Re: dynamic counter-proposal
William Gropp (gropp@mcs.anl.gov)
Fri, 10 May 1996 15:56:46 -0500
| > 2) There are other issues that are likely to need to know if the spawn is
| > creating processes that are an extension of an existing set of a new set.
| > For example, the user may want to gang or co-schedule the new processes with
| > the existing set.
|
| But surely the application writer already knows at the time of the spawn call
| whether he/she wants (for example) gang scheduling? In the old proposal, this
| would probably be indicated via the 'info' argument, right? Same deal with the
| new proposal.
No, it is a "quality of implementation" issue. An implementation may want to
provide the service of gang scheduling for MPI applications by default in one
case and not in the other, allowing the user to override the default (null
info) choice if different behavior is desired. For most of the MPI
applications that we see, gang-scheduling is what the user is expecting (or at
least hoping for and designing for), and when they don't get it, they suffer
performance problems. Of course, I'm talking about MPPs again.
Bill