RELCCSD integral sorting/ordering

7 views
Skip to first unread message

Johann Pototschnig

unread,
Mar 25, 2022, 8:56:54 AMMar 25
to dirac-users
Dear RELCCSD experts,

I am trying to run RELCCSD on a supercomputer and the integral sorting/ordering at the beginning of RELCCSD is taking quite some time.

Does this step make use of any kind of parallelization?
Are there some other ways to speed it up?
Which MOLTRA scheme will be efficient?

best wishes,
Johann Pototschnig

Andre Gomes

unread,
Mar 25, 2022, 9:18:47 AMMar 25
to dirac...@googlegroups.com
hi,

as far as i understand it, at least for scheme 6  the code is parallelized in the sense that it will deal with integrals from different nodes.

but then it operates in a sort of round-robin scheme, with each rank reading data from file and then sharing it. so at the end of the day it feels like a sequential step.

i cannot say there'd be an improvement by using scheme 4.

best,

andre





--
You received this message because you are subscribed to the Google Groups "dirac-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dirac-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/e083f5e6-092f-433a-83bb-1ded962d3cb5n%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages