10 ms·
Indeed -- memory- or communication-bound. Somewhere there's a Dell "roofline" plot indicating IPC in some codes running in an unspecified way. See also John Mc
by gnufx 8y ago
Indeed -- memory- or communication-bound. Somewhere there's a Dell "roofline" plot indicating IPC in some codes running in an unspecified way. See also John McCalpin's writings.
Things doing 3-D FFTs (common in materials science) tend to spend a lot of time in an MPI collective at sufficient scale. I spent half an hour profiling and then changing an MPI parameter for 30% improvement in one case at not very large scale.