Hi,Is there a way to figure the best possible PE Configuration on hybrid execution for CESM, so as to get the highest throughput.I am using a cluster with 8 nodes and 16 cores/node.
It depends which machine you are using. Examples of pe_layout can be found on the CESM webapge. For insatnce:http://www.cesm.ucar.edu/models/cesm1.2/timing/
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.