From 797a64ead43ae9bf60fffe451a90469b7b211b81 Mon Sep 17 00:00:00 2001
From: Juan ESCOBAR <juan.escobar@aero.obs-mip.fr>
Date: Fri, 15 Apr 2022 19:01:35 +0200
Subject: [PATCH] Juan 15/04/2022:MNH/ZSOLVER/turb_hor_thermo_flux.f90, Cray
GPU Opt/Bug bypass, add present_cr + acc_nv , where needeed
---
src/MNH/turb_hor_thermo_flux.f90 | 191 +++++---------
src/ZSOLVER/turb_hor_thermo_flux.f90 | 374 ++++++---------------------
2 files changed, 155 insertions(+), 410 deletions(-)
diff --git a/src/MNH/turb_hor_thermo_flux.f90 b/src/MNH/turb_hor_thermo_flux.f90
index 4e0bd16f6..52cff5188 100644
--- a/src/MNH/turb_hor_thermo_flux.f90
+++ b/src/MNH/turb_hor_thermo_flux.f90
@@ -332,9 +332,7 @@ ZFLX(:,:,IKE+1) = ZFLX(:,:,IKE)
CALL MXM_DEVICE( PK, ZTMP1_DEVICE )
CALL GX_M_U_DEVICE(1,IKU,1,PTHLM,PDXX,PDZZ,PDZX,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCSHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -355,18 +353,18 @@ ZFLX(:,:,IKB:IKB) = -XCSHF * MXM( PK(:,:,IKB:IKB) ) * &
#else
CALL MXM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DXM_DEVICE( PTHLM(:,:,IKB:IKB), ZTMP2_DEVICE(:,:,1:1) )
-!$acc kernels
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PTHLM(:,:,IKB+2) &
+ZCOEFF(:,:,IKB+1)*PTHLM(:,:,IKB+1) &
+ZCOEFF(:,:,IKB )*PTHLM(:,:,IKB )
!$acc end kernels
CALL MXM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1))
-!$acc kernels
+!$acc kernels present_cr(ZFLX,ZTMP1_DEVICE)
ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDXX(:,:,IKB) - ZTMP4_DEVICE(:,:,1) &
*0.5* ( PDZX(:,:,IKB+1)+PDZX(:,:,IKB)) &
* PINV_PDXX(:,:,IKB) )
-!$acc end kernels
+! acc end kernels
#endif
! extrapolates the flux under the ground so that the vertical average with
! the IKB flux gives the ground value ( warning the tangential surface
@@ -375,17 +373,13 @@ ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
ZFLX(:,:,IKB-1:IKB-1) = 2. * MXM( SPREAD( PSFTHM(:,:)* PDIRCOSXW(:,:), 3,1) ) &
- ZFLX(:,:,IKB:IKB)
#else
-!$acc kernels
-!PW: BUG: commented 'acc loop independent collapse(2)' to workaround compiler bug (NVHPC 21.1)
-#ifdef MNH_COMPILER_NVHPC
-!acc loop independent collapse(2)
-#endif
+! acc kernels
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU )
ZTMP1_DEVICE(JI,JJ,1) = PSFTHM(JI,JJ)* PDIRCOSXW(JI,JJ)
END DO
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
-!$acc kernels
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
#endif
@@ -404,36 +398,28 @@ END IF
IF (.NOT. LFLAT) THEN
CALL MXM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL DXF_DEVICE(ZTMP2_DEVICE, ZTMP3_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PDZX(JI,JJ,JK)*ZTMP4_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MXF_DEVICE(ZTMP2_DEVICE, ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP4_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
@@ -505,23 +491,23 @@ IF ( KRRL >= 1 ) THEN
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:) *ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE)
- !$acc kernels
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = PDZX(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP6_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP7_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = 2.*( ZTMP2_DEVICE(:,:,:) +ZTMP7_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
@@ -529,11 +515,11 @@ IF ( KRRL >= 1 ) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PATHETA(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL DXF_DEVICE( ZTMP6_DEVICE, ZTMP2_DEVICE)
- !$acc kernels
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
@@ -636,7 +622,7 @@ IF (KSPLT==1 .AND. LLES_CALL) THEN
ZTMP3_DEVICE(:,:,:) = ZTMP1_DEVICE(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE(ZTMP3_DEVICE,ZTMP1_DEVICE)
- CALL MZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
+ CALL MZF_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
CALL LES_MEAN_SUBGRID( ZTMP2_DEVICE,X_LES_RES_ddxa_W_SBG_UaThl , .TRUE. )
!
