Parallelise loops with OpenMP.

git-svn-id: http://svn.cactuscode.org/arrangements/CactusNumerical/MoL/trunk@129 578cdeb0-5ea1-4b81-8215-5a3b8777ee0b

9 files changed, 56 insertions, 0 deletions

diff --git a/src/GenericRK.c b/src/GenericRK.c
index f9622c7..6fc2dc9 100644
--- a/src/GenericRK.c
+++ b/src/GenericRK.c
@@ -140,6 +140,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
            beta);
 #endif
 
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateVar[index] = (*Original_Delta_Time) / cctkGH->cctk_timefac * beta * RHSVar[index];
@@ -203,6 +204,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
       
       if ( (alpha > MoL_Tiny)||(alpha < -MoL_Tiny) )
       {
+#pragma omp parallel for
         for (index = 0; index < totalsize; index++)
         {
           UpdateVar[index] += alpha * ScratchVar[index];
@@ -248,6 +250,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
                           MoL_Intermediate_Steps -
                           (*MoL_Intermediate_Step)) * totalsize, totalsize);
 #endif
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         ScratchVar[index] = UpdateVar[index];
@@ -278,6 +281,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
     RHSComplexVar = (CCTK_COMPLEX const *)CCTK_VarDataPtrI(cctkGH, 0, 
                                                            RHSVariableIndex[var]);
 
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateComplexVar[index] = CCTK_CmplxMul(Complex_Delta_Time,
@@ -310,6 +314,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
       
       if ( (alpha > MoL_Tiny)||(alpha < -MoL_Tiny) )
       {
+#pragma omp parallel for
         for (index = 0; index < totalsize; index++)
         {
           UpdateComplexVar[index] =
@@ -333,6 +338,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
                                                 MoL_Intermediate_Steps -
                                                 (*MoL_Intermediate_Step)) * 
                                               totalsize];
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         ScratchComplexVar[index] = UpdateComplexVar[index];
@@ -365,6 +371,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
 
     arraytotalsize = ArrayScratchSizes[var];
 
+#pragma omp parallel for
     for (index = 0; index < arraytotalsize; index++)
     {
       UpdateVar[index] = (*Original_Delta_Time) / cctkGH->cctk_timefac * beta * RHSVar[index];
@@ -393,6 +400,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
       
       if ( (alpha > MoL_Tiny)||(alpha < -MoL_Tiny) )
       {
+#pragma omp parallel for
         for (index = 0; index < arraytotalsize; index++)
         {
           UpdateVar[index] += alpha * ScratchVar[index];
@@ -441,6 +449,7 @@ void MoL_GenericRKAdd(CCTK_ARGUMENTS)
                  CurrentArrayScratchSize +
                  arrayscratchlocation);
 #endif
+#pragma omp parallel for
       for (index = 0; index < arraytotalsize; index++)
       {
         ScratchVar[index] = UpdateVar[index];
diff --git a/src/ICN.c b/src/ICN.c
index 7f75a35..c9a6114 100644
--- a/src/ICN.c
+++ b/src/ICN.c
@@ -120,6 +120,7 @@ void MoL_ICNAdd(CCTK_ARGUMENTS)
     RHSVar = (CCTK_REAL*)CCTK_VarDataPtrI(cctkGH, 0, 
                                           RHSVariableIndex[var]);
     
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateVar[index] = OldVar[index] + CCTK_DELTA_TIME * RHSVar[index];
@@ -155,6 +156,7 @@ void MoL_ICNAdd(CCTK_ARGUMENTS)
 /*                  "This proc array total size is %d.",  */
 /*                  arraytotalsize); */
 
+#pragma omp parallel for
     for (index = 0; index < arraytotalsize; index++)
     {
       UpdateVar[index] = OldVar[index] + CCTK_DELTA_TIME * RHSVar[index];
@@ -174,6 +176,7 @@ void MoL_ICNAdd(CCTK_ARGUMENTS)
     RHSComplexVar = (CCTK_COMPLEX*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                     RHSComplexVariableIndex[var]);
     
