Message-Passing Computingpeople.scs.carleton.ca/~achan/teaching/2002-comp... · Message Tag Used to differentiate between different types of messages being sent. Message tag is carried

Slides for Parallel Programming Techniques and Applications Using Networked Workstations and Parallel Computers by Barry Wilkinson and Michael Allen,Prentice Hall Upper Saddle River New Jersey, USA, ISBN 0-13-671710-1. 2002 by Prentice Hall Inc. All rights reserved.

Slide 41

Message-Passing Computing

Chapter 2

Slide 42

Basics of Message-Passing Programming using user-level message passing libraries

Two primary mechanisms needed:

1. A method of creating separate processes for execution on

different computers

2. A method of sending and receiving messages

Slide 43

Single Program Multiple Data (SPMD) model

Different processes merged into one program. Within program,control statements select different parts for each processor toexecute. All executables started together - static process creation.

Sourcefile

Executables

Processor 0 Processor n − 1

Compile to suitprocessor

Basic MPI way

Slide 44

Process 1

Process 2spawn();

Time

Start executionof process 2

Multiple Program Multiple Data (MPMD) Model

Separate programs for each processor. Master-slave approachusually taken. One processor executes master process. Otherprocesses started from within master process - dynamic processcreation.

PVM way

Slide 45

Process 1 Process 2

send(&x, 2);

recv(&y, 1);

x y

Movementof data

Basic “point-to-point” Send and Receive Routines

Passing a message between processes using send() and recv()library calls:

Generic syntax (actual formats later)

Slide 46

Synchronous Message Passing

Routines that actually return when message transfer completed.

Synchronous send routine

Waits until complete message can be accepted by the receivingprocess before sending the message.

Synchronous receive routine

Waits until the message it is expecting arrives.

Synchronous routines intrinsically perform two actions: Theytransfer data and they synchronize processes.

Slide 47

Synchronous send() and recv() library calls using 3-way protocolProcess 1 Process 2

send();

recv();Suspend

Time

processAcknowledgment

MessageBoth processescontinue

(a) When send() occurs before recv()

Process 1 Process 2

recv();

send();Suspend

Time

process

Acknowledgment

MessageBoth processescontinue

(b) When recv() occurs before send()

Request to send

Slide 48

Asynchronous Message Passing

Routines that do not wait for actions to complete before returning.

Usually require local storage for messages.

More than one version depending upon the actual semantics for

returning.

In general, they do not synchronize processes but allow processes

to move forward sooner. Must be used with care.

Slide 49

MPI Definitions of Blocking and Non-Blocking

Blocking - return after their local actions complete, though the

message transfer may not have been completed.

Non-blocking - return immediately.

Assumes that data storage to be used for transfer not modified by

subsequent statements prior to tbeing used for transfer, and it is left

to the programmer to ensure this.

Notices these terms may have different interpretations in other

systems.)

Slide 50

Process 1 Process 2

send();

recv();

Message buffer

Readmessage buffer

Continueprocess

Time

How message-passing routines can return before message transfer completed

Message buffer needed between source and destination to hold

message:

Slide 51

Asynchronous (blocking) routines changing to synchronous routines

Once local actions completed and message is safely on its way,

sending process can continue with subsequent work.

Buffers only of finite length and a point could be reached when send

routine held up because all available buffer space exhausted.

Then, send routine will wait until storage becomes re-available - i.e

then routine behaves as a synchronous routine.

Slide 52

Message Tag

Used to differentiate between different types of messages being

sent.

Message tag is carried within message.

If special type matching is not required, a wild card message tag is

used, so that the recv() will match with any send().

Slide 53

Process 1 Process 2

send(&x,2,5);

recv(&y,1,5);

x y

Movementof data

Message Tag Example

To send a message, x, with message tag 5 from a source process,

1, to a destination process, 2, and assign to y:

Waits for a message from process 1 with a tag of 5

Slide 54

“Group” message passing routines

Apart from point-to-point message passing routines, have routines

that send message(s) to a group of processes or receive

message(s) from a group of processes - higher efficiency than

separate point-to-point routines although not absolutely necessary.

