Visible to Intel only — GUID: GUID-C4CFD699-7493-408F-9E00-A2CF829FDD54
Measuring Communication and Computation Overlap
Measuring Pure Communication Time
Iallgather
Iallgather_pure
Iallgatherv
Iallgatherv_pure
Iallreduce
Iallreduce_pure
Ialltoall
Ialltoall_pure
Ialltoallv
Ialltoallv_pure
Ibarrier
Ibarrier_pure
Ibcast
Ibcast_pure
Igather
Igather_pure
Igatherv
Igatherv_pure
Ireduce
Ireduce_pure
Ireduce_scatter
Ireduce_scatter_pure
Iscatter
Iscatter_pure
Iscatterv
Iscatterv_pure
Visible to Intel only — GUID: GUID-C4CFD699-7493-408F-9E00-A2CF829FDD54
PingPing
Use PingPing to measure a PingPing message passing between two ranks in each pair of ranks.
Property |
Description |
---|---|
Measured pattern |
The ranks are broken into pairs: (0, N/2), (1, N/2 + 1), (2, N/2 + 2) and so on. The number of ranks N must be even. |
MPI routines |
MPI_Irecv, MPI_Send, MPI_Wait |
MPI data type |
MPI_BYTE |
The benchmark reports latency (in microseconds), bandwidth (in million bytes per second), and message rate (in messages per second).
NOTE:
IMB-P2P PingPing and IMB-MPI1 PingPing are different benchmarks (IMB-MPI1 PingPing uses MPI_Isend, MPI_Recv, and MPI_Wait).