Parallel Computing for Data Science

120 CHAPTER 5. SHARED-MEMORY: C

5.2.3 Analysis

Now, take a look at burst():

void b urst ( double ∗x , int nx , int k ,

int ∗ startmax , int ∗endmax , double ∗ maxval )

{

int nth ; // number of thre a d s

#pragma omp p a r a l l e l

{ int p e r s t a r t , // START OF PARALLEL BLOCK

perlen , // p er i od len g t h

. . .

. . .

. . .

∗ startmax = mystartmax ;

∗endmax = myendmax ;

}

}

} // END OF PARALLEL BLOCK

This is really the crux of OpenMP. Note the pragma:

#pragma omp parallel

This instruction to the compiler unleashes a team of threads. Each of the

threads will execute the block that follows,

2

with certain rules governing

the local variables:

Consider the variable ...