Slide 55

bcast();

buf

bcast();

data

bcast();

datadata

Process 0 Process n − 1Process 1

Action

Code

Broadcast

Sending same message to all processes concerned with problem.

Multicast - sending same message to defined group of processes.

MPI form

/* Receive data from master */msgtype = 0;pvm_recv(-1, msgtype);pvm_upkint(&nproc, 1, 1);pvm_upkint(tids, nproc, 1);pvm_upkint(&n, 1, 1);pvm_upkint(data, n, 1);

/* Determine my tid */for (i=0; i<nproc; i++)

if(mytid==tids[i]){me = i;break;}

/* Add my portion Of data */x = n/nproc;low = me * x;high = low + x;for(i = low; i < high; i++)

sum += data[i];

/* Send result to master */pvm_initsend(PvmDataDefault);pvm_pkint(&me, 1, 1);pvm_pkint(&sum, 1, 1);msgtype = 5;master = pvm_parent();pvm_send(master, msgtype);

/* Exit PVM */pvm_exit(); return(0);

}

Broadcast data

Receive results

Slide 70

MPI (Message Passing Interface)

Standard developed by group of academics and industrial partners

to foster more widespread use and portability.

Defines routines, not implementation.

Several free implementations exist.

Slide 71

MPI

Process Creation and Execution

Purposely not defined and will depend upon the implementation.

Only static process creation is supported in MPI version 1. All

processes must be defined prior to execution and started together.

Orginally SPMD model of computation.

MPMD also possible with static creation - each program to be

started together specified.

Slide 72

Communicators

Defines scope of a communication operation.

Processes have ranks associated with communicator.

Initially, all processes enrolled in a “universe” called

MPI_COMM_WORLD, and each process is given a unique rank, a

number from 0 to n − 1, where there are n processes.

Other communicators can be established for groups of processes.

Slide 73

Using the SPMD Computational Model

main (int argc, char *argv[]){MPI_Init(&argc, &argv);..MPI_Comm_rank(MPI_COMM_WORLD, &myrank);/*find process rank */if (myrank == 0)

master();else

slave();..MPI_Finalize();}

where master() and slave() are procedures to be executed by

master process and slave process, respectively.

Slide 74

“Unsafe” Message Passing

MPI specifically addresses unsafe message passing.

Slide 75

Unsafe message passing with libraries

lib()

send(…,1,…);

recv(…,0,…);

Process 0 Process 1

send(…,1,…);

recv(…,0,…);(a) Intended behavior

(b) Possible behaviorlib()

lib()

send(…,1,…);

recv(…,0,…);

Process 0 Process 1

send(…,1,…);

recv(…,0,…);

Destination

Source

Slide 76

MPI Solution

“Communicators”

A communication domain that defines a set of processes that are

allowed to communicate between themselves.

The communication domain of the library can be separated from

that of a user program.

Used in all point-to-point and collective MPI message-passing

communications.

Slide 77

Default Communicator

MPI_COMM_WORLD, exists as the first communicator for all the

processes existing in the application.

A set of MPI routines exists for forming communicators.

Processes have a “rank” in a communicator.

Slide 78

Point-to-Point Communication

PVM style packing and unpacking data is generally avoided by the

use of an MPI datatype being defined.

Slide 79

Blocking Routines

Return when they are locally complete - when location used to hold

message can be used again or altered without affecting message

being sent.

A blocking send will send the message and return. This does not

mean that the message has been received, just that the process is

free to move on without adversely affecting the message.

