| 635 | }}} |
| 636 | |---------------- |
| 637 | {{{#!td style="vertical-align:top;width: 150px" |
| 638 | [=#collective_wait '''collective_wait'''] |
| 639 | }}} |
| 640 | {{{#!td style="vertical-align:top;width: 50px" |
| 641 | L |
| 642 | }}} |
| 643 | {{{#!td style="vertical-align:top;width: 75px" |
| 644 | see right |
| 645 | }}} |
| 646 | {{{#!td |
| 647 | Set barriers in front of collective MPI operations.\\\\ |
| 648 | Depending on the communication network in use, setting of barriers (MPI_BARRIER) in front of collective MPI operations (MPI_ALLTOALL, MPI_ALLREDUCE) may significantly speedup the code (about 10% on SGI-ICE-systems with Infiniband fat tree). Therefore, {{{collective_wait}}} is {{{.TRUE.}}} by default on SGI-systems (if hostname(3:5) = "{{{sgi}}}", given with '''mrun'''-option {{{-h}}}), but {{{.FALSE.}}} on all other systems. |