1 / 13

Message Passing Interface (MPI)

Message Passing Interface (MPI). Jonathan Carroll-Nellenback CIRC Summer School. Background. MPI – message passing interface Language independent communications protocol MPI-1, MPI-2, and MPI-3 standards

Download Presentation

Message Passing Interface (MPI)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Message Passing Interface (MPI) Jonathan Carroll-Nellenback CIRC Summer School

  2. Background • MPI – message passing interface • Language independent communications protocol • MPI-1, MPI-2, and MPI-3 standards • Implementations typically consist of a specific set of routines callable from C, C++, or Fortran • MPICH - 3.1.2 (implements MPI – 3) • Open MPI – 1.8 (Implements MPI – 2) • IMPI – Intel MPI • MVAPICH, • Commercial implementations • Bindings for Python, Java, etc... http://web.eecs.utk.edu/~dongarra/WEB-PAGES/SPRING-2006/mpi-quick-ref.pdf

  3. Outline • First MPI Program • /public/jcarrol5/mpi/example1.f90 • MPI_Init – initializes mpi • MPI_Comm_rank – gets the current task rank within the communicator (starting at 0) • MPI_Comm_size – gets the size of the communicator • MPI_Finalize – closes mpi • MPI_Reduce – Collective communication • MPI_Allreduce – Collective communication • Compiling • module load openmpi • mpif90 example1.f90 (mpicc example1.c) • srun –p debug –n 4 –o output_%ta.out

  4. Exercise 1 • Parallelize exercise1.f90 or exercise1.c

  5. Collective Communication Routines • /public/jcarrol5/mpi/example2.f90 • 1 to All • MPI_Bcast – Broadcasts the same data to all ranks • MPI_Scatter – Evenly Distributes data to all ranks • MPI_Scatterv – Unevenly distributes data to all ranks • All to 1 • MPI_Reduce – Performs a reduction operation towards a single rank • MPI_Gather – Collects evenly distributed data on one rank • MPI_Gatherv – Collects unevenly distributed data on one rank • All to All • MPI_Allreduce – Performs a reduction and broadcasts the result • MPI_Allgather – Collects evenly distributed data onto all ranks • MPI_Allgatherv – Collects unevenly distributed data onto all ranks • MPI_Alltoall – Scatter/Gather • MPI_Alltoallv – Scatterv/Gatherv

  6. Important Constants • Reduction Operations – MPI_MAX, MPI_MIN, MPI_SUM, MPI_PROD, MPI_BAND, MPI_BOR, MPI_BXOR, MPI_LAND, MPI_LOR, MPI_LXOR • C Data types – MPI_CHAR, MPI_SHORT, MPI_INT, MPI_LONG, MPI_UNSIGNED_CHAR, MPI_UNSIGNED_SHORT, MPI_UNSIGNED, MPI_UNSIGNED_LONG, MPI_FLOAT, MPI_DOUBLE, MPI_LONG_DOUBLE, MPI_BYTE, MPI_PACKED • Fortran Data types – MPI_CHARACTER, MPI_INTEGER, MPI_REAL, MPI_LOGICAL, MPI_INTEGER1, MPI_INTEGER2, MPI_INTEGER4, MPI_REAL2, MPI_REAL4, MPI_REAL8, MPI_DOUBLE_PRECISION, MPI_COMPLEX, MPI_DOUBLE_COMPLEX, MPI_BYTE, MPI_PACKED

  7. Exercise 2 • Parallelize exercise2.f90 or exercise2.c

  8. Basic Sending and Receiving • /public/jcarrol5/mpi/example3.f90 • Tags – additional identifiers on messages • MPI_Send • MPI_Recv

  9. Exercise 3 • Modify your program from exercise2 to not use any global communication routines.

  10. Sending modes • Blocking vs Non-blocking • Non Blocking sends and receives will immediately return control to the calling routine. However, they usually will require buffering and testing later on to see whether the send/recv has completed. • Good for overlapping communication with computation • May lead to extra buffering • Synchronous vs Asynchronous • Synchronous sends require a matching recv to be called before returning. Blocking only if recv has not been posted. Does not require any additional buffering. • Buffered vsNonBuffered • Buffered sends explicitly buffer the data to be sent so that the calling routine can release the memory. • Ready send • Assumes that the receiver has already posted the recv.

  11. Send Routines • /public/jcarrol5/mpi/example4.f90 • MPI_Send – May or may not block • MPI_Bsend – May buffer – returns immediately • MPI_Ssend – Synchronous Send (returns after matching recv posted) • MPi_Rsend – Ready send (matching recv must be posted) • MPI_Isend – Nonblocking send (must check for completion) • MPI_Ibsend – Nonblocking buffered send • MPI_Issend – Nonblocking synchronous send • MPI_Irsend - Nonblocking ready send • MPI_Recv – Blocking receive • MPI_IRecv – Nonblocking receive

  12. Exercise 5 • Rewrite exercise 3 using ready sends(rsend), synchronous sends (ssend), and nonblocking sends (isend) and see if it is any faster.

  13. Communicators and Groups • /public/jcarrol5/mpi/example5.f90 • MPI starts with one communicator (MPI_COMM_WORLD) • Separate communicator groups can be formed using • MPI_Comm_split • Or you can extract the group belonging to mpi_comm_world and create subgroups through various routines. • Multiple communicators can use the same group.

More Related