BF OpenMp b4b

manual/impl/switch.tex

@@ -280,7 +280,7 @@ \subsubsection{~Optional switches} \label{sub:opt_switch}
 \sit{ompx}{Idem, but for directives used only for exclusive OpenMP
     parallelization.}  
 \sit{omph}{Idem, but for directives used only for hybrid MPI-OpenMP
-    parallelization.}
+    parallelization. Note that when using OMPH the test switch T1 cannot be used.}
 \sit{pdlib}{Domain Decomposition for Explicit and Implicit Solver on triangular unstructured grids. ({\code ParMetis} is required for this option)}
 \end{slist}
 Note that these switches can only be used in certain combinations, as enforced

model/bin/make_makefile.sh

@@ -97,9 +97,9 @@
               shared mpp mpiexp thread GSE prop \
               stress s_ln source stab s_nl snls s_bot s_db miche s_tr s_bs \
               dstress s_ice s_is reflection s_xx \
-              wind windx wcor rwind curr currx mgwind mgprop mggse \
+              wind windx rwind curr currx mgwind mgprop mggse \
               subsec tdyn dss0 pdif tide refrx ig rotag arctic nnt mprf \
-              cou oasis agcm ogcm igcm trknc setup pdlib memck uost rstwind
+              cou oasis agcm ogcm igcm trknc setup pdlib memck uost dbtest
   do
     case $type in
 #sort:mach:
@@ -163,6 +163,11 @@
                ID='directive controlled threading'
                TS='OMP'
                OK='OMPG OMPX OMPH' ;;
+#sort:dbtest:
+      dbtest ) TY='upto1'
+               ID='test for T1 conflict eith OMPH'
+               TS='T1'
+               OK='T1' ;;
 #sort:GSE:
       GSE    ) TY='one'
                ID='GSE aleviation'
@@ -247,23 +252,11 @@
       windx  ) TY='one'
                ID='wind interpolation in space'
                OK='WNX0 WNX1 WNX2' ;;
-#sort:wcor:
-      wcor   ) TY='upto1'
-               ID='wind speed correction'
-               TS='WCOR'
-               OK='WCOR' ;;
 #sort:rwind:
       rwind  ) TY='upto1'
                ID='wind vs. current definition'
                TS='RWND'
                OK='RWND' ;;
-
-#sort:rstwind:
-      rstwind  ) TY='upto1'
-               ID='wind in restart for wmesmf'
-               TS='WRST'
-               OK='WRST' ;;
-
 #sort:curr:
       curr   ) TY='one'
                ID='current interpolation in time'
@@ -508,6 +501,7 @@
       memck  ) memck=$sw ;;
       setup  ) setup=$sw ;;
       uost   ) uost=$sw ;;
+      dbtest ) dbtest=$sw ;;
               *    ) ;;
     esac
   done
@@ -677,6 +671,14 @@
       echo ' ' ; exit 11
   fi
 
+  if [ "$dbtest" = 'T1' ] && [ "$thread2" = 'OMPH' ]
+  then
+      echo ' '
+      echo "   *** !/T1 cannot be used in combination with !/OMPH"
+      echo "       Remove OMPH if desire is to test T1 GRINFO."
+      echo ' ' ; exit 11
+  fi
+
   case $s_nl in
    NL0) nl=$NULL
         nlx=$NULL ;;
@@ -863,7 +865,6 @@
          ww3_ounp ww3_gspl ww3_gint ww3_bound ww3_bounc ww3_systrk $tideprog"
   progs="$progs ww3_multi_esmf  ww3_uprstr"
   progs="$progs libww3"
-  progs="$progs libww3.so"
 
