Scribd
Upload
32 views

Openmp Programming: Aiichiro Nakano

The document provides an overview of OpenMP, a programming interface for shared-memory parallel programming using compiler directives. OpenMP uses a fork-join model where the master thread spawns worker threads to perform parallel tasks and then joins the workers upon completion. OpenMP programs use shared memory and synchronization mechanisms to avoid race conditions. Examples are provided to demonstrate calculating pi in parallel using OpenMP.

Uploaded by

Spin Fotonio
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

Openmp Programming: Aiichiro Nakano

The document provides an overview of OpenMP, a programming interface for shared-memory parallel programming using compiler directives. OpenMP uses a fork-join model where the master thread spawns worker threads to perform parallel tasks and then joins the workers upon completion. OpenMP programs use shared memory and synchronization mechanisms to avoid race conditions. Examples are provided to demonstrate calculating pi in parallel using OpenMP.

Uploaded by

Spin Fotonio
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

OpenMP Programming !

Aiichiro Nakano!
Collaboratory for Advanced Computing & Simulations! Department of Computer Science! Department of Physics & Astronomy! Department of Chemical Engineering & Materials Science! University of Southern California! Email: [email protected]!

OpenMP !
!Portable application program interface (API) for shared-memory
!OpenMP = Open specications for Multi Processing! !OpenMP homepage !
! www.openmp.org!

parallel programming based on multi-threading by compiler directives!

!OpenMP tutorial !
! www.llnl.gov/computing/tutorials/openMP!

!Process: an instance of program running!


!Thread: a sequence of instructions being executed, possibly sharing resources with other threads within a process!
send write receive read

share

MPI (distributed memory)!

OpenMP (shared memory)!

OpenMP Programming Model !


Fork-join parallelism ! !Fork: master thread spawns a team of threads as needed! !Join: when the team of threads complete the statements in the parallel section, they terminate synchronously, leaving only the master thread !

!OpenMP threads communicate by sharing variables!

OpenMP Example: omp_example.c !


#include <stdio.h>! #include <omp.h>! void main () {! !int nthreads,tid;! !nthreads = omp_get_num_threads();! !printf("Sequential section: # of threads = %d\n",nthreads);! !/* Fork multi-threads with own copies of variable */! !#pragma omp parallel private(tid)! !{! ! !/* Obtain & print thread id */! ! !tid = omp_get_thread_num();! ! !printf("Parallel section: Hello world from thread %d\n",tid); ! ! !/* Only master thread does this */! ! !if (tid == 0) {! ! ! !nthreads = omp_get_num_threads();! ! ! !printf("Parallel section: # of threads = %d\n",nthreads);}! !} /* All created threads terminate */! }!

!Obtain the number of threads & my thread ID! !By default, all variables are shared unless selectively changing storage attributes using private clauses !

parallel section !

OpenMP Example: omp_example.c !


!Compilation on hpc-login2.usc.edu! !source /usr/usc/mpich/default/mx-intel/setup.csh!
!icc -o omp_example omp_example.c -openmp!

!PBS script!
!#!/bin/bash! !#PBS -l nodes=1:ppn=2,arch=x86_64! !#PBS -l walltime=00:00:59! Set the # of threads !#PBS -o omp_example.out! !#PBS -j oe! using environment !#PBS -N omp_example! parameter! !OMP_NUM_THREADS=2! !WORK_HOME=/home/rcf-proj2/an/anakano/hpc/cs596! !cd $WORK_HOME! !./omp_example!

!Output!
!Sequential section: # of threads = 1! !Parallel section: Hello world from thread 1! !Parallel section: Hello world from thread 0! !Parallel section: # of threads = 2!

Setting the Number of Threads !


#include <stdio.h>! #include <omp.h>! void main () {! !int nthreads,tid;! !omp_set_num_threads(2);! !nthreads = omp_get_num_threads();! !printf("Sequential section: # of threads = %d\n",nthreads);! !/* Fork multi-threads with own copies of variable */! !#pragma omp parallel private(tid) ! ! {! ! !/* Obtain & print thread id */! ! !tid = omp_get_thread_num();! ! !printf("Parallel section: Hello world from thread %d\n",tid);! ! !/* Only master thread does this */! ! !if (tid == 0) {! ! ! !nthreads = omp_get_num_threads();! ! ! !printf("Parallel section: # of threads = %d\n",nthreads);! ! !}! !} /* All created threads terminate */! }!

!Setting the number of threads to be used in parallel sections within the program (no need to set OMP_NUM_THREADS); see omp_example_set.c!

OpenMP Programming Model !


!OpenMP is typically used to parallelize loops! !Use synchronization mechanisms to avoid race conditions (i.e., the result changes for different thread schedules)! !Critical section: only one thread at a time can enter !
#pragma omp parallel! {! !...! !#pragma omp critical! !{! ! !...! !} " ...! }!

Threads wait their turn only one at a time executes the critical section!

Example: Calculating ! !
!Numerical integration! 1 4 dx = # "0 2 1+ x !Discretization:! !" = 1/N: step = 1/NBIN! !xi = (i+0.5)" (i = 0,,N-1)!
N "1 i =01 +

4 xi2

$ %&

#include <stdio.h>! #define NBIN 100000! void main() {! !int i; double step,x,sum=0.0,pi;! !step = 1.0/NBIN;! !for (i=0; i<NBIN; i++) {! ! !x = (i+0.5)*step;! ! !sum += 4.0/(1.0+x*x);}! !pi = sum*step;! !printf( PI = %f\n ,pi);! }!

OpenMP Program: omp_pi_critical.c !


#include <stdio.h>! #include <omp.h>! #define NBIN 100000! void main() {! !double step,sum=0.0,pi;! !step = 1.0/NBIN;! !# pragma omp parallel! ! {! ! !int nthreads,tid,i;! ! !double x;! ! !nthreads = omp_get_num_threads();! ! !tid = omp_get_thread_num();! ! !for (i=tid; i<NBIN; i+=nthreads) { ! ! ! !x = (i+0.5)*step;! ! ! !#pragma omp critical! ! ! !sum += 4.0/(1.0+x*x);! ! !}! ! }! !pi = sum*step;! !printf("PI = %f\n",pi);! }!

Shared variables! Private (local) variables! This has to be atomic!

Thread-private variables: Either declare private or dene within a parallel section!

Avoid Critical Section: omp_pi.c !


Data privatization!
#include <stdio.h>! #include <omp.h>! #define NBIN 100000! #define MAX_THREADS 8! void main() {! !int nthreads,tid;! !double step,sum[MAX_THREADS]={0.0},pi=0.0;! !step = 1.0/NBIN;! !#pragma omp parallel private(tid)! Array of partial sums! ! {! for multi-threads! ! !int i;! ! !double x;! ! !nthreads = omp_get_num_threads();! ! !tid = omp_get_thread_num();! ! !for (i=tid; i<NBIN; i+=nthreads) {! ! ! !x = (i+0.5)*step;! ! ! !sum[tid] += 4.0/(1.0+x*x);! ! !}! ! }! !for(tid=0; tid<nthreads; tid++) pi += sum[tid]*step;! !printf("PI = %f\n",pi);! }!

You might also like