CALL GX_M_M_DEVICE(PTHLM,PDXX,PDZZ,PDZX,ZTMP1_DEVICE)
@@ -779,9 +765,7 @@ END IF
CALL MXM_DEVICE( PK, ZTMP1_DEVICE )
CALL GX_M_U_DEVICE(1,IKU,1,PRM(:,:,:,1),PDXX,PDZZ,PDZX,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -792,13 +776,13 @@ END DO
! gradient
CALL MXM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DXM_DEVICE(PRM(:,:,IKB:IKB,1), ZTMP2_DEVICE(:,:,1:1))
-!$acc kernels
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PRM(:,:,IKB+2,1) &
+ZCOEFF(:,:,IKB+1)*PRM(:,:,IKB+1,1) &
+ZCOEFF(:,:,IKB )*PRM(:,:,IKB ,1)
!$acc end kernels
CALL MXM_DEVICE(ZTMP3_DEVICE(:,:,1:1),ZTMP4_DEVICE(:,:,1:1))
-!$acc kernels
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDXX(:,:,IKB) &
-ZTMP4_DEVICE(:,:,1) &
@@ -810,7 +794,7 @@ END DO
ZTMP1_DEVICE(:,:,1) = PSFRM(:,:)* PDIRCOSXW(:,:)
!$acc end kernels
CALL MXM_DEVICE(ZTMP1_DEVICE(:,:,1:1),ZTMP2_DEVICE(:,:,1:1))
- !$acc kernels
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
@@ -820,45 +804,35 @@ END DO
IF (.NOT. LFLAT) THEN
CALL MXM_DEVICE(PRHODJ,ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL DXF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PDZX(JI,JJ,JK)*ZTMP4_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MXF_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP4_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
!$acc end kernels
- CALL DZF_DEVICE( ZTMP2_DEVICE, ZTMP4_DEVICE )
+ CALL DZF_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRRS(JI,JJ,JK,1) = PRRS(JI,JJ,JK,1) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP4_DEVICE(JI,JJ,JK)
END DO
@@ -877,36 +851,33 @@ END DO
! Compute the equivalent tendancy for Rc and Ri
!
IF ( KRRL >= 1 ) THEN
- !$acc kernels
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:)
ZTMP2_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP8_DEVICE )
IF (.NOT. LFLAT) THEN
- !$acc kernels
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP8_DEVICE(:,:,:) * ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP4_DEVICE, ZTMP3_DEVICE )
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
CALL MZM_DEVICE( ZTMP2_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = PDZX(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP6_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP7_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZFLXC,ZTMP6_DEVICE)
ZFLXC(:,:,:) = ZFLXC(:,:,:) + 2.*( ZTMP3_DEVICE(:,:,:) + ZTMP7_DEVICE(:,:,:) )
- !$acc end kernels
- !
!
- !$acc kernels
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP6_DEVICE, ZTMP3_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP8_DEVICE(:,:,:) * ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL DXF_DEVICE(ZTMP4_DEVICE, ZTMP5_DEVICE)
@@ -1169,9 +1140,7 @@ END IF
CALL MYM_DEVICE( PK, ZTMP1_DEVICE )
CALL GY_M_V_DEVICE(1,IKU,1,PTHLM,PDYY,PDZZ,PDZY,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCSHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1188,13 +1157,13 @@ END IF
! gradient
CALL MYM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DYM_DEVICE(PTHLM(:,:,IKB:IKB), ZTMP2_DEVICE(:,:,1:1) )
-!$acc kernels
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PTHLM(:,:,IKB+2) &
+ZCOEFF(:,:,IKB+1)*PTHLM(:,:,IKB+1) &
+ZCOEFF(:,:,IKB )*PTHLM(:,:,IKB )
!$acc end kernels
CALL MYM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1) )
-!$acc kernels
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDYY(:,:,IKB) &
- ZTMP4_DEVICE(:,:,1) &
@@ -1208,7 +1177,7 @@ ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
ZTMP1_DEVICE(:,:,1) = PSFTHM(:,:)* PDIRCOSYW(:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
-!$acc kernels
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
!
