From: Vermaas, Josh (vermaasj_at_msu.edu)
Date: Tue Aug 30 2022 - 08:56:34 CDT
Hi Shivam,
Libmvec is the gnu vector math library, and namd3 is unable to find it on your system, which means that it can’t run. Is there any reason why that wouldn’t be in the library path on your HPC resources? Plausible reasons might include using something other than the GNU compiler stack (intel?), but that would in some sense be unusual.
-Josh
From: <owner-namd-l_at_ks.uiuc.edu> on behalf of SHIVAM TIWARI <t.shivam_at_iitg.ac.in>
Reply-To: "namd-l_at_ks.uiuc.edu" <namd-l_at_ks.uiuc.edu>, SHIVAM TIWARI <t.shivam_at_iitg.ac.in>
Date: Tuesday, August 30, 2022 at 7:39 AM
To: namd-l <namd-l_at_ks.uiuc.edu>
Subject: namd-l: error while running simulation with NAMD3.0alpha12 and 13
Dear all,
I have a system with around 2,63000 atoms, where some atoms are kept fixed. I am trying to run the simulation with namd 3.0 on my HPC which has NVIDIA tesla K40. I downloaded alpha13 single node version from namd3 page, and uploaded the same on my account on HPC; however, when I m trying to run the simulation with it I am getting the following error:
./NAMD_3.0alpha13_Linux-x86_64-multicore-CUDA/namd3: error while loading shared libraries: libmvec.so.1: cannot open shared object file: No such file or directory
then I tried with alpha12 version but again I m getting the same error. However, when I tried with alpha8 version it didn't ran with it either, but the error was different, it was about the fixed atoms command in my configuration file, since alpha8 doesn't have the fixed atom feature included in it (as mentioned on namd3 webpage). So I tried removing the fixed atoms command and running it, this time the simulation was running, but when I was checking the progress on log file (tail -f logfile) the simulation is not proceeding after it has produced the initial benchmark information and no output files were produced.
Also when I looked at the performance of namd 3.0 alpha8 (1 node, 1gpu, 1core) in comparison with namd2.14 multicore cuda( 1 GPU, 10 cores,single node) I didn't see any gain in the performance, both were giving me around 4 ns/day.
Please help me resolve these issues, also please suggest the best strategy to gain the best performance out of the available resources. The job submission script that I am using is as follows:
#!/bin/sh
#BATCH -A
#SBATCH --nodes 1
#SBATCH --ntasks-per-node=10
#SBATCH --account=project
#SBATCH --gres=gpu:1
#SBATCH --qos=project
#SBATCH --partition=high
#SBATCH --mem=10gb
#SBATCH --error=job.%J.err
#SBATCH --output=job.%J.out
#SBATCH --time=10-00:00:00
#SBATCH --no-requeue
namdexec=../NAMD_3.0alpha12_Linux-x86_64-multicore-CUDA/namd3
${namdexec} +p $SLURM_NTASKS +setcpuaffinity prod_pans.conf > tst.log
regards
shivam
This archive was generated by hypermail 2.1.6 : Tue Dec 13 2022 - 14:32:44 CST