Skip to content

Latest commit

 

History

History
32 lines (27 loc) · 1.48 KB

README-nonblocking.md

File metadata and controls

32 lines (27 loc) · 1.48 KB
Raw

Message chain with non-blocking communication

Write a program where every MPI task sends data to the next one. Let ntasks be the number of the tasks, and myid the rank of the current task. Your program should work as follows:

  • Every task with a rank less than ntasks-1 sends a message to task myid+1. For example, task 0 sends a message to task 1.
  • The message content is an integer array where each element is initialised to myid.
  • The message tag is the receiver's rank.
  • The sender prints out the number of elements it sends and the tag it used.
  • All tasks with rank > 0 receive messages.
  • Each receiver prints out their myid and the first element in the received array.

  1. Implement the program described above using non-blocking communication, i.e. MPI_Isend, MPI_Irecv, and MPI_Wait. Utilize MPI_PROC_NULL when treating the special cases of the first and the last task. You may start from scratch or use the skeleton code (cpp/skeleton.cpp, c/skeleton.c or fortran/skeleton.F90) as a starting point.

  2. The skeleton code prints out the time spent in communication. Investigate the timings with different numbers of MPI tasks (e.g. 2, 4, 8, 16, ...). Compare the results to the implementation with MPI_Sends and MPI_Recv's and pay attention especially to rank 0. Can you explain the behaviour?

  3. Write a version that uses MPI_Waitall instead of MPI_Waits.