@@ -1218,45 +1187,35 @@ IF (.NOT. L2D) THEN
IF (.NOT. LFLAT) THEN
CALL MYM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP1_DEVICE, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PDZY(JI,JJ,JK)*ZTMP2_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MYF_DEVICE(ZTMP1_DEVICE, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
!$acc end kernels
CALL DZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRTHLS(JI,JJ,JK) = PRTHLS(JI,JJ,JK) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1282,32 +1241,32 @@ IF ( KRRL >= 1 .AND. .NOT. L2D) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PATHETA(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP4_DEVICE, ZTMP3_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP4_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = PDZY(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP4_DEVICE, ZTMP5_DEVICE)
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP4_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = 2.*( ZTMP3_DEVICE(:,:,:) + ZTMP4_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
- !$acc kernels
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL DYF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP6_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP3_DEVICE, ZTMP5_DEVICE )
@@ -1441,7 +1400,7 @@ IF (KRR/=0) THEN
+ZCOEFF(:,:,IKB :IKB )*PRM(:,:,IKB :IKB ,1) ) &
*0.5* ( PDZY(:,:,IKB+1:IKB+1)+PDZY(:,:,IKB:IKB)) &
* PINV_PDYY(:,:,IKB:IKB) )
-! extrapolates the flux under the ground so that the vertical average with
+! extrapolates the flux under the ground so that the vertical average with
! the IKB flux gives the ground value ( warning the tangential surface
! flux has been set to 0 for the moment !! to be improved )
ZFLX(:,:,IKB-1:IKB-1) = 2. * MYM( SPREAD( PSFRM(:,:)* PDIRCOSYW(:,:), 3,1) ) &
@@ -1548,9 +1507,7 @@ IF (KRR/=0) THEN
CALL MYM_DEVICE( PK, ZTMP1_DEVICE )
CALL GY_M_V_DEVICE(1,IKU,1,PRM(:,:,:,1),PDYY,PDZZ,PDZY, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO !CONCURRENT
@@ -1566,39 +1523,37 @@ IF (KRR/=0) THEN
! gradient
CALL MYM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DYM_DEVICE(PRM(:,:,IKB:IKB,1), ZTMP2_DEVICE(:,:,1:1))
- !$acc kernels
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PRM(:,:,IKB+2,1) &
+ZCOEFF(:,:,IKB+1)*PRM(:,:,IKB+1,1) &
+ZCOEFF(:,:,IKB )*PRM(:,:,IKB ,1)
!$acc end kernels
CALL MYM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1) )
- !$acc kernels
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDYY(:,:,IKB) &
- ZTMP4_DEVICE(:,:,1) &
*0.5* ( PDZY(:,:,IKB+1)+PDZY(:,:,IKB)) &
* PINV_PDYY(:,:,IKB) )
!$acc end kernels
-! extrapolates the flux under the ground so that the vertical average with
+! extrapolates the flux under the ground so that the vertical average with
! the IKB flux gives the ground value ( warning the tangential surface
! flux has been set to 0 for the moment !! to be improved )
!$acc kernels
ZTMP1_DEVICE(:,:,1) = PSFRM(:,:)* PDIRCOSYW(:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
- !$acc kernels
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
-!
+ !
! Add this source to the conservative mixing ratio sources
!
IF (.NOT. L2D) THEN
IF (.NOT. LFLAT) THEN
CALL MYM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
@@ -1606,27 +1561,21 @@ IF (KRR/=0) THEN
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PDZY(JI,JJ,JK)*ZTMP2_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MYF_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
@@ -1634,9 +1583,7 @@ IF (KRR/=0) THEN
CALL DZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
!
