[Wien] Missing k-points in SGE job

[Atta-fynn, Raymond]

Dear users,

I was running parallel jobs on a supercomputer using SGE and WIEN2k_08 when I noticed a strange problem which causes my job to crash. Suppose that I'm running 84 k-points on 6 nodes, then each node will handle 14 k-points and so I should have six sets of k-point files each containing 14 k-points: case.klist_1 [k-points 1-14], case.klist_2 [k-points 15-28], case.klist_3 [k-points 29-42], case.klist_4 [k-points 43-56], case.klist_5 [k-points 57-70] and case.klist_6 [k-points 71-84].

However, the last k-point subset case.klist_6 always has the first two k-points missing from the list (instead of 71-84 in the example above, it is 73-84). This happen in all other parallel jobs; the last k-point subset always has the first two k-points in the list missing. I was wondering if someone with experience in SGE can help me figure this out.

Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20091118/aefd6292/attachment.htm>