Cornell Theory Center

---

This is the presentation layer of a two-part module. For an explanation of the layers and how to navigate within and between them, return to the top page of this module.

1. Overview of MPI

   1.1 What Is MPI?

   1.2 What does MPI offer?

   1.3 MPI at the Cornell Theory Center

   1.4 How to use MPI

2. MPI programs

   2.1 Format of MPI routines

   2.2 MPI routines

   2.3 An MPI sample program

3. MPI messages

   3.1 Data

   3.2 Envelope

4. Communicators

   4.1 Why have communicators?

   4.2 Communicators and process groups

5. Summary

References Lab Exercises Evaluation

1. Overview

• Message Passing Interface standard

• The first standard and portable message passing library with good performance

• "Standard" by consensus of MPI Forum participants from over 40 organizations

• Finished and published in May 1994, updated in June 1995

• MPI 2 complete as of July 1997. Extends (does not change) MPI.

• Documents that define MPI are on the World Wide Web at Argonne National Lab

1.2 What does MPI offer?

• Standardization - on many levels

• Portability - to existing and new systems

• Performance - comparable to vendors' proprietary libraries

• Richness - extensive functionality, many quality implementations

1.3 MPI at the Cornell Theory Center

• Two implementations are installed on CTC's SP system
  1. IBM's MPI product for RS/6000 clusters and SP systems

  2. MPICH, a portable implementation from Argonne National Lab and Mississippi State University

  This workshop focuses on the IBM implementation.

• Performance of MPI is comparable to MPL, an IBM proprietary library developed specifically for the SP architecture

• IBM's MPI and MPL products have now been merged so that it is possible to mix MPI and MPL calls in the same program

• IBM MPI includes "MPE" routines which provide enhanced functionality and performance (at cost of portability)

1.4 How to Use MPI

• When possible, start with a debugged serial version

• Design parallel algorithm

• Write code, making calls to MPI library

• Compile and run using IBM's Parallel Environment
  (covered in the IBM SP Parallel Environment module)

• Run with a few nodes first, increase number gradually

2. MPI programs

2.1 Format of MPI routines

• C bindings
  • rc = MPI_Xxxxx(parameter, ... )
  • rc is error code, = MPI_SUCCESS if successful

• Fortran bindings
  • call MPI_XXXXX(parameter,..., ierror)
  • case insensitive
  • ierror is error code, = MPI_SUCCESS if successful

• Exception: timing functions return double precision reals

• Header file required
  • #include "mpi.h" for C programs
    
  • include 'mpif.h' for Fortran programs
    

2.2 MPI routines

• MPI has over 125 functions, but you can do much with just 6:

  • Initialize for communications

    MPI_INIT: initializes the MPI environment

    MPI_COMM_SIZE: returns the number of processes

    MPI_COMM_RANK: returns this process's number (rank)

  • Communicate to share data between processes

    MPI_SEND: sends a message

    MPI_RECV: receives a message

  • Exit in a "clean" fashion when done communicating
    

    MPI_FINALIZE

2.3 An MPI sample program (Fortran)

Click here for C version

   program hello
   include 'mpif.h'
   integer rank, size, ierror, tag, status(MPI_STATUS_SIZE)
   character(12) message
	
   call MPI_INIT(ierror)
   call MPI_COMM_SIZE(MPI_COMM_WORLD, size, ierror)
   call MPI_COMM_RANK(MPI_COMM_WORLD, rank, ierror)
   tag = 100

   if (rank .eq. 0) then
      message = 'Hello, world'
      do i=1, size-1
         call MPI_SEND(message, 12, MPI_CHARACTER, i, tag,
  &                    MPI_COMM_WORLD, ierror)
      enddo

   else
    	 call MPI_RECV(message, 12, MPI_CHARACTER, 0, tag,
  &                    MPI_COMM_WORLD, status, ierror)
   endif

   print*, 'node', rank, ':', message
   call MPI_FINALIZE(ierror)
   end

3. MPI messages

 call MPI_SEND(startbuf,count,datatype,
                   \     |    /
                    ---DATA--- 

                 dest,tag,comm,  ierror)
                   \   |   /
                    ENVELOPE

3.1 Data

• Arguments
  • startbuf (starting location of data)
  • count (number of elements)

    receive >= send

  • datatype (basic or derived)

    receiver = send (unless MPI_PACKED)

• MPI basic datatypes for Fortran

    ---------------------------------------
    MPI Datatype           Fortran Datatype  
    ---------------------------------------
    MPI_INTEGER            INTEGER           
    MPI_REAL               REAL              
    MPI_DOUBLE_PRECISION   DOUBLE PRECISION  
    MPI_COMPLEX            COMPLEX           
    MPI_LOGICAL            LOGICAL           
    MPI_CHARACTER          CHARACTER(1)      
    MPI_BYTE                                 
    MPI_PACKED                               
    ---------------------------------------
  

  • C datatypes

• Derived datatypes
  • mixed datatypes
  • non-contiguous data

3.2 Envelope

• Destination or source
  • rank in a communicator
  • receive = sender or MPI_ANY_SOURCE
• Tag
  • integer chosen by programmer
  • receive = sender or MPI_ANY_TAG
• Communicator
  • defines communication "space"
  • receive = send
• Analogy: bill collection agency

4. Communicators

4.1 Why have communicators?

• If you are writing a library, can you choose safe tags?
• Example: variable (and possibly incorrect) behavior if no communicator

Animation

4.2 Communicators and process groups

• Programmers can define a process group
• And define new communicator(s) for the process group

5. Summary

• MPI program: 6 basic calls
  • MPI_INIT
  • MPI_COMM_RANK
  • MPI_COMM_SIZE
  • MPI_SEND
  • MPI_RECV
  • MPI_FINALIZE

• MPI messages
  • data (startbuf, count, datatype)
  • envelope (destination/source, tag, communicator)

• Communicators

References

Message Passing Interface Forum (1995) MPI: A Message Passing Interface Standard. June 12, 1995. Available online or in postscript.

IBM manual for their implementation of MPI is available online

CTC's MPI Documentation

Lab exercises for MPI Basics

Please complete this short evaluation form. Thank you!