Hi expert,
I meet with a very special problem which is from our customer by using LS-DYNA in windows hpc server 2008 sp1.
When we submit a job for LS-DYNA application computing on one compute node by job scheduler, the job will fail just after submission or after a period of computing without clear error message output. It seems that the
job is crashed.
But if we let the LS-DYNA compute directly on the same compute node, it can compute successfully.
Also we monitor the memory usage of the compute node when the LS-DYNA application is computing by using the "view performance chart" and find that the memory of the compute node will be exhausted suddenly by the computing
which through job scheduler but the another one (not through job scheduler) not.
[by the way, how can i upload a image in this thread? so i can upload the performance chart image for the usage of the memory of the compute node when computing for the two situation]
So i wonder if there is any bug with the job scheduler or any compatibility issue with LS-DYNA or others?
Here is some information for your reference.
HPC Version: windows hpc server 2008 with hpc pack sp1
Compute Node: 32G Memeory, 2*4 Cores Intel Xeon CPU
Network: enterprise 1Gb, private 1Gb, application infiniband
LS-DYNA verison: ls971d R2
7600.1224 SMP
LS-DYNA command line:
c:\SmtcProgram\LSDYNA\program\ls971_d_7600.1224_winx64_p\ls971_d_7600.1224_winx64_p.exe i=deg.k ncpu=8 memory=1500MB
Thanks
