Good Reductions

Correct reduction:

c$omp parallel private(suml,i)

c$omp& shared(sum,a,n)

suml = 0.0

c$omp do

do 10 i=1,n

10 suml = suml + a(i)

c$omp critical

sum = sum + suml

c$omp end critical

c$omp end parallel

Using Reduction does the same:

c$omp parallel private(i)

c$omp& shared(a,n)

c$omp& reduction(+:sum)

c$omp do

do 10 i=1,n

10 sum = sum + a(i)

c$omp end parallel

Previous slide

Next slide

Back to first slide

View graphic version