Skip to content
Snippets Groups Projects
Commit c81edcf3 authored by Fabio Roberto Vitello's avatar Fabio Roberto Vitello
Browse files

Update aprod.c

parent 8e801e93
No related branches found
No related tags found
No related merge requests found
Show whitespace changes
Inline Side-by-side
Showing
with 15 additions and 86 deletions
aprod.c
+ 15
− 86
View file @ c81edcf3
...@@ -13,15 +13,11 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -13,15 +13,11 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
double *systemMatrix, long int *matrixIndex, int *instrCol, int *instrConstrIlung, struct comData comlsqr, time_t *ompSec) double *systemMatrix, long int *matrixIndex, int *instrCol, int *instrConstrIlung, struct comData comlsqr, time_t *ompSec)
{ {
// Parallel definitions // Parallel definitions
int myid, nproc; int myid, nproc;
long int *mapNoss, *mapNcoeff; long int *mapNoss, *mapNcoeff;
int nthreads, tid, ntasks; int nthreads, tid, ntasks;
long **mapForThread; long **mapForThread;
/// struct comData *comlsqr;
//
FILE *fk, *fk0; FILE *fk, *fk0;
...@@ -77,18 +73,8 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -77,18 +73,8 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
long offsetGlobParam = comlsqr.offsetGlobParam; long offsetGlobParam = comlsqr.offsetGlobParam;
/* for(int n=0;n<ntasks;n++)
comlsqr->mapForThread[n]=(long *) calloc(3,sizeof(long));
*/
//nthreads = 1;
tid = 0; tid = 0;
FILE *fp1, *fp2;
// fp1=fopen("test1_aprod","w");
// fp2=fopen("test2_aprod","w");
if (mode != 1 && mode != 2) if (mode != 1 && mode != 2)
{ {
...@@ -103,7 +89,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -103,7 +89,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
time_t startTime = time(NULL); time_t startTime = time(NULL);
//// #pragma omp parallel private(myid, sum, k, l1, l2, l, j, tid, nthreads, i2, na) shared(mapNoss, instrCol, comlsqr, vVect, systemMatrix, matrixIndex, knownTerms, j2)
{ {
myid = comlsqr.myid; myid = comlsqr.myid;
...@@ -128,8 +113,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -128,8 +113,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
int nGlobVal = nAstroPSolved + nAttP + nInstrPSolved; int nGlobVal = nAstroPSolved + nAttP + nInstrPSolved;
jstartAstro = miValAstro - offLocalAstro; jstartAstro = miValAstro - offLocalAstro;
//FV_ EDIT ompSs
for(int nt=0; nt < ntasks; nt++ ) for(int nt=0; nt < ntasks; nt++ )
{ {
...@@ -221,8 +204,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -221,8 +204,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
{ {
sum = 0.0; sum = 0.0;
offExtConstr = mapNcoeff[myid] + iexc * nOfElextObs; offExtConstr = mapNcoeff[myid] + iexc * nOfElextObs;
//FV_ EDIT ompSs
//#pragma omp for
for (int j3 = 0; j3 < numOfExtStar * nAstroPSolved; j3++) for (int j3 = 0; j3 < numOfExtStar * nAstroPSolved; j3++)
sum += systemMatrix[offExtConstr + j3] * vVect[j3]; sum += systemMatrix[offExtConstr + j3] * vVect[j3];
...@@ -230,16 +211,13 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -230,16 +211,13 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
{ {
offExtAtt = offExtConstr + numOfExtStar * nAstroPSolved + nax * numOfExtAttCol; offExtAtt = offExtConstr + numOfExtStar * nAstroPSolved + nax * numOfExtAttCol;
vVIx = offExtAttConstr + nax * nDegFreedomAtt; vVIx = offExtAttConstr + nax * nDegFreedomAtt;
//FV_ EDIT ompSs
//#pragma omp for
for (int j3 = 0; j3 < numOfExtAttCol; j3++) for (int j3 = 0; j3 < numOfExtAttCol; j3++)
sum += systemMatrix[offExtAtt + j3] * vVect[vVIx + j3]; sum += systemMatrix[offExtAtt + j3] * vVect[vVIx + j3];
} }
//FV_ EDIT ompSs
//#pragma omp atomic
knownTerms[ktIx + iexc] += sum; knownTerms[ktIx + iexc] += sum;
} //for iexc } //for iexc
