j
k
j a
j l
A small clarification ... I always used 64 mpi processes to run my code (hence, the 30-40 speedup performace is because of your new code).
Xavier
Back to the thread
Back to the list