[mpiwg-sessions] FW: MPI_Intercomm_from_group folow-up

[Pritchard Jr., Howard]

--
Howard Pritchard
B Schedule
HPC-ENV
Office 9, 2nd floor Research Park
TA-03, Building 4200, Room 203
Los Alamos National Laboratory


From: "Tony, Skelljum" <Tony-Skjellum at utc.edu<mailto:Tony-Skjellum at utc.edu>>
Date: Wednesday, January 9, 2019 at 10:47 AM
To: Howard Pritchard <howardp at lanl.gov<mailto:howardp at lanl.gov>>
Cc: MPI Sessions working group <mpiwg-sessions at lists.mpi-forum.org<mailto:mpiwg-sessions at lists.mpi-forum.org>>
Subject: Re: MPI_Intercomm_from_group folow-up


Howard, OK


1) What happens when non-group members call the intercomm function

2) What happens when not-all group members call the intercomm function

--> Dan convinced me that this is disallowed by the semantics as written

--> I want to explore weakening those semantics to see where that leads.


My further thought is we should explore

  *   Arms-length versions of these functions where it is intentionally allowed for non-members to call functions and not all members to call functions--for FT reasons, maybe in cases where we just need pt2pt communication
  *   When we introduce topology to groups vs. comms, intercomms with topology of groups on each side also pose an interesting new graph-to-graph connectivity idea that we might exploit for scalability and maybe for FT


Thanks,
Tony


Anthony Skjellum, PhD

Professor of Computer Science and Chair of Excellence

Director, SimCenter

University of Tennessee at Chattanooga (UTC)

tony-skjellum at utc.edu<mailto:tony-skjellum at utc.edu>  [or skjellum at gmail.com<mailto:skjellum at gmail.com>]

cell: 205-807-4968


________________________________
From: Pritchard Jr., Howard <howardp at lanl.gov<mailto:howardp at lanl.gov>>
Sent: Wednesday, January 9, 2019 12:21:03 PM
To: Skjellum, Anthony
Cc: MPI Sessions working group
Subject: MPI_Intercomm_from_group folow-up

HI Tony,

The audio was really bad a few minutes ago and I couldn’t understand what you were saying.
Could you summarize on the WG list what you were suggesting?

Thanks,

Howard

--
Howard Pritchard
B Schedule
HPC-ENV

Office 9, 2nd floor Research Park

TA-03, Building 4200, Room 203

Los Alamos National Laboratory

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-sessions/attachments/20190109/4fd033c5/attachment.html>