} }
////////////////////////////////////////////////////// //////////////////////////////////////////////////////
/// Mode 1 BarConstr /// Mode 1 BarConstr
if (nEqBarConstr) if (nEqBarConstr)
...@@ -251,12 +229,8 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -251,12 +229,8 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
{ {
sum = 0.0; sum = 0.0;
offBarConstrIx = offBarConstr + iexc * nOfElBarObs; offBarConstrIx = offBarConstr + iexc * nOfElBarObs;
//FV_ EDIT ompSs
//#pragma omp for
for (int j3 = 0; j3 < numOfBarStar * nAstroPSolved; j3++) for (int j3 = 0; j3 < numOfBarStar * nAstroPSolved; j3++)
sum += systemMatrix[offBarConstrIx + j3] * vVect[j3]; sum += systemMatrix[offBarConstrIx + j3] * vVect[j3];
//FV_ EDIT ompSs
//#pragma omp atomic
knownTerms[ktIx + iexc] += sum; knownTerms[ktIx + iexc] += sum;
} //for iexc } //for iexc
} }
...@@ -283,15 +257,11 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -283,15 +257,11 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
offSetInstr += instrConstrIlung[m]; offSetInstr += instrConstrIlung[m];
} }
offvV = mapNoss[myid] * nInstrPSolved + offSetInstr; offvV = mapNoss[myid] * nInstrPSolved + offSetInstr;
//FV_ EDIT ompSs
//#pragma omp for
for (int j3 = 0; j3 < instrConstrIlung[i1]; j3++) for (int j3 = 0; j3 < instrConstrIlung[i1]; j3++)
{ {
vVix = instrCol[offvV + j3]; vVix = instrCol[offvV + j3];
sum += systemMatrix[offSetInstrInc + j3] * vVect[offSetInstrConstr1 + vVix]; sum += systemMatrix[offSetInstrInc + j3] * vVect[offSetInstrConstr1 + vVix];
} }
//FV_ EDIT ompSs
//#pragma omp atomic
knownTerms[ktIx + i1] += sum; knownTerms[ktIx + i1] += sum;
} }
} }
...@@ -308,7 +278,7 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -308,7 +278,7 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
} }
else //mode==2 else //mode==2
{ //if(mode {
time_t startTime = time(NULL); time_t startTime = time(NULL);
...@@ -316,19 +286,12 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -316,19 +286,12 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
myid = comlsqr.myid; myid = comlsqr.myid;
printf("vVect[1]: %f:\n", vVect[1] ); printf("vVect[1]: %f:\n", vVect[1] );
//qui lavorare
/*
#ifdef OMP
tid = omp_get_thread_num();
nthreads = omp_get_num_threads();
#endif
*/
///////////////////////////////////////////////////// /////////////////////////////////////////////////////
/// Mode 2 Astrometric Sect /// Mode 2 Astrometric Sect
if (nAstroPSolved) if (nAstroPSolved)
{ {
long offLocalAstro = comlsqr.mapStar[myid][0] * nAstroPSolved; long offLocalAstro = comlsqr.mapStar[myid][0] * nAstroPSolved;
// for (long ix = mapForThread[tid][0]; ix < mapForThread[tid][2]; ix++)
long lset; long lset;
long jstartAstro; long jstartAstro;
...@@ -373,10 +336,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -373,10 +336,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
aprodM2AstroP(ix, nt, comlsqr, vVect, systemMatrix, matrixIndex, knownTerms); aprodM2AstroP(ix, nt, comlsqr, vVect, systemMatrix, matrixIndex, knownTerms);
} //for ix } //for ix
} }
// #pragma omp taskwait
} }
...@@ -395,12 +354,9 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -395,12 +354,9 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
long lset; long lset;
// printf("ntask: %d\n",ntasks);
for(int nt=0; nt < ntasks; nt++ ) for(int nt=0; nt < ntasks; nt++ )
{ {
// #pragma omp task shared(nparam,nInstrPSolved,matrixIndex) in (nt) commutative(vVect)
// #pragma omp task commutative(vVect)
#pragma omp task label(nAttP) commutative(vVect)//out(vVect[mix : mix + (nAttAxes-1) * nDegFreedomAtt]) #pragma omp task label(nAttP) commutative(vVect)//out(vVect[mix : mix + (nAttAxes-1) * nDegFreedomAtt])
{ {
