From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Thu Jul 27 2017 - 11:44:13 CDT
Hi:
As I have already in part posted in the last two days, I am trying to run a
FEP simulation on a NextScale Knights Landing cluster. Totally
unsuccessfully in one-node attempts, while they run on my linux-intel 4CPUs
desktop..
Now I have run on the same cluster, one node, same input type, classical MD*,
*which runs without problems.
The job file used for both classical MD and FEP after commanding "module
load env-knl":
*#########################*#!/bin/bash
#PBS -l select=1:ncpus=64:mpiprocs=126:mem=86GB:mcdram=cache:numa=quadrant
#PBS -l walltime=00:05:00
#PBS -o npt-06.out
#PBS -e npt-06.err
#PBS -A my account
# go to submission directory
cd $PBS_O_WORKDIR
# load namd
module load profile/knl
module load autoload namd/2.12_knl
module help namd/2.12_knl
#launch NAMD over 4*64=256 cores
# namd2 +p 256 npt-05.conf > npt-05.log
# mpirun -perhost 1 -n 1 namd2 +ppn 126 npt-06.conf +pemap 4-66+68 + commap
67 > npt-06.log
mpirun -perhost 1 -n 1 namd2 +ppn 256 npt-07.conf +pemap 0-63+64+128+192 >
npt-07.log
###########################
The first line of command reports error "unknown +p" while the other two
(for one node) are OK for MD (adjusting the number of MPI processes in the
first #PBS line).
In contrast, with FEP all lines of command fail, the first one like for MD,
the other two with error:
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= PID 58639 RUNNING AT r065c01s03-hfi.marconi.cineca.it
= EXIT CODE: 11
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= PID 58639 RUNNING AT r065c01s03-hfi.marconi.cineca.it
= EXIT CODE: 11
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
============================
Is any special flag lacking in our KNL compilation that prevents knl
working? I could ask, and report here, (if useful) how namd12 has been
compiled
Thanks for advice
francesco pietra
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:28 CST