slurm job array
Hi,
I want to run an acusolve problem 4 times with same geometry, ic, bc, etc. except the inlet velocity=1,2,3,4 m/s. I want to use slurm job array for the same. Do you think I need to create separate directories for each problem? I will manually modify the input files. Do you think it would conflict the solution and log files?
Thanks.
Best Answer
-
If you intend to keep the same problem name, the runs could all be in the same directory, but I would run each consecutively - not concurrently. (The runtime/solution files might be all separated by run number - so maybe even concurrently would be OK in the same directory. Maybe test that first.) The solution files would then be separated by run number under the same problem name.
If you want to run them concurrently you could run with different directories - but you could also run within the same directory via:
Specify a different problem name for each run and specify the relevant input file for that run
acuRun -pb <problem_name> -inp <input_file_name> ...
2
Answers
-
If you intend to keep the same problem name, the runs could all be in the same directory, but I would run each consecutively - not concurrently. (The runtime/solution files might be all separated by run number - so maybe even concurrently would be OK in the same directory. Maybe test that first.) The solution files would then be separated by run number under the same problem name.
If you want to run them concurrently you could run with different directories - but you could also run within the same directory via:
Specify a different problem name for each run and specify the relevant input file for that run
acuRun -pb <problem_name> -inp <input_file_name> ...
2 -
acupro_21778 said:
If you intend to keep the same problem name, the runs could all be in the same directory, but I would run each consecutively - not concurrently. (The runtime/solution files might be all separated by run number - so maybe even concurrently would be OK in the same directory. Maybe test that first.) The solution files would then be separated by run number under the same problem name.
If you want to run them concurrently you could run with different directories - but you could also run within the same directory via:
Specify a different problem name for each run and specify the relevant input file for that run
acuRun -pb <problem_name> -inp <input_file_name> ...
Thank you.
0