1.A.2 Parallel Loop Model

• Note threads, shared, local

• program
  
  c$omp parallel do
  c$omp& shared(A) private(I)
  do I=1,100
  . . .
  enddo
  c$omp end parallel
  end

A is shared

Parallel Do

I=1

I=2

I=3

I=4

I=5

I=6

A

Several, e.g. 4, threads

created on entry

Iterations

scheduled

on threads

I is local

Threads

barrier at end

Threads either spin or

sleep between regions

Previous slide

Next slide

Back to first slide

View graphic version