Dear FLUKA experts,
some of my jobs are failing with the message
Abort called from SJPCMP reason 3500?? Run stopped!
while other seeds run through without problems. I attached one failing job. The custom source was linked with ldpmqmd.
ranSIS100_Dump_Z36_A84_Q16_E460MeVu_07001.ran (1.6 KB) ranSIS100_Dump_Z36_A84_Q16_E460MeVu_07002.ran (1.6 KB) SIS100_Dump_Z36_A84_Q16_E460MeVu_07.inp (6.4 KB) SIS100_Dump_Z36_A84_Q16_E460MeVu_07001.err (7.4 KB) SIS100_Dump_Z36_A84_Q16_E460MeVu_07001.log (134 Bytes) SIS100_Dump_Z36_A84_Q16_E460MeVu_07001.out (129.3 KB) source_twissbeam.f (11.1 KB)
I have a FLUKA job
SIS100_Dump_Z92_A238_Q28_E275MeVu_03.inp (3.4 KB) source_twissbeam.f (11.1 KB)
(source linked with ldpmqmd)
that keeps crashing/aborting. I have tried it with my local FLUKA installation (fluka_4-1.0.Linux-gfor7_amd64.deb) and a cluster installation of my IT department (fluka-4-1.0.Linux-gfor9.tgz).
My local version fluka_4-1.0.Linux-gfor7_amd64.deb aborts with the
message reported earlier:
ranSIS100_Dump_Z92_A238_Q28_E275MeVu_03001.ran (1.6 KB) ranSIS100_Dump_Z92_A238_Q28_E275MeVu_03002.ran (1.6 KB) SIS100_Dump_Z92_A238_Q28_E275MeVu_03001.err (47.2 KB) SIS100_Dump_Z36_A84_Q16_E460MeVu_07001.log (134 Bytes) SIS100_Dump_Z92_A238_Q28_E275MeVu_03001.out (65.7 KB)
The cluster version fluka-4-1.0.Linux-gfor9.tgz on the other hand crashes with
At line 54 of file preeq/xinneu.f
Fortran runtime error: Index ‘102’ of dimension 1 of array ‘zznxse’ above upper bound of 100
ranSIS100_Dump_Z92_A238_Q28_E275MeVu_03001.ran (1.6 KB) ranSIS100_Dump_Z92_A238_Q28_E275MeVu_03002.ran (1.6 KB) SIS100_Dump_Z92_A238_Q28_E275MeVu_03001.err (22.7 KB) SIS100_Dump_Z92_A238_Q28_E275MeVu_03001.log (674 Bytes) SIS100_Dump_Z92_A238_Q28_E275MeVu_03001.out (64.2 KB)
As you can see, not only the errors seem to be different, but they also seem to encounter different problems. The cluster version fails after some 20k primaries while my local version exceeds 40k primaries.
Note: it seems the first .log file here is replaced with the one of an earlier report by the forum software since they are identical. Let me assure you that the STOP 3500?? log file of this job is indeed the same.
Thanks @rmartin. Both the STOP 3500?? and STOP 2500 crashes, related to a spin initialization issue for some unfortunate nuclei, are being solved (we may contact you privately to further test the fix with your well probing user case). As for the zznxse one, it’s a pre-equilibrium issue with a superheavy nucleus we shall address. Sorry for the inconvenience.
Thanks again for your crash reports!
All three bugs have been corrected and the respective fixes will be made available with the next minor release of the code, to take place early 2021.
With kind regards,