Changeset 1257 for palm/trunk/SOURCE/pres.f90

Timestamp:

Nov 8, 2013 3:18:40 PM (10 years ago)

Author:

raasch

Message:

New:
---

openACC porting of timestep calculation
(modules, timestep, time_integration)

Changed:

---

openACC loop directives and vector clauses removed (because they do not give any performance improvement with PGI
compiler versions > 13.6)
(advec_ws, buoyancy, coriolis, diffusion_e, diffusion_s, diffusion_u, diffusion_v, diffusion_w, diffusivities, exchange_horiz, fft_xy, pres, production_e, transpose, tridia_solver, wall_fluxes)

openACC loop independent clauses added
(boundary_conds, prandtl_fluxes, pres)

openACC declare create statements moved after FORTRAN declaration statement
(diffusion_u, diffusion_v, diffusion_w, fft_xy, poisfft, production_e, tridia_solver)

openACC end parallel replaced by end parallel loop
(flow_statistics, pres)

openACC "kernels do" replaced by "kernels loop"
(prandtl_fluxes)

output format for theta* changed to avoid output of *
(run_control)

Errors:

---

bugfix for calculation of advective timestep (old version may cause wrong timesteps in case of
vertixcally stretched grids)
Attention: standard run-control output has changed!
(timestep)

File:

• palm/trunk/SOURCE/pres.f90 (modified) (10 diffs)

palm/trunk/SOURCE/pres.f90

@@ -20 +20 @@
 ! Current revisions:
 ! ------------------
-!
+! openacc loop and loop vector clauses removed, independent clauses added,
+! end parallel replaced by end parallel loop
 !
 ! Former revisions:
@@ -321 +322 @@
        !$OMP PARALLEL DO SCHEDULE( STATIC )
        !$acc kernels present( d )
-       !$acc loop
        DO  i = nxl, nxr
           DO  j = nys, nyn
-             !$acc loop vector(32)
              DO  k = nzb+1, nzt
                 d(k,j,i) = 0.0
@@ -391 +390 @@
        ENDDO
     ENDDO
-    !$acc end parallel
+    !$acc end parallel loop
     localsum = localsum + threadsum * dt_3d * &
                           weight_pres(intermediate_timestep_count)
@@ -421 +420 @@
        !$OMP PARALLEL DO
        !$acc kernels present( d, tend )
-       !$acc loop
        DO  i = nxl, nxr
           DO  j = nys, nyn
-             !$acc loop vector( 32 )
              DO  k = nzb+1, nzt
                 tend(k,j,i) = d(k,j,i)
@@ -558 +555 @@
        !$OMP DO
        !$acc kernels present( p, tend, weight_substep )
-       !$acc loop
+       !$acc loop independent
        DO  i = nxl-1, nxr+1
+          !$acc loop independent
           DO  j = nys-1, nyn+1
-             !$acc loop vector( 32 )
+             !$acc loop independent
              DO  k = nzb, nzt+1
                 p(k,j,i) = tend(k,j,i) * &
@@ -575 +573 @@
        !$OMP DO
        !$acc kernels present( p, tend, weight_substep )
-       !$acc loop
+       !$acc loop independent
        DO  i = nxl-1, nxr+1
+          !$acc loop independent
           DO  j = nys-1, nyn+1
-             !$acc loop vector( 32 )
+             !$acc loop independent
              DO  k = nzb, nzt+1
                 p(k,j,i) = p(k,j,i) + tend(k,j,i) * &
@@ -605 +604 @@
     !$OMP DO
     !$acc kernels present( ddzu, nzb_u_inner, nzb_v_inner, nzb_w_inner, tend, u, v, w, weight_pres )
-    !$acc loop
+    !$acc loop independent
     DO  i = nxl, nxr   
-       DO  j = nys, nyn
-          !$acc loop vector( 32 )
+       !$acc loop independent
+       DO  j = nys, nyn
+          !$acc loop independent
           DO  k = 1, nzt
              IF ( k > nzb_w_inner(j,i) )  THEN
@@ -616 +616 @@
              ENDIF
           ENDDO
-          !$acc loop vector( 32 )
+          !$acc loop independent
           DO  k = 1, nzt
              IF ( k > nzb_u_inner(j,i) )  THEN
@@ -624 +624 @@
              ENDIF
           ENDDO
-          !$acc loop vector( 32 )
+          !$acc loop independent
           DO  k = 1, nzt
              IF ( k > nzb_v_inner(j,i) )  THEN
@@ -773 +773 @@
        ENDDO
     ENDDO
-    !$acc end parallel
+    !$acc end parallel loop
 #endif