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateComplexVar[index] = CCTK_CmplxAdd(OldComplexVar[index],
@@ -269,6 +272,7 @@ void MoL_ICNAverage(CCTK_ARGUMENTS)
     RHSVar = (CCTK_REAL*)CCTK_VarDataPtrI(cctkGH, 0, 
                                           RHSVariableIndex[var]);
     
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
 /*       UpdateVar[index] = 0.5 * (UpdateVar[index] + OldVar[index]); */
@@ -302,6 +306,7 @@ void MoL_ICNAverage(CCTK_ARGUMENTS)
       arraytotalsize *= arraydata.lsh[arraydim];
     }
 
+#pragma omp parallel for
     for (index = 0; index < arraytotalsize; index++)
     {
 /*       UpdateVar[index] = 0.5 * (UpdateVar[index] + OldVar[index]); */
@@ -323,6 +328,7 @@ void MoL_ICNAverage(CCTK_ARGUMENTS)
     RHSComplexVar = (CCTK_COMPLEX*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                     RHSComplexVariableIndex[var]);
     
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateComplexVar[index] = CCTK_CmplxMul(Complex_Half, CCTK_CmplxAdd(UpdateComplexVar[index], OldComplexVar[index]));
diff --git a/src/InitialCopy.c b/src/InitialCopy.c
index 5b72393..b421908 100644
--- a/src/InitialCopy.c
+++ b/src/InitialCopy.c
@@ -556,6 +556,7 @@ void MoL_InitRHS(CCTK_ARGUMENTS)
                                           RHSVariableIndex[var]);
     if (RHSVar)
     {
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         RHSVar[index] = 0;
@@ -594,6 +595,7 @@ void MoL_InitRHS(CCTK_ARGUMENTS)
     {
       if (RHSVar)
       {
+#pragma omp parallel for
         for (index = 0; index < arraytotalsize; index++)
         {
           RHSVar[index] = 0;
@@ -636,6 +638,7 @@ void MoL_InitRHS(CCTK_ARGUMENTS)
                                                     RHSComplexVariableIndex[var]);
     if (RHSVar)
     {
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         RHSVar[index] = 0;
diff --git a/src/RK2.c b/src/RK2.c
index 1301e26..58dd03f 100644
--- a/src/RK2.c
+++ b/src/RK2.c
@@ -118,6 +118,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
           RHSVar = (CCTK_REAL const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                 RHSVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += CCTK_DELTA_TIME * RHSVar[index];
@@ -146,6 +147,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
             arraytotalsize *= arraydata.lsh[arraydim];
           }
 
+#pragma omp parallel for
           for (index = 0; index < arraytotalsize; index++)
           {
             UpdateVar[index] += CCTK_DELTA_TIME * RHSVar[index];
@@ -163,6 +165,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
           RHSComplexVar = (CCTK_COMPLEX const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                           RHSComplexVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateComplexVar[index] = CCTK_CmplxAdd(UpdateComplexVar[index],
@@ -186,6 +189,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
           RHSVar = (CCTK_REAL const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                 RHSVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] = 0.5 * (OldVar[index] + UpdateVar[index]) + 
@@ -218,6 +222,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
             arraytotalsize *= arraydata.lsh[arraydim];
           }
           
+#pragma omp parallel for
           for (index = 0; index < arraytotalsize; index++)
           {
             UpdateVar[index] = 0.5 * (OldVar[index] + UpdateVar[index]) + 
@@ -238,6 +243,7 @@ void MoL_RK2Add(CCTK_ARGUMENTS)
           RHSComplexVar = (CCTK_COMPLEX const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                           RHSComplexVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateComplexVar[index] =
diff --git a/src/RK3.c b/src/RK3.c
index 68329a2..f3b5e9c 100644
--- a/src/RK3.c
+++ b/src/RK3.c
@@ -121,6 +121,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSVar = (CCTK_REAL*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                 RHSVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += CCTK_DELTA_TIME * RHSVar[index];
@@ -149,6 +150,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
             arraytotalsize *= arraydata.lsh[arraydim];
           }
 
