[CHiPPS-tickets] [COIN-OR High-Performance Parallel Search] #23: MPI BLIS hangs during termination check...

[COIN-OR High-Performance Parallel Search]

#23: MPI BLIS hangs during termination check...
-----------------------+----------------------------------------------------
Reporter:  nedwards    |        Owner:  yanxu
    Type:  defect      |       Status:  new  
Priority:  major       |    Component:  ALPS 
 Version:  stable/0.9  |   Resolution:       
Keywords:              |  
-----------------------+----------------------------------------------------
Comment (by nedwards):

 Replying to [comment:2 nedwards]:
 > I may have been too quick to claim success...
 >
 > I think my quick fix can result in dead-lock...

 Well, I've been running instances using this fix for a few days and
 haven't observed any deadlock - I was just too impatient for the
 termination check to finish.

 I've been checking the time spent in the explicit MPI_Wait and generally
 it is not too bad. My quick and dirty instrumentation shows the
 occassional multi-second wait, but this conservative approach doesn't seem
 to be causing a big slowdown.

 {{{
     MPI_Status sentStatus;
     struct timeval tv0,tv1;
     gettimeofday(&tv0,NULL);
     MPI_Wait(&forwardRequestL_, &sentStatus);
     MPI_Wait(&forwardRequestR_, &sentStatus);
     gettimeofday(&tv1,NULL);
     float delta = (tv1.tv_sec+tv1.tv_usec/1000000.0)
                   - (tv0.tv_sec+tv0.tv_usec/1000000.0);
     if (delta >= 0.001) {
       std::cout << " PROC " <<  globalRank_ << " L and R forward
 completed. (" << delta << " sec.)" << std::endl;
     }
 }}}

-- 
Ticket URL: <https://projects.coin-or.org/CHiPPS/ticket/23#comment:4>
COIN-OR High-Performance Parallel Search <http://projects.coin-or.org/CHiPPS>
A framework for data-intensive tree-search algorithms.