Assessing gains from parallel computation on supercomputers
-
1
Stanford University
info
Year of publication: 2013
Issue: 10
Pages: 1-14
Type: Working paper
Abstract
We assess gains from parallel computation on Backlight supercomputer. We find that information transfers are expensive. To make parallel computation efficient, a task per core must be sufficiently large, ranging from few seconds to one minute depending on the number of cores employed. For small problems, the shared memory programming (OpenMP) leads to a higher efficiency of parallelization than the distributive memory programming (MPI).