please consider the environment before printing this e-mail
AMBER mailing list
Received on Thu Apr 04 2019 - 08:30:03 PDT
Custom Search