+#pragma omp parallel for
           for (index = 0; index < arraytotalsize; index++)
           {
             UpdateVar[index] += CCTK_DELTA_TIME * RHSVar[index];
@@ -166,6 +168,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSComplexVar = (CCTK_COMPLEX*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                           RHSComplexVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateComplexVar[index] = CCTK_CmplxAdd(UpdateComplexVar[index],
@@ -190,6 +193,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSVar = (CCTK_REAL const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                 RHSVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] = 0.25 * (3*OldVar[index] +
@@ -222,6 +226,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
             arraytotalsize *= arraydata.lsh[arraydim];
           }
 
+#pragma omp parallel for
           for (index = 0; index < arraytotalsize; index++)
           {
             UpdateVar[index] = 0.25*(3*OldVar[index] + UpdateVar[index]) +
@@ -242,6 +247,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSComplexVar = (CCTK_COMPLEX const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                           RHSComplexVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateComplexVar[index] =
@@ -271,6 +277,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSVar = (CCTK_REAL const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                 RHSVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] = (OldVar[index] + 2*UpdateVar[index]) * one_third
@@ -303,6 +310,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
             arraytotalsize *= arraydata.lsh[arraydim];
           }
           
+#pragma omp parallel for
           for (index = 0; index < arraytotalsize; index++)
           {
             UpdateVar[index] = (OldVar[index] + 2*UpdateVar[index]) * one_third
@@ -323,6 +331,7 @@ void MoL_RK3Add(CCTK_ARGUMENTS)
           RHSComplexVar = (CCTK_COMPLEX const*)CCTK_VarDataPtrI(cctkGH, 0, 
                                                           RHSComplexVariableIndex[var]);
           
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateComplexVar[index] =
diff --git a/src/RK4.c b/src/RK4.c
index c96377a..250c847 100644
--- a/src/RK4.c
+++ b/src/RK4.c
@@ -156,6 +156,7 @@ CCTK_WARN(0, "not implemented");
            beta);
 #endif
 
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       UpdateVar[index] = OldVar[index] + 
@@ -176,6 +177,7 @@ CCTK_WARN(0, "not implemented");
     /* scratch storage */
     if ((*MoL_Intermediate_Step) == MoL_Intermediate_Steps)
     {
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         ScratchVar[index] = 0;
@@ -184,6 +186,7 @@ CCTK_WARN(0, "not implemented");
 
     if ((*MoL_Intermediate_Step)>1)
     {
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         ScratchVar[index] += alpha * UpdateVar[index];
@@ -191,6 +194,7 @@ CCTK_WARN(0, "not implemented");
     }
     else
     {
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] += ScratchVar[index] - 4.0 / 3.0 * OldVar[index];
@@ -240,6 +244,7 @@ CCTK_WARN(0, "not done");
 
     ScratchVar = &ArrayScratchSpace[arrayscratchlocation];
 
+#pragma omp parallel for
     for (index = 0; index < arraytotalsize; index++)
     {
       UpdateVar[index] = OldVar[index] +
@@ -248,6 +253,7 @@ CCTK_WARN(0, "not done");
 
     if ((*MoL_Intermediate_Step) == MoL_Intermediate_Steps)
     {
+#pragma omp parallel for
       for (index = 0; index < arraytotalsize; index++)
       {
         ScratchVar[index] = 0;
@@ -256,6 +262,7 @@ CCTK_WARN(0, "not done");
     