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRRS(JI,JJ,JK,1) = PRRS(JI,JJ,JK,1) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1661,32 +1608,32 @@ IF (KRR/=0) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP5_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = PDZY(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZTMP5_DEVICE(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = ZFLXC(:,:,:) + 2.*( ZTMP4_DEVICE(:,:,:) + ZTMP2_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
- !$acc kernels
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP2_DEVICE(:,:,:) = ZTMP3_DEVICE(:,:,:) * PINV_PDYY(:,:,:)
!$acc end kernels
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
- !$acc kernels
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP2_DEVICE(:,:,:) = ZTMP1_DEVICE(:,:,:)* PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP2_DEVICE, ZTMP4_DEVICE )
diff --git a/src/ZSOLVER/turb_hor_thermo_flux.f90 b/src/ZSOLVER/turb_hor_thermo_flux.f90
index a767a3163..52cff5188 100644
--- a/src/ZSOLVER/turb_hor_thermo_flux.f90
+++ b/src/ZSOLVER/turb_hor_thermo_flux.f90
@@ -332,9 +332,7 @@ ZFLX(:,:,IKE+1) = ZFLX(:,:,IKE)
CALL MXM_DEVICE( PK, ZTMP1_DEVICE )
CALL GX_M_U_DEVICE(1,IKU,1,PTHLM,PDXX,PDZZ,PDZX,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCSHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -355,21 +353,13 @@ ZFLX(:,:,IKB:IKB) = -XCSHF * MXM( PK(:,:,IKB:IKB) ) * &
#else
CALL MXM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DXM_DEVICE( PTHLM(:,:,IKB:IKB), ZTMP2_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZTMP3_DEVICE)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PTHLM(:,:,IKB+2) &
+ZCOEFF(:,:,IKB+1)*PTHLM(:,:,IKB+1) &
+ZCOEFF(:,:,IKB )*PTHLM(:,:,IKB )
!$acc end kernels
CALL MXM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1))
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZFLX)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZFLX,ZTMP1_DEVICE)
ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDXX(:,:,IKB) - ZTMP4_DEVICE(:,:,1) &
*0.5* ( PDZX(:,:,IKB+1)+PDZX(:,:,IKB)) &
@@ -389,11 +379,7 @@ DO CONCURRENT ( JI=1:JIU,JJ=1:JJU )
END DO
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZFLX)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
#endif
@@ -412,36 +398,28 @@ END IF
IF (.NOT. LFLAT) THEN
CALL MXM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL DXF_DEVICE(ZTMP2_DEVICE, ZTMP3_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PDZX(JI,JJ,JK)*ZTMP4_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MXF_DEVICE(ZTMP2_DEVICE, ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP4_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
@@ -513,43 +491,23 @@ IF ( KRRL >= 1 ) THEN
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP1_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:) *ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE)
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP1_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = PDZX(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP6_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP7_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLXC)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = 2.*( ZTMP2_DEVICE(:,:,:) +ZTMP7_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
@@ -557,19 +515,11 @@ IF ( KRRL >= 1 ) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PATHETA(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL DXF_DEVICE( ZTMP6_DEVICE, ZTMP2_DEVICE)
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP3_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
@@ -672,7 +622,7 @@ IF (KSPLT==1 .AND. LLES_CALL) THEN
ZTMP3_DEVICE(:,:,:) = ZTMP1_DEVICE(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE(ZTMP3_DEVICE,ZTMP1_DEVICE)
- CALL MZF_DEVICE( ZTMP1_DEVICE,ZTMP2_DEVICE)
+ CALL MZF_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
CALL LES_MEAN_SUBGRID( ZTMP2_DEVICE,X_LES_RES_ddxa_W_SBG_UaThl , .TRUE. )
!
CALL GX_M_M_DEVICE(PTHLM,PDXX,PDZZ,PDZX,ZTMP1_DEVICE)
@@ -815,9 +765,7 @@ END IF
CALL MXM_DEVICE( PK, ZTMP1_DEVICE )
CALL GX_M_U_DEVICE(1,IKU,1,PRM(:,:,:,1),PDXX,PDZZ,PDZX,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
-!$acc loop independent collapse(3)
-#endif
+!$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -828,21 +776,13 @@ END DO
! gradient
CALL MXM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DXM_DEVICE(PRM(:,:,IKB:IKB,1), ZTMP2_DEVICE(:,:,1:1))
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZTMP3_DEVICE)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PRM(:,:,IKB+2,1) &
+ZCOEFF(:,:,IKB+1)*PRM(:,:,IKB+1,1) &
+ZCOEFF(:,:,IKB )*PRM(:,:,IKB ,1)
!$acc end kernels
CALL MXM_DEVICE(ZTMP3_DEVICE(:,:,1:1),ZTMP4_DEVICE(:,:,1:1))
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZFLX)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDXX(:,:,IKB) &
-ZTMP4_DEVICE(:,:,1) &
@@ -854,11 +794,7 @@ END DO
ZTMP1_DEVICE(:,:,1) = PSFRM(:,:)* PDIRCOSXW(:,:)
!$acc end kernels
CALL MXM_DEVICE(ZTMP1_DEVICE(:,:,1:1),ZTMP2_DEVICE(:,:,1:1))
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLX)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
@@ -868,45 +804,35 @@ END DO
IF (.NOT. LFLAT) THEN
CALL MXM_DEVICE(PRHODJ,ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL DXF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDXX(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PDZX(JI,JJ,JK)*ZTMP4_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MXF_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP4_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
!$acc end kernels
- CALL DZF_DEVICE( ZTMP2_DEVICE, ZTMP4_DEVICE)
+ CALL DZF_DEVICE(ZTMP2_DEVICE,ZTMP4_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRRS(JI,JJ,JK,1) = PRRS(JI,JJ,JK,1) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP4_DEVICE(JI,JJ,JK)
END DO
@@ -925,57 +851,33 @@ END DO
! Compute the equivalent tendancy for Rc and Ri
!
