All,
I've committed r.557 into subversion, which is a simple modification to
SetupHarnessPipeline.py. It lists every element of PBS_NODEFILE twice
in our nodelist, which actually doubles our MPI universe size and
doubles the number of stageslices on each node.
This seems to work without breaking anything and yields about a speedup
of about 1.5.
I managed to run 5 of our 6 stages with 6 CCDs/node (our original number
from the design) and get each image through the pipeline in about 93
seconds (latency of 93 sec).
It's worth noting that this speedup of 1.5 for using both cores is
different from the speedup of 2 for halving the number of CCDs/node, and
probably more realistic.
Cheers,
-Jon
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