     if ((*MoL_Intermediate_Step)>1)
     {
+#pragma omp parallel for
       for (index = 0; index < arraytotalsize; index++)
       {
         ScratchVar[index] += alpha * UpdateVar[index];
@@ -263,6 +270,7 @@ CCTK_WARN(0, "not done");
     }
     else
     {
+#pragma omp parallel for
       for (index = 0; index < arraytotalsize; index++)
       {
         UpdateVar[index] += ScratchVar[index] - 4.0 / 3.0 * OldVar[index];
diff --git a/src/RK45.c b/src/RK45.c
index 3ca2456..6173d77 100644
--- a/src/RK45.c
+++ b/src/RK45.c
@@ -181,6 +181,7 @@ void MoL_RK45Add(CCTK_ARGUMENTS)
                                               + MoL_Num_Evolved_Vars * 
                                               (MoL_Intermediate_Steps - 
                                                (*MoL_Intermediate_Step)));
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       ScratchVar[index] = tmp * RHSVar[index];
@@ -207,6 +208,7 @@ void MoL_RK45Add(CCTK_ARGUMENTS)
     if (*MoL_Intermediate_Step - 1)
     { 
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -226,6 +228,7 @@ void MoL_RK45Add(CCTK_ARGUMENTS)
         
         if ( (beta > MoL_Tiny)||(beta < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += beta * ScratchVar[index];
@@ -238,6 +241,7 @@ void MoL_RK45Add(CCTK_ARGUMENTS)
     else
     {
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -257,6 +261,7 @@ void MoL_RK45Add(CCTK_ARGUMENTS)
         
         if ( (gamma > MoL_Tiny)||(gamma < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += gamma * ScratchVar[index];
diff --git a/src/RK65.c b/src/RK65.c
index 22bd26b..e98f37a 100644
--- a/src/RK65.c
+++ b/src/RK65.c
@@ -140,6 +140,7 @@ void MoL_RK65Add(CCTK_ARGUMENTS)
                                               + MoL_Num_Evolved_Vars * 
                                               (MoL_Intermediate_Steps - 
                                                (*MoL_Intermediate_Step)));
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       ScratchVar[index] = tmp * RHSVar[index];
@@ -166,6 +167,7 @@ void MoL_RK65Add(CCTK_ARGUMENTS)
     if (*MoL_Intermediate_Step - 1)
     { 
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -185,6 +187,7 @@ void MoL_RK65Add(CCTK_ARGUMENTS)
         
         if ( (beta > MoL_Tiny)||(beta < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += beta * ScratchVar[index];
@@ -197,6 +200,7 @@ void MoL_RK65Add(CCTK_ARGUMENTS)
     else
     {
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -216,6 +220,7 @@ void MoL_RK65Add(CCTK_ARGUMENTS)
         
         if ( (gamma > MoL_Tiny)||(gamma < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += gamma * ScratchVar[index];
diff --git a/src/RK87.c b/src/RK87.c
index 3ce19b6..f806edd 100644
--- a/src/RK87.c
+++ b/src/RK87.c
@@ -161,6 +161,7 @@ void MoL_RK87Add(CCTK_ARGUMENTS)
                                               + MoL_Num_Evolved_Vars * 
                                               (MoL_Intermediate_Steps - 
                                                (*MoL_Intermediate_Step)));
+#pragma omp parallel for
     for (index = 0; index < totalsize; index++)
     {
       ScratchVar[index] = tmp * RHSVar[index];
@@ -188,6 +189,7 @@ void MoL_RK87Add(CCTK_ARGUMENTS)
     if (*MoL_Intermediate_Step - 1)
     { 
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -207,6 +209,7 @@ void MoL_RK87Add(CCTK_ARGUMENTS)
         
         if ( (beta > MoL_Tiny)||(beta < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += beta * ScratchVar[index];
@@ -219,6 +222,7 @@ void MoL_RK87Add(CCTK_ARGUMENTS)
     else
     {
 
+#pragma omp parallel for
       for (index = 0; index < totalsize; index++)
       {
         UpdateVar[index] = OldVar[index];
@@ -238,6 +242,7 @@ void MoL_RK87Add(CCTK_ARGUMENTS)
         
         if ( (gamma > MoL_Tiny)||(gamma < -MoL_Tiny) )
         {
+#pragma omp parallel for
           for (index = 0; index < totalsize; index++)
           {
             UpdateVar[index] += gamma * ScratchVar[index];