Intel parallel programming

 Dharmendra Savaliya
 { Subham Yadav
 Bhaumik Patel
 Darshak Shah}

 How to make Processors always busy?
 Fork join method For OpenMP.
 What is MPI?

 STUDY PROBLEM, SEQUENTIAL PROGRAM
 LOOK FOR OPPORTUNITIES FOR PARALLELISM
 TRY TO KEEP ALL PROCESSORS BUSY DOING USEFUL WORK

 DOMAIN DECOMPOSITION
 TASK DECOMPOSITION
 Dependence Graph

First, decide how
data elements
should be divided
among processors.
Second, decide
which tasks each
processor should be
doing
Intel White paper : www.intel.com/pressroom/archive/reference/

First, divide tasks
among processors
Second, decide which
data elements are
going to be accessed
(read and/or written)
by which
processors

Special kind of task decomposition

1. for (i = 0; i < 3; i++)
a[i] = b[i] / 2.0;

2. for (i = 1; i < 4; i++)
a[i] = a[i-1] * b[i];

3. a = f(x, y, z);
b = g(w, x);
t = a + b;
c = h(z);
s = t / c;

 OpenMP is an API for parallel programming
 First developed by the OpenMP Architecture Review
in 1997, a standard Designed for shared-memory
multiprocessors
 Set of compiler directives, library functions, and
environment variables, but not a language
 Can be used with C, C++, or Fortran
 Based on fork/join model of threads

Strengths
 Well-suited for Domain decompositions
 Available on Unix and Windows
Weaknesses
 Not good for Task decompositions
 Race condition due to dependancy
16Implementing Domain Decompositions

The compiler directive
 #pragma omp parallel for
 tells the compiler that the for loop which
immediately follows can be executed in parallel
 The number of loop iterations must be computable at
run time before loop executes
 Loop must not contain a break, return, or exit
 Loop must not contain a goto to a label outside loop

 The hello World OpenMP Prog.
#include <stdio.h>
#include <omp.h>
int main()
{
#pragma omp parallel
printf("Hello World n");
return 0;
}
 Since fork/join is a source of loop, we
want to maximize the amount of work
done for each fork/join
Master
Master
W0 W1 W2

 Message Passing interface
 Used on Distributed memory MIMD architectures
• MPI specifies the API for message passing
(communication related routines)
 Distributed Memory
Processes have only local memory and must use
some other mechanism (e.g., message passing) to
exchange information.
 Advantage: programmers have explicit control
over data distribution and communication

#include "mpi.h"
#include <stdio.h>
int main( int argc, char *argv[] )
{
MPI_Init( &argc, &argv );
printf( "Hello worldn" );
MPI_Finalize();
return 0;
}
• Mpi.h contains MPI
definitioins and types.
• MPI program must start
with MPI_init
• MPI program must exit
with MPI_Finalize
• MPI functions are just
library routines that can be
used on top of the regular
C, C++

 MPI_Send(start, count, datatype, dest, tag, comm)
 MPI_Recv(start, count, datatype, source, tag,
comm, status)
 The Simple MPI (six functions that make most of programs work):
 MPI_INIT
 MPI_FINALIZE
 MPI_COMM_SIZE
 MPI_COMM_RANK
 MPI_SEND
 MPI_RECV

 MPI OpenMP
 Distributed memory Shared memory model
model
 Distributed network on Multi-core processors
 Message based Directive based
 Flexible and expressive Easier to program and
debug

If, We use OpenMP and MPI For C Programming
But in this case performance is less then MPI.
So, Better to use OpenMP or MPI Separately.

 http://www.intel-software-academic-
program.com/pages/courses

Intel parallel programming

More Related Content

What's hot

Viewers also liked

Similar to Intel parallel programming

Recently uploaded

Intel parallel programming