IF ( KRRL >= 1 ) THEN
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP2_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:)
ZTMP2_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL MXM_DEVICE( ZTMP1_DEVICE, ZTMP8_DEVICE )
IF (.NOT. LFLAT) THEN
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP4_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP8_DEVICE(:,:,:) * ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP4_DEVICE, ZTMP3_DEVICE )
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
CALL MZM_DEVICE( ZTMP2_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = PDZX(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP6_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP7_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLXC,ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLXC,ZTMP6_DEVICE)
ZFLXC(:,:,:) = ZFLXC(:,:,:) + 2.*( ZTMP3_DEVICE(:,:,:) + ZTMP7_DEVICE(:,:,:) )
!
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP6_DEVICE, ZTMP3_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP4_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP8_DEVICE(:,:,:) * ZFLX(:,:,:)*PINV_PDXX(:,:,:)
!$acc end kernels
CALL DXF_DEVICE(ZTMP4_DEVICE, ZTMP5_DEVICE)
@@ -1238,9 +1140,7 @@ END IF
CALL MYM_DEVICE( PK, ZTMP1_DEVICE )
CALL GY_M_V_DEVICE(1,IKU,1,PTHLM,PDYY,PDZZ,PDZY,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCSHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1257,21 +1157,13 @@ END IF
! gradient
CALL MYM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DYM_DEVICE(PTHLM(:,:,IKB:IKB), ZTMP2_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZTMP3_DEVICE)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PTHLM(:,:,IKB+2) &
+ZCOEFF(:,:,IKB+1)*PTHLM(:,:,IKB+1) &
+ZCOEFF(:,:,IKB )*PTHLM(:,:,IKB )
!$acc end kernels
CALL MYM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZFLX)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDYY(:,:,IKB) &
- ZTMP4_DEVICE(:,:,1) &
@@ -1285,11 +1177,7 @@ ZFLX(:,:,IKB) = -XCSHF * ZTMP1_DEVICE(:,:,1) * &
ZTMP1_DEVICE(:,:,1) = PSFTHM(:,:)* PDIRCOSYW(:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
-!$acc kernels present(ZFLX)
-#else
-!$acc kernels
-#endif
+!$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
!
@@ -1299,45 +1187,35 @@ IF (.NOT. L2D) THEN
IF (.NOT. LFLAT) THEN
CALL MYM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP1_DEVICE, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PDZY(JI,JJ,JK)*ZTMP2_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MYF_DEVICE(ZTMP1_DEVICE, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
!$acc end kernels
CALL DZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRTHLS(JI,JJ,JK) = PRTHLS(JI,JJ,JK) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1363,60 +1241,32 @@ IF ( KRRL >= 1 .AND. .NOT. L2D) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PATHETA(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP4_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP4_DEVICE, ZTMP3_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP4_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP4_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP4_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP4_DEVICE)
ZTMP4_DEVICE(:,:,:) = PDZY(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP4_DEVICE, ZTMP5_DEVICE)
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP4_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP6_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP6_DEVICE)
ZTMP6_DEVICE(:,:,:) = ZTMP4_DEVICE(:,:,:)*ZTMP5_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP6_DEVICE, ZTMP4_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLXC)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = 2.*( ZTMP3_DEVICE(:,:,:) + ZTMP4_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP3_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL DYF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP3_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP6_DEVICE(:,:,:)*PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP3_DEVICE, ZTMP5_DEVICE )
@@ -1573,42 +1423,42 @@ IF (KRR/=0) THEN
!
