Performance of a Code Migration for the Simulation of Supersonic Ejector Flow to SMP, MIC, and GPU Using OpenMP, OpenMP+LEO, and OpenACC Directives

Listing 10

Format of the OpenACC directives used to parallelize the loops.

!$acc data &

!$acc copyin(deltae,a_x,b_x,K_x)  &

!acc local(DF1Dn,DF2Dn,DF3Dn,DF4Dn)  &

!WHILE OF TIME

!$acc kernels

 !$acc loop independent

 DO j=2,(nj-1)

     !$acc loop independent

     DO i=2,(ni-1)

     END DO

 END DO

!$acc end kernels

IF  (MOD(k,it_display) == 0) THEN

!$acc update host(U,V,ro,P,T,M,ET) async

CALL LAYER_WRITE(DBLE(k))

END IF

END DO ! end of while in time

!END OF WHILE

!$acc end data