Slide 80

Parameters of the blocking send

MPI_Send(buf, count, datatype, dest, tag, comm)

Address of

Number of items

Datatype of

Rank of destination

Message tag

Communicator

send buffer

to send

each item

process

Slide 81

Parameters of the blocking receive

MPI_Recv(buf, count, datatype, src, tag, comm, status)

Address of

Maximum number

Datatype of

Rank of source

Message tag

Communicator

receive buffer

of items to receive

each item

process

Statusafter operation

Slide 82

Example

To send an integer x from process 0 to process 1,

MPI_Comm_rank(MPI_COMM_WORLD,&myrank); /* find rank */

if (myrank == 0) {int x;MPI_Send(&x, 1, MPI_INT, 1, msgtag, MPI_COMM_WORLD);

} else if (myrank == 1) {int x;MPI_Recv(&x, 1, MPI_INT, 0,msgtag,MPI_COMM_WORLD,status);

}

Slide 83

Nonblocking Routines

Nonblocking send - MPI_Isend(), will return “immediately” even

before source location is safe to be altered.

Nonblocking receive - MPI_Irecv(), will return even if there is no

message to accept.

Slide 84

Nonblocking Routine Formats

MPI_Isend(buf, count, datatype, dest, tag, comm, request)

MPI_Irecv(buf, count, datatype, source, tag, comm, request)

Completion detected by MPI_Wait() and MPI_Test().

MPI_Wait() waits until operation completed and returns then.

MPI_Test() returns with flag set indicating whether operation

completed at that time.

Need to know whether particular operation completed.

Determined by accessing the request parameter.

Slide 85

Example

To send an integer x from process 0 to process 1 and allow process

0 to continue,

MPI_Comm_rank(MPI_COMM_WORLD, &myrank);/* find rank */

if (myrank == 0) {int x;MPI_Isend(&x,1,MPI_INT, 1, msgtag, MPI_COMM_WORLD, req1);compute();MPI_Wait(req1, status);

} else if (myrank == 1) {int x;MPI_Recv(&x,1,MPI_INT,0,msgtag, MPI_COMM_WORLD, status);

}

Slide 86

Four Send Communication Modes

Standard Mode Send Not assumed that corresponding receive routine has started.Amount of buffering not defined by MPI. If buffering provided, sendcould complete before receive reached.

Buffered ModeSend may start and return before a matching receive. Necessary tospecify buffer space via routine MPI_Buffer_attach().

Synchronous ModeSend and receive can start before each other but can only completetogether.

Ready ModeSend can only start if matching receive already reached, otherwiseerror. Use with care.

Slide 87

Each of the four modes can be applied to both blocking and

nonblocking send routines.

Only the standard mode is available for the blocking and

nonblocking receive routines.

Any type of send routine can be used with any type of receive

routine.

Slide 88

Collective Communication

Involves set of processes, defined by an intra-communicator.

Message tags not present.

Broadcast and Scatter Routines

The principal collective operations operating upon data are

MPI_Bcast() - Broadcast from root to all other processesMPI_Gather() - Gather values for group of processesMPI_Scatter() - Scatters buffer in parts to group of processesMPI_Alltoall() - Sends data from all processes to all processesMPI_Reduce() - Combine values on all processes to single valueMPI_Reduce_scatter() - Combine values and scatter resultsMPI_Scan() - Compute prefix reductions of data on processes

Slide 89

Example

To gather items from the group of processes into process 0, using

dynamically allocated memory in the root process, we might use

int data[10]; /*data to be gathered from processes*/.

MPI_Comm_rank(MPI_COMM_WORLD, &myrank); /* find rank */if (myrank == 0) {

MPI_Comm_size(MPI_COMM_WORLD, &grp_size); /*find group size*/buf = (int *)malloc(grp_size*10*sizeof(int));/*allocate memory*/

}MPI_Gather(data,10,MPI_INT,buf,grp_size*10,MPI_INT,0,MPI_COMM_WORLD);

Note that MPI_Gather() gathers from all processes, including root.

Slide 90

Barrier

As in all message-passing systems, MPI provides a means of

synchronizing processes by stopping each one until they all have

reached a specific “barrier” call.