   for prog in $progs
   do
@@ -1072,13 +1073,6 @@
              source="w3triamd w3srcemd $dsx $flx $ln $st $nl $bt $ic $is $db $tr $bs $xx $refcode $igcode $uostmd"
                  IO='w3iogrmd w3iogomd w3iopomd w3iotrmd w3iorsmd w3iobcmd w3iosfmd w3partmd'
                 aux="constants w3servmd w3timemd $tidecode w3arrymd w3dispmd w3cspcmd w3gsrumd" ;;
-     libww3.so) IDstring='Object file archive'
-               core='w3fldsmd w3initmd w3wavemd w3wdasmd w3updtmd'
-               data='wmmdatmd w3gdatmd w3wdatmd w3adatmd w3idatmd w3odatmd'
-               prop="$pr"
-             source="w3triamd w3srcemd $dsx $flx $ln $st $nl $bt $ic $is $db $tr $bs $xx $refcode $igcode $uostmd"
-                 IO='w3iogrmd w3iogomd w3iopomd w3iotrmd w3iorsmd w3iobcmd w3iosfmd w3partmd'
-                aux="constants w3servmd w3timemd $tidecode w3arrymd w3dispmd w3cspcmd w3gsrumd" ;;  
      ww3_uprstr) IDstring='Update Restart File' 
               core= 
 	          data='wmmdatmd w3triamd w3gdatmd w3wdatmd w3adatmd w3idatmd w3odatmd' 
@@ -1098,7 +1092,7 @@
       filesl="$data $core $prop $source $IO $aux"
     # if program name is libww3, then
     # the target is compile and create archive
-    elif [ "$prog" = "libww3" ] ||  [ "$prog" = "libww3.so" ]
+    elif [ "$prog" = "libww3" ]
     then
       d_string="$prog"' : $(aPo)/'
       files="$aux $core $data $prop $source $IO"
@@ -1143,19 +1137,6 @@
       done
       echo "	@cd \$(aPo); $ar_cmd $lib $objs" >> makefile
       echo ' '                                   >> makefile
-    # if program name is libww3.so, then
-    # the target is compile and create archive
-    elif [ "$prog" = "libww3.so" ]
-    then
-      lib=$prog
-      objs=""
-      for file in $filesl
-      do
-        objs="$objs $file.o"
-      done
-      echo "	@cd \$(aPo); ld -o $lib -shared $objs" >> makefile
-      echo ' '                                   >> makefile
-
     else
       echo '	@$(aPb)/link '"$filesl"          >> makefile
       echo ' '                                   >> makefile

model/ftn/w3iogomd.ftn

@@ -1462,7 +1462,7 @@
 ! 2.c Finalize integration over band and update mean arrays
 !
 !
-!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA,FACTOR,FACTOR2,KD,FKD,USSCO,M1)
+!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA,FACTOR,FACTOR2,MA,MC,MB,KD,FKD,USSCO,M1)
 !
         DO JSEA=1, NSEAL
           CALL INIT_GET_ISEA(ISEA, JSEA)
@@ -1680,7 +1680,7 @@
 !
         END DO
 
-!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA)
+!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA,IX,IY,STEX,STEY,STED,ITL,IK)
 !
         DO JSEA=1, NSEAL
 !/DIST        ISEA      = IAPROC + (JSEA-1)*NAPROC
@@ -2239,7 +2239,7 @@
 !
 ! 6.b Loop over local sea points
 !
-!/OMPG/!$OMP PARALLEL DO PRIVATE(ISEA,JSEA,IX,IY,I)
+!/OMPG/!$OMP PARALLEL DO PRIVATE(ISEA,JSEA,IX,IY,I,J)
 !
           DO JSEA=1, NSEAL
             CALL INIT_GET_ISEA(ISEA, JSEA)
@@ -3730,7 +3730,7 @@
 !
          DO ITH=1, NTH
 !
-!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA,FACTOR)
+!/OMPG/!$OMP PARALLEL DO PRIVATE(JSEA,ISEA)
 !
             DO JSEA=1, NSEAL
 !/DIST                ISEA         = IAPROC + (JSEA-1)*NAPROC
@@ -3747,7 +3747,7 @@
 !
 !
 !/OMPG/!$OMP PARALLEL DO
-!PRIVATE(JSEA,ISEA,FACTOR,FACTOR2,KD,FKD,USSCO,M1)
+!PRIVATE(JSEA,ISEA,FACTOR,FACTOR2,KD,FKD,USSCO,M1,MINDIFF,IB)
 !
          DO JSEA=1, NSEAL
 !/DIST              ISEA         = IAPROC + (JSEA-1)*NAPROC
@@ -3786,11 +3786,11 @@
                ENDDO
                MINDIFF=1.e8
                !Put spectral energey into whichever band central wavenumber fits in
-               NB=NK!allocated to size2*NK! STKBND_IN(NK)
                USSP(JSEA,Spc2Bnd(IK))    =  USSP(JSEA,Spc2Bnd(IK)) + ABX(JSEA)*USSCO
                USSP(JSEA,NK+Spc2BND(IK)) =  USSP(JSEA,NK+Spc2Bnd(IK)) + ABY(JSEA)*USSCO
             ENDIF
          END DO
+!/OMPG/!$OMP END PARALLEL DO
       END DO
 !
       RETURN