IF ( KRRL >= 1 .AND. .NOT. L2D) THEN ! Sub-grid condensation
IF (.NOT. LFLAT) THEN
- ZFLXC(:,:,:) = ZFLXC(:,:,:) &
- + 2.*( MXF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:) ) &
- + MZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( &
- PDZY(:,:,:)*(MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) ) )&
+ ZFLXC(:,:,:) = ZFLXC(:,:,:) &
+ + 2.*( MXF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:) ) &
+ + MZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( &
+ PDZY(:,:,:)*(MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) ) ) &
)
IF ( KRRI >= 1 ) THEN
- PRRS(:,:,:,2) = PRRS(:,:,:,2) + 2. * &
- (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY ) &
- + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
- (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) )&
- * PINV_PDZZ(:,:,:) ) &
+ PRRS(:,:,:,2) = PRRS(:,:,:,2) + 2. * &
+ (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY ) &
+ + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
+ (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) ) &
+ * PINV_PDZZ(:,:,:) ) &
)*(1.0-PFRAC_ICE(:,:,:))
- PRRS(:,:,:,4) = PRRS(:,:,:,4) + 2. * &
- (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY ) &
- + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
- (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) )&
- * PINV_PDZZ(:,:,:) ) &
+ PRRS(:,:,:,4) = PRRS(:,:,:,4) + 2. * &
+ (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY ) &
+ + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
+ (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) ) &
+ * PINV_PDZZ(:,:,:) ) &
)*PFRAC_ICE(:,:,:)
ELSE
- PRRS(:,:,:,2) = PRRS(:,:,:,2) + 2. * &
- (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY ) &
- + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
- (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) )&
- * PINV_PDZZ(:,:,:) ) &
+ PRRS(:,:,:,2) = PRRS(:,:,:,2) + 2. * &
+ (- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY ) &
+ + DZF( MZM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*MYF( PDZY(:,:,:)* &
+ (MZM( ZFLX(:,:,:)*PINV_PDYY(:,:,:) )) ) &
+ * PINV_PDZZ(:,:,:) ) &
)
END IF
ELSE
ZFLXC(:,:,:) = ZFLXC(:,:,:) + 2.*MXF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX )
IF ( KRRI >= 1 ) THEN
PRRS(:,:,:,2) = PRRS(:,:,:,2) - 2. * &
- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY )*(1.0-PFRAC_ICE(:,:,:))
+ DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY )*(1.0-PFRAC_ICE(:,:,:))
PRRS(:,:,:,4) = PRRS(:,:,:,4) - 2. * &
- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY )*PFRAC_ICE(:,:,:)
+ DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY )*PFRAC_ICE(:,:,:)
ELSE
PRRS(:,:,:,2) = PRRS(:,:,:,2) - 2. * &
- DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)/PDYY )
+ DYF( MYM( PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:) )*ZFLX(:,:,:)*PINV_PDYY )
END IF
END IF
END IF
@@ -1657,9 +1507,7 @@ IF (KRR/=0) THEN
CALL MYM_DEVICE( PK, ZTMP1_DEVICE )
CALL GY_M_V_DEVICE(1,IKU,1,PRM(:,:,:,1),PDYY,PDZZ,PDZY, ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZFLX(JI,JJ,JK) = -XCHF * ZTMP1_DEVICE(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK)
END DO !CONCURRENT
@@ -1675,21 +1523,13 @@ IF (KRR/=0) THEN
! gradient
CALL MYM_DEVICE( PK(:,:,IKB:IKB), ZTMP1_DEVICE(:,:,1:1) )
CALL DYM_DEVICE(PRM(:,:,IKB:IKB,1), ZTMP2_DEVICE(:,:,1:1))
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP3_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,1) = ZCOEFF(:,:,IKB+2)*PRM(:,:,IKB+2,1) &
+ZCOEFF(:,:,IKB+1)*PRM(:,:,IKB+1,1) &
+ZCOEFF(:,:,IKB )*PRM(:,:,IKB ,1)
!$acc end kernels
CALL MYM_DEVICE( ZTMP3_DEVICE(:,:,1:1), ZTMP4_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLX)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB) = -XCHF * ZTMP1_DEVICE(:,:,1) * &
( ZTMP2_DEVICE(:,:,1) * PINV_PDYY(:,:,IKB) &
- ZTMP4_DEVICE(:,:,1) &
@@ -1703,11 +1543,7 @@ IF (KRR/=0) THEN
ZTMP1_DEVICE(:,:,1) = PSFRM(:,:)* PDIRCOSYW(:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE(:,:,1:1), ZTMP2_DEVICE(:,:,1:1) )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLX)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLX)
ZFLX(:,:,IKB-1) = 2. * ZTMP2_DEVICE(:,:,1) - ZFLX(:,:,IKB)
!$acc end kernels
!