Slide 91

Sample MPI program.#include “mpi.h”#include <stdio.h>#include <math.h>#define MAXSIZE 1000void main(int argc, char *argv){

int myid, numprocs;int data[MAXSIZE], i, x, low, high, myresult, result;char fn[255];char *fp;MPI_Init(&argc,&argv);MPI_Comm_size(MPI_COMM_WORLD,&numprocs);MPI_Comm_rank(MPI_COMM_WORLD,&myid);if (myid == 0) { /* Open input file and initialize data */

strcpy(fn,getenv(“HOME”));strcat(fn,”/MPI/rand_data.txt”);if ((fp = fopen(fn,”r”)) == NULL) {

printf(“Can’t open the input file: %s\n\n”, fn);exit(1);

}for(i = 0; i < MAXSIZE; i++) fscanf(fp,”%d”, &data[i]);

}/* broadcast data */MPI_Bcast(data, MAXSIZE, MPI_INT, 0, MPI_COMM_WORLD);

/* Add my portion Of data */x = n/nproc;low = myid * x;high = low + x;for(i = low; i < high; i++)

myresult += data[i];printf(“I got %d from %d\n”, myresult, myid);

/* Compute global sum */MPI_Reduce(&myresult, &result, 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD);if (myid == 0) printf(“The sum is %d.\n”, result);MPI_Finalize();

}

Slide 92

Process 1

Process 2

Process 3

TimeComputingWaitingMessage-passing system routineMessage

Debugging and Evaluating Parallel ProgramsVisualization Tools

Programs can be watched as they are executed in a space-time

diagram (or process-time diagram):

Slide 93

PVM has a visualization tool called XPVM.

Implementations of visualization tools are available for MPI. An

example is the Upshot program visualization system.

Slide 94

Evaluating Programs EmpiricallyMeasuring Execution Time

To measure the execution time between point L1 and point L2 in the

code, we might have a construction such as

.L1: time(&t1); /* start timer */

.L2: time(&t2); /* stop timer */

.elapsed_time = difftime(t2, t1); /* elapsed_time = t2 - t1 */printf(“Elapsed time = %5.2f seconds”, elapsed_time);

MPI provides the routine MPI_Wtime() for returning time (in

seconds).

Slide 95

Home Page

http://www.cs.unc.edu/par_prog

Slide 96

Basic Instructions for Compiling/Executing PVM Programs

Preliminaries

• Set up paths

• Create required directory structure

• Modify makefile to match your source file

• Create a file (hostfile) listing machines to be used(optional)

Details described on home page.

Slide 97

Compiling/executing PVM programs

Convenient to have two command line windows.To start PVM:At one command line:

pvmreturning a pvm prompt (>) To compile PVM programsAt another command line in pvm3/src/:

aimk fileTo execute PVM programAt same command line in pvm3/bin/?/ (where ? is name of OS)

fileTo terminate pvmAt 1st command line (>):

quit

Slide 98

Basic Instructions for Compiling/Executing MPI Programs

Preliminaries

• Set up paths

• Create required directory structure

• Create a file (hostfile) listing machines to be used(required)

Details described on home page.

Slide 99

Hostfile

Before starting MPI for the first time, need to create a hostfile

Sample hostfile

ws404#is-sm1 //Currently not executing, commentedpvm1 //Active processors, UNCC sun cluster called pvm1 - pvm8pvm2pvm3pvm4pvm5pvm6pvm7pvm8

Slide 100

Compiling/executing (SPMD) MPI program

For LAM MPI version 6.5.2. At a command line:

To start MPI:First time: lamboot -v hostfileSubsequently: lambootTo compile MPI programs:

mpicc -o file file.cor mpiCC -o file file.cppTo execute MPI program:

mpirun -v -np no_processors fileTo remove processes for reboot

lamclean -vTerminate LAM

lamhaltIf fails

wipe -v lamhost

Slide 101

Compiling/Executing Multiple MPI Programs

Create a file specifying programs:

Example

1 master and 2 slaves, “appfile” contains

n0 mastern0-1 slave

To execute:mpirun -v appfile

Sample output3292 master running on n0 (o)3296 slave running on n0 (o)412 slave running on n1

Slide 102

Intentionally blank

Message-Passing Computingpeople.scs.carleton.ca/~achan/teaching/2002-comp... · Message Tag Used to differentiate between different types of messages being sent. Message tag is carried - [PDF Document] (2024)

FAQs

What is the message passing method? ›

Message passing is a method of communication and synchronization between processes in operating systems. It allows processes to exchange data, signals, requests, and responses through a shared medium, such as a message queue, a pipe, a socket, or a shared memory.