model/ftn/w3pro1md.ftn

@@ -583,7 +583,7 @@
         NYMAX=NY-1
         IF ( ICLOSE.EQ.ICLOSE_TRPL ) NYMAX=NY
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IX, IY, IXY)
+!/OMPH/!$OMP PARALLEL DO PRIVATE (IX, IY, IXY, VCB)
 !
         DO IX=1, NX
            DO IY=1, NYMAX
@@ -627,7 +627,7 @@
         IF ( ICLOSE.EQ.ICLOSE_TRPL ) THEN
            IY=NY
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IXY, VCB, IX, IY)
+!/OMPH/!$OMP PARALLEL DO PRIVATE (IXY, VCB, IX)
 !
            DO IX=1, NX
               IXY    = IY +(IX-1)*NY

model/ftn/w3pro3md.ftn

@@ -960,7 +960,7 @@
       IF ( FLCUR ) THEN
 !/T          WRITE (NDST,9022)
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IXY)
+!/OMPH/!$OMP PARALLEL DO PRIVATE (ISEA,IXY)
 !
           DO ISEA=1, NSEA
             IXY         = MAPSF(ISEA,3)
@@ -974,7 +974,7 @@
 !
         END IF
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IX, IY, IXY, CP, CQ)
+!/OMPH/!$OMP PARALLEL DO PRIVATE (ISEA,IX, IY, IXY, CP, CQ)
 !
       DO ISEA=1, NSEA
          IX     = MAPSF(ISEA,1)

model/ftn/w3uno2md.ftn

@@ -815,7 +815,7 @@
 !/T1      WRITE (NDST,9010)
 !/T1      WRITE (NDST,9011) NB0, 'CENTRAL'
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IP, IXY, CFL, IXYC, IXYD, QB)  
+!/OMPH/!$OMP PARALLEL DO PRIVATE (IP, IXY, CFL, IXYC, IXYD, QB, QBO, IY2, IX2, QN)  
 !
       DO IP=1, NB0
 !
@@ -910,7 +910,7 @@
 !/C90/!DIR$ IVDEP
 !/NEC/!CDIR NODEP
 !
-!/OMPH/!$OMP PARALLEL DO PRIVATE (IP, IXY, JN, JP )
+!/OMPH/!$OMP PARALLEL DO PRIVATE (IP, IXY, JN, JP, QOLD )
 !
       DO IP=1, NACT
 !

model/ftn/w3uqckmd.ftn

@@ -1,3 +1,4 @@
+
 #include "w3macros.h"
 !/ ------------------------------------------------------------------- /
       MODULE W3UQCKMD
@@ -877,7 +878,7 @@
 !/T1      WRITE (NDST,9011) NB0, 'CENTRAL'
 !
 !/OMPH/!$OMP PARALLEL DO PRIVATE (IP, IXY, CFL, IXYC, QB, IXYU, IXYD, &
-!/OMPH/!$OMP&                     DQ, DQNZ, QCN, QBN, QBR, CFAC )
+!/OMPH/!$OMP&                     DQ, DQNZ, QCN, QBN, QBR, CFAC, IY, IX )
 !
       DO IP=1, NB0
 !