@@ -1717,9 +1553,7 @@ IF (KRR/=0) THEN
IF (.NOT. LFLAT) THEN
CALL MYM_DEVICE(PRHODJ, ZTMP1_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP2_DEVICE(JI,JJ,JK) = ZTMP1_DEVICE(JI,JJ,JK) * ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
@@ -1727,27 +1561,21 @@ IF (KRR/=0) THEN
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
!
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = ZFLX(JI,JJ,JK) * PINV_PDYY(JI,JJ,JK)
END DO
!$acc end kernels
CALL MZM_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PDZY(JI,JJ,JK)*ZTMP2_DEVICE(JI,JJ,JK)
END DO
!$acc end kernels
CALL MYF_DEVICE(ZTMP1_DEVICE,ZTMP2_DEVICE)
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
ZTMP1_DEVICE(JI,JJ,JK) = PMZM_PRHODJ(JI,JJ,JK) * ZTMP2_DEVICE(JI,JJ,JK) * PINV_PDZZ(JI,JJ,JK)
END DO
@@ -1755,9 +1583,7 @@ IF (KRR/=0) THEN
CALL DZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
!
!$acc kernels
-#ifdef MNH_COMPILER_NVHPC
- !$acc loop independent collapse(3)
-#endif
+ !$acc_nv loop independent collapse(3)
DO CONCURRENT ( JI=1:JIU,JJ=1:JJU,JK=1:JKU)
PRRS(JI,JJ,JK,1) = PRRS(JI,JJ,JK,1) - ZTMP3_DEVICE(JI,JJ,JK) + ZTMP2_DEVICE(JI,JJ,JK)
END DO
@@ -1782,60 +1608,32 @@ IF (KRR/=0) THEN
ZTMP1_DEVICE(:,:,:) = PRHODJ(:,:,:)*PAMOIST(:,:,:)*PSRCM(:,:,:)
!$acc end kernels
CALL MYM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP3_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP3_DEVICE)
ZTMP3_DEVICE(:,:,:) = ZTMP2_DEVICE(:,:,:)*ZFLX(:,:,:)
!$acc end kernels
CALL MXF_DEVICE( ZTMP3_DEVICE, ZTMP4_DEVICE )
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP5_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP1_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZFLX(:,:,:)*PINV_PDYY(:,:,:)
!$acc end kernels
CALL MZM_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP1_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = PDZY(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MYF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP1_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP1_DEVICE)
ZTMP1_DEVICE(:,:,:) = ZTMP5_DEVICE(:,:,:)*ZTMP2_DEVICE(:,:,:)
!$acc end kernels
CALL MZF_DEVICE( ZTMP1_DEVICE, ZTMP2_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZFLXC)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZFLXC)
ZFLXC(:,:,:) = ZFLXC(:,:,:) + 2.*( ZTMP4_DEVICE(:,:,:) + ZTMP2_DEVICE(:,:,:) )
!$acc end kernels
IF ( KRRI >= 1 ) THEN
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP2_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP2_DEVICE(:,:,:) = ZTMP3_DEVICE(:,:,:) * PINV_PDYY(:,:,:)
!$acc end kernels
CALL DYF_DEVICE( ZTMP2_DEVICE, ZTMP3_DEVICE )
-#ifdef MNH_COMPILER_CCE
- !$acc kernels present(ZTMP2_DEVICE)
-#else
- !$acc kernels
-#endif
+ !$acc kernels present_cr(ZTMP2_DEVICE)
ZTMP2_DEVICE(:,:,:) = ZTMP1_DEVICE(:,:,:)* PINV_PDZZ(:,:,:)
!$acc end kernels
CALL DZF_DEVICE( ZTMP2_DEVICE, ZTMP4_DEVICE )
--
GitLab