1
1
openmpi/orte/mca/errmgr
Ralph Castain 8f496b01b7 Try automatically adding local spawn threads to parallelize the fork/exec process to speed up the launch on large SMPs. Harvest the threads after initial spawn to minimize any impact on running jobs.
Change the determination of #spawn threads to be done on basis of #local procs in first job being spawned. Someone can look at an optimization that handles subsequent dynamic spawns that might be larger in size.

Leave the threads running, but blocked, for the life of the daemon, and use them to harvest the local procs as they terminate. This helps short-lived jobs in particular.

Add MCA params to set:
  * max number of spawn threads (default: 4)
  * set a specific number of spawn threads (default: -1, indicating no set number)
  * cutoff - minimum number of local procs before using spawn threads (default: 32)

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2017-11-29 19:54:00 -08:00
..
base Update OPAL and ORTE for thread safety 2017-06-06 12:30:57 -07:00
default_app mca: Dynamic components link against project lib 2017-08-24 11:56:16 -04:00
default_hnp mca: Dynamic components link against project lib 2017-08-24 11:56:16 -04:00
default_orted Try automatically adding local spawn threads to parallelize the fork/exec process to speed up the launch on large SMPs. Harvest the threads after initial spawn to minimize any impact on running jobs. 2017-11-29 19:54:00 -08:00
default_tool mca: Dynamic components link against project lib 2017-08-24 11:56:16 -04:00
dvm mca: Dynamic components link against project lib 2017-08-24 11:56:16 -04:00
errmgr.h Update OPAL and ORTE for thread safety 2017-06-06 12:30:57 -07:00
Makefile.am Purge whitespace from the repo 2015-06-23 20:59:57 -07:00