Get More Info ›

What is the message passing function? ›

Message passing is a technique for invoking behavior (i.e., running a program) on a computer. In contrast to the traditional technique of calling a program by name, message passing uses an object model to distinguish the general function from the specific implementations.

Keep Reading ›

What is the difference between shared memory and message passing? ›

Message passing should be used if communication needs to be done across different machines or platforms or if data inconsistency or race conditions need to be avoided. Shared memory should be used when communication is happening on the same machine or when high speed and efficiency of data transfer are desired.

Show Me More ›

What is message passing between objects? ›

Message passing in OOP is a mechanism for objects to communicate and interact with each other by sending messages. It involves invoking methods on objects, which can lead to the exchange of information, execution of a specific behaviour, or modification of an object's state.

Learn More ›

What are the disadvantages of message passing? ›

Message passing has some disadvantages compared to shared memory, such as slower and more expensive performance due to data copying and transferring, as well as increased complexity from explicit coding of the message format, content, and logic.

Get More Info ›

How the processes communicate through message passing? ›

In message-passing systems, processes communicate with one another by sending and receiving messages over a communication channel. So how the arrangement should be done? The pattern of the connection provided by the channel is described by some topology systems. The collection of the channels are called a network.

Learn More ›

What is the difference between message passing and system calls? ›

Message passing is a higher level concept of one process sending a message to another. It is implemented by a system ( kernel ) call, asking the kernel to pass the message to the other process. System calls ask the kernel to perform various services for the process.

Learn More ›

Is message passing faster than shared memory? ›

Message Passing is particularly useful in a distributed environment where the communicating processes may reside on different, network connected, systems. Message passing architectures are usually easier to implement but are also usually slower than shared memory architectures.

Read On ›

What is the difference between data binding and message passing? ›

Binding refers to the linking of a procedure call to the code to be executed in response to the call. The process of programming in which communication is involved is known as message passing.

Find Out More ›

What is a real life example of message passing? ›

Message passing is similar to the idea of sending and receiving messages in real life. Suppose we are at a hotel and we want to order room service. For this, we write a message (place an order) to the room service department and put it in the internal messaging system of the hotel.

Is message passing synchronous or asynchronous? ›

Answer: Asynchronous message passing allows the sender to continue execution without waiting for a response, enabling simultaneous processing and separation of components. Synchronous message passing induces a blocking behavior where the sender waits for the receiver's answer.

What is an example of a message passing interface? ›

A popular example is MPI_Send , which allows one specified process to send a message to a second specified process.

What is message passing algorithm? ›

Message passing algorithm which is an iterative decoding algorithm factorizes the global function of many variables into product of simpler local functions, whose arguments are the subset of variables. In order to visualize this factorization we use factor graph.

What is a message passing routine? ›

Basic Message-Passing Routines

The key operations of sending and receiving data are done through message buffers. PVM uses a message tag (msgtag), attached to a message to differentiate between types of messages being sent. Both message tag and source wild cards are available.

What is the principle of message passing programming? ›

The message-passing programming paradigm requires that the parallelism is coded explicitly by the programmer. That is, the programmer is responsible for analyzing the underlying serial algorithm/application and identifying ways by which he or she can decompose the computations and extract concurrency.

Get More Info ›

What are message passing interface methods? ›

MPI Communication Methods

It involves the transfer of a message from one process to a particular process in the same communicator. MPI provides blocking (synchronous) and non-blocking (asynchronous) Point-to-Point communication.

Message-Passing Computingpeople.scs.carleton.ca/~achan/teaching/2002-comp... · Message Tag Used to differentiate between different types of messages being sent. Message tag is carried - [PDF Document] (2024)

FAQs

What is the message passing method? ›

Is message passing synchronous or asynchronous? ›