From: Victor Kwan (vkwan8_at_uwo.ca)
Date: Tue Nov 19 2019 - 14:57:56 CST
Use the multicore cuda build if you are running it on one node, it is as simple as
namd2 +p20 +idlepoll config > log
On Tue, Nov 19, 2019 at 3:50 PM Daipayan Sarkar <sdaipayan_at_gmail.com<mailto:sdaipayan_at_gmail.com>> wrote:
I have compiled NAMD-2.13 version netlrts with cuda-10.0. The job scheduler software is torque and have been facing issues for submitting a job. Below is my NAMD job submission script. I submit the bash script (at the end) using torque job scheduler:
qsub -l nodes=1:ppn=20:gpus=1,naccesspolicy=singleuser submit.sh
Using "$charm ++local +p20 $namd +idlepoll ++ppn 19 equil.0.inp > equil.0.log" gives error
++ppn not recognized
Removing +ppn gives the following error:
Reason: FATAL ERROR: Number of devices (1) is not a multiple of number of processes (20). Sharing devices between processes is inefficient. Specify +ignoresharing (each process uses all visible devices) if not all devices are visible to each process, otherwise adjust number of processes to evenly divide number of devices, specify subset of devices with +devices argument (e.g., +devices 0,2), or multiply list shared devices (e.g., +devices 0,1,2,0).
Please advice on how to proceed.
==== submit.sh ======
#$charm ++local +p20 $namd +idlepoll ++ppn 19 equil.0.inp > equil.0.log
$charm ++local +p20 +idlepoll $namd equil.0.inp > equil.0.log
This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:12 CST