Meeting Agenda

March 5th - 7th, 2012

Agenda

Monday, March 5, 2012

Time Title Link Video phone number WebEx Password
1:00pm - 2:00pm Plenary Session: Hybrid working group -
Clarify MPI behavior when multiple MPI processes run in the same address space - formal reading Ticket #310
2:00pm - 3:00pm Plenary Session: Hybrid working group -
Provide a mechanism for migrating threads between MPI processes - formal reading Ticket #311
3:00pm - 4:00pm Plenary Session: Hybrid working group -
Helper Threads - formal reading Ticket #217
4:00pm - 4:15pm Break
4:15pm - 6:30pm Planary Session: Fault Tolerance working group Formal Readings -
User-Level Failure Mitigation - Ticket #323,
Clarify MPI_ERRORS_ARE_FATAL scope of abort - Ticket #324
6:30pm - 7:30pm Dinner on own
7:30pm - 10:00pm Planary Session: Fault Tolerance working group Formal Readings -
User-Level Failure Mitigation - Ticket #323,
Clarify MPI_ERRORS_ARE_FATAL scope of abort - Ticket #324

Tuesday, March 6, 2012

Time Title Link Video phone number WebEx Password
8:45am - 9:30am Existing ticket 0 -
Mention C++ typedef deprecations - ticket #185,
Minor Text Correction / MPI_COMM_GET_NAME + MPI_GET_PROCESSOR_NAME - ticket #207,
MPI-2.2 typo - missing parenthesis on page 272, line 25 - ticket #215,
Typo in the description of MPI_GROUP_INCL - ticket #254,
Missing headline for MPI_GET_PROCESSOR_NAME - ticket #255,
Minor typos in pt2pt chapter - ticket #262,
wrong formula for the upper bound - ticket #263,
Typo in One-Sided chapter 11.4.4 - ticket #267,
Formatting issue in One-Sided chapter 11.4.4 - ticket #268,
MPI 2.1 Clarification - MPI_Cart_map with num_dims=0 - ticket #162,
s/legal/something_better/g - ticket #182,
MPI_MAX_OBJECT_NAME and MPI_Type/win_get_name - ticket #219,
Typo-Correction in ALLTOALLV - ticket #312
9:30am - 10:00am Plenary Session: MPI_info object - formal reading - Ticket #271
10:00am - 10:15am Votes
10:15am - 10:30am Break
10:30am - 11:30am Votes
11:30am - 12:30pm Working lunch: Plenary Session: Formal Readings:
Add Immediate versions of nonblocking collective I/O routines (Koziel) - ticket #273,
Noncollective Communicator Creation(Dinan) - ticket #286
12:30pm - 4:30pm Plenary Session: Formal Readings:
Update MPI_Intercomm_create to use collective tag space(Dinan) - ticket #305,
Update examples to not use deprecated constructs (where possible)(Tillier) - ticket #278,
Remove C++ Bindings(Tillier) - ticket #281,
MPI_UNWEIGHTED should not be NULL - ticket #294
Change MPI_OFFSET external32 size to 16 Bytes - ticket #196,
MPI_WAIT/TEST language unclear about non blocking I/O, Generalized, or collective requests - ticket #321,
Topology awareness in MPI_Dims_create(Balaji) - ticket #195,
Move MPI-1 deprecated functions to new Removed Interfaces chapter(Tillier) - ticket #303
4:30pm - 4:45pm Break
4:45pm - 6:30pm Plenary Session: Formal Readings:
Clarification that MPI_Cart_map and MPI_Graph_map are local calls(Traff/Rabenseifner) - ticket #158,
For reductions: Grouping MPI_AINT, MPI_OFFSET, MPI_COUNT as Multi-language types(Ritzdorf/Rabenseifner) - ticket #187,
Clarify ambiguous sentence for MPI_FINALIZE(Snir) - ticket #227,
MPI_PROC_NULL behavior for MPI_PROBE and MPI_IPROBE not directly defined(Squyres) - ticket #256,
Text updates for MPI_DIST_GRAPH(Moody) - ticket #259
6:30pm - 7:30pm Dinner on own
7:30pm - 9:30pm Plenary Session: Formal Readings:
Consistent use of [] for input arrays(Tillier) - ticket #126,
K&R style on 6.7.6(Tillier) - ticket #293

Wednesday, March 7, 2012

Time Title Link Video phone number WebEx Password
9:00am - 10:00am Plenary Session: Formal Reading:
Update examples to not use deprecated constructs (where possible)(Tillier) - ticket #278,
Plenary Session: Relaxing MPI_TYPE_GET_ENVELOPE combiner rules
10:00am - 12:00pm Planary Session: Formal Readings
User-Level Failure Mitigation - Ticket #323,
MPI3 Fault Tolerance - Files - Ticket #326,
MPI3 Fault Tolerance - Dynamic process management - Ticket #327,
Clarify MPI behavior when multiple MPI processes run in the same address space - Ticket #310,
Provide a mechanism for migrating threads between MPI processes - Ticket #311,
MPI_INIT & MPI_FINALIZE - Ticket #313
12:00pm - 12:30pm Working Lunch: Plenary Session: MPI 3.0 status
12:30pm - 3:30pm Plenary Session: Formal Readings:
Nonblocking Communicator Duplication - ticket #168,
MPI_IN_PLACE in Gather (Hoefler) - ticket #163,
Fix issue with definition of nonblocking in One Sided Chapter (Hoefler) - ticket #300,
Fix MPI_PROC_NULL behavior for mprobe/improbe/mrecv/imrecv - ticket #328,
Topology awareness in MPI_Dims_create - ticket #195,
C++ binding for Dist_graph_neighbors_count is badly broken - ticket #192,
Correct error related to MPI_REQUEST_FREE - ticket #317

Votes

Ticket Zero Votes

  • #0: Big ticket 0 glob: #162, #182, #185, #207, #215, #219, #254, #255, #262, #263, #312, #322
  • #125: Use of [] in output arrays
  • #126: Consistent use of [] for input arrays
  • #293: K&R style on 6.7.6
  • #321: MPI_WAIT/TEST language unclear about non blocking I/O, Generalized, or collective requests.
  • #158: Clarification that MPI_Cart_map and MPI_Graph_map are local calls
  • #259: Text updates for MPI_DIST_GRAPH

First Vote

  • #308: RMA - Remove sentence on offset in datatypes
  • #309: RMA - Clarify usage of status for request-based RMA operations
  • #284: Allocate a shared memory window
  • #280: Add a new datatype creation routine - MPI_Type_create_hindexed_block
  • #168: Nonblocking Communicator Duplication
  • #272: Remove non-blocking collective C++ bindings
  • #286: Noncollective Communicator Creation
  • #305: Update MPI_Intercomm_create to use collective tag space

Second Vote

  • #298: Adding clarification to MPI_WIN_LOCK_ALL
  • #229: New MPI-3 Fortran Support - #A: Overview over all related tickets
  • #29: Change displs arguments in neighborhood w collectives to MPI_Aint