Closed mrakitin closed 6 years ago
That's a strange issue and it looks to be dependent on the script used for calculations. With https://github.com/mrakitin/sirepo_bugs/blob/master/issue_335/SRWLIB_VirtBL_ESM_01.py SRW produces similar results using both commands:
$ mpiexec -n 16 python SRWLIB_VirtBL_ESM_01.py --sm --und_g=35 --sm_mag=2 --sm_type=1 --sm_ei=10 --sm_ef=1200 --sm_ne=400 --op_r=30 --sm_rx=12e-03 --sm_ry=12e-03 --sm_fn=sm_flux_par.dat
$ python SRWLIB_VirtBL_ESM_01.py --sm --und_g=35 --sm_mag=2 --sm_type=1 --sm_ei=10 --sm_ef=1200 --sm_ne=400 --op_r=30 --sm_rx=12e-03 --sm_ry=12e-03 --sm_fn=sm_flux_seq.dat
With the analogous https://github.com/mrakitin/sirepo_bugs/blob/master/issue_335/mpi_run2.py generated in Sirepo it gives different results with and w/o mpiexec.
By adding of --sm_na=1
you can reproduce the issue.
Added @ochubar as a watcher to this ticket.
SRW - Discrepancies in the animated flux report on different servers (https://github.com/radiasoft/sirepo/issues/335).
Oleg does not observe the issue on his version of SRW. Need to debug it.