Computational Chemistry -- fpppp #2b

• Private variables are initialized by each thread

• Iterations are handed out by parallel do

• Schedule handing out of iterations to minimize load imbalance

SUBROUTINE TWLDRV(..., VEE,FXYZ, ...)

C BEGIN LOOPS OVER SHELLS

!$omp PARALLEL

!$omp& shared(VEE, ... ,FXYZ)

!$omp& shared(<about 80 shared>)

!$omp& private(<about 180 private>)

!$omp& private(VEE_LOCAL, ...)

VEE_LOCAL = 0

!$omp do schedule(dynamic, 1)

DO 1001 ISHELL=NSHELL,1,-1

FQ4 = 0

. . . <about 1000 lines of code> . . .

Previous slide

Next slide

Back to first slide

View graphic version