Figure 
shows the transfer time between two nodes for
small messages (up to 1024 bytes). This figure does not show the results for
PVMe with PvmDataRaw, since they are the same as the results
for PvmDataInPlace.
Figure: Latency on the IBM SP-2: PVM3, PVMe, MPL
We computed the latencies from Figure using a least squares interpolation.
They are given in the following table.
The latency of pvm_psend-pvm_precv is roughly the same as that of mpc_bsend-mpc_brecv. For the same reason as on the CM-5 and the Paragon, PvmDataRaw gives a lower latency than PvmDataInPlace.