model/ftn/w3wavemd.ftn

@@ -1490,7 +1490,7 @@
                       NKCFL=NK
 !/T                   NKCFL=1
 !
-!/OMPG/!$OMP PARALLEL DO PRIVATE (JSEA,ISEA,IX,IY) SCHEDULE (DYNAMIC,1)
+!/OMPG/!$OMP PARALLEL DO PRIVATE (JSEA,ISEA) SCHEDULE (DYNAMIC,1)
 !
                       DO JSEA=1, NSEAL
                         CALL INIT_GET_ISEA(ISEA, JSEA)
@@ -1585,7 +1585,7 @@
           IF ( FLCTH .OR. FLCK ) THEN
               DO ITLOC=1, ITLOCH
 !
-!/OMPG/!$OMP PARALLEL PRIVATE (JSEA,ISEA,IX,IY,DEPTH)
+!/OMPG/!$OMP PARALLEL PRIVATE (JSEA,ISEA,IX,IY,DEPTH,IXrel)
 !/OMPG/!$OMP DO SCHEDULE (DYNAMIC,1)
 !
 !/DEBUGRUN                WRITE(740+IAPROC,*) ' ITLOC=', ITLOC
@@ -1725,14 +1725,14 @@
 !/MPI         CALL MPI_STARTALL (NRQSG1, IRQSG1(1,2), IERR_MPI)
 !/MPI       END IF
 !
-!/OMPX/!$OMP PARALLEL PRIVATE (ISPEC,FIELD)
+!!/OMPX/!$OMP PARALLEL PRIVATE (ISPEC,FIELD)
 !
 !/DEBUGRUN        WRITE(740+IAPROC,*) 'W3WAVE, step 6.14'
 !/DEBUGRUN        FLUSH(740+IAPROC)
 
             IF ( FLOMP ) ALLOCATE ( FIELD(1-NY:NY*(NX+2)) )
 !
-!/OMPX/!$OMP DO SCHEDULE (DYNAMIC,1)
+!!/OMPX/!$OMP DO SCHEDULE (DYNAMIC,1)
 !
             DO ISPEC=1, NSPEC
               IF ( IAPPRO(ISPEC) .EQ. IAPROC ) THEN
@@ -1781,11 +1781,11 @@
 !/MEMCHECK       call printMallInfo(IAPROC,mallInfos)
 
 
-!/OMPX/!$OMP END DO
+!!/OMPX/!$OMP END DO
 
               IF ( FLOMP ) DEALLOCATE ( FIELD )
 
-!/OMPX/!$OMP END PARALLEL
+!!/OMPX/!$OMP END PARALLEL
 
 !Li   Initialise IK IX IY in case ARC option is not used to avoid warnings.
               IK=1
@@ -1873,7 +1873,7 @@
           IF ( FLCTH .OR. FLCK ) THEN
               DO ITLOC=ITLOCH+1, NTLOC
 !
-!/OMPG/!$OMP PARALLEL PRIVATE (JSEA,ISEA,IX,IY,DEPTH)
+!/OMPG/!$OMP PARALLEL PRIVATE (JSEA,ISEA,IX,IY,DEPTH,IXrel)
 !/OMPG/!$OMP DO SCHEDULE (DYNAMIC,1)
 !
 !/DEBUGRUN                WRITE(740+IAPROC,*) ' ITLOC=', ITLOC
@@ -1964,7 +1964,7 @@
 !/PDLIB!/DEBUGSRC          END IF
 !
 !/OMPG/!$OMP PARALLEL PRIVATE (JSEA,ISEA,IX,IY,DELA,DELX,DELY,        &
-!/OMPG/!$OMP&                  REFLEC,REFLED,D50,PSIC,TMP1,TMP2,TMP3)
+!/OMPG/!$OMP&                  REFLEC,REFLED,D50,PSIC,TMP1,TMP2,TMP3,TMP4)
 !/OMPG/!$OMP DO SCHEDULE (DYNAMIC,1)
 !
               DO JSEA=1, NSEAL

regtests/bin/matrix_ncep

@@ -121,7 +121,7 @@ fi
   export     shrd='y' # Do shared architecture tests
   export     dist='y' # Do distributed architecture (MPI) tests
   export      omp='y' # Threaded (OpenMP) tests
-  export     hybd='n' # Hybrid options
+  export     hybd='y' # Hybrid options
 
   export   prop1D='y' # 1-D propagation tests (ww3_tp1.X)
   export   prop2D='y' # 2-D propagation tests (ww3_tp2.X)