for (long ix = mapForThread[nt][0]; ix < mapForThread[nt][2]; ix++) for (long ix = mapForThread[nt][0]; ix < mapForThread[nt][2]; ix++)
...@@ -436,11 +392,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -436,11 +392,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
} }
// long lset;
///#pragma omp for
} }
///////////////////////////////////////////////////// /////////////////////////////////////////////////////
/// Mode 2 Instrument Sect /// Mode 2 Instrument Sect
...@@ -457,8 +408,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -457,8 +408,6 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
for(int nt=0; nt < ntasks; nt++ ) for(int nt=0; nt < ntasks; nt++ )
{ {
// #pragma omp task shared(vVect,nparam,offLset,nInstrPSolved) in (nt)
// #pragma omp task commutative(vVect)
#pragma omp task label(nOfInstrConstr) commutative(vVect) //out(vVect[mix : mix + (nAttAxes-1) * nDegFreedomAtt]) #pragma omp task label(nOfInstrConstr) commutative(vVect) //out(vVect[mix : mix + (nAttAxes-1) * nDegFreedomAtt])
{ {
long lset; long lset;
...@@ -487,15 +436,13 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -487,15 +436,13 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
aprodM2InstrP(ix, nt, comlsqr, vVect, systemMatrix, instrCol, knownTerms); aprodM2InstrP(ix, nt, comlsqr, vVect, systemMatrix, instrCol, knownTerms);
} //for ix } //for ix
} }
// #pragma omp taskwait
} }
} }
/////////////////////////////////////////////////////
///} //pragma
*ompSec += time(NULL) - startTime; *ompSec += time(NULL) - startTime;
if (comlsqr.itn <= 2 && (myid == 0 || myid == nproc - 1 || debugMode == 1)) if (comlsqr.itn <= 2 && (myid == 0 || myid == nproc - 1 || debugMode == 1))
printf("PE=%d AprodTiming: mode=2.1 OmpSec=%ld\n", myid, time(NULL) - startTime); printf("PE=%d AprodTiming: mode=2.1 OmpSec=%ld\n", myid, time(NULL) - startTime);
///////////////////////////////////////////////////// /////////////////////////////////////////////////////
/// Mode 2 Global Sect /// Mode 2 Global Sect
if (nGlobP ) if (nGlobP )
...@@ -533,29 +480,15 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -533,29 +480,15 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
} }
//////////////////////////////////////////////////// ////////////////////////////////////////////////////
startTime = time(NULL); startTime = time(NULL);
//FV_ EDIT ompSs
// #pragma omp parallel private(myid, yi, localSum, tid, nthreads, i2, j2, na) shared(mapNoss, vVect, systemMatrix, knownTerms, k, j3)
{
myid = comlsqr.myid; myid = comlsqr.myid;
//FV_ EDIT ompSs
/*
//#ifdef OMP
tid = omp_get_thread_num();
nthreads = omp_get_num_threads();
#endif
*/
localSum = 0.0;
localSum = 0.0;
printf("\t\t\t\nEqExtConstr: %d nEqBarConstr: %d nOfInstrConstr: %d", nEqExtConstr,nEqBarConstr,nOfInstrConstr); printf("\t\t\t\nEqExtConstr: %d nEqBarConstr: %d nOfInstrConstr: %d", nEqExtConstr,nEqBarConstr,nOfInstrConstr);
#pragma omp taskwait #pragma omp taskwait
//////////////////////////////////////////////////////
/// Mode 2 ExtConstr
if (nEqExtConstr && true) if (nEqExtConstr && true)
{ {
time_t startTime_nEqExtConstr = time(NULL); time_t startTime_nEqExtConstr = time(NULL);
...@@ -679,13 +612,9 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms, ...@@ -679,13 +612,9 @@ void aprod(int mode, long int m, long int n, double *vVect, double *knownTerms,
} }
//////////// ////////////
} //pragma
*ompSec += time(NULL) - startTime; *ompSec += time(NULL) - startTime;
if (comlsqr.itn <= 2 && (myid == 0 || myid == nproc - 1 || debugMode == 1)) if (comlsqr.itn <= 2 && (myid == 0 || myid == nproc - 1 || debugMode == 1))
printf("PE=%d AprodTiming: mode=2.2 OmpSec=%ld\n", myid, time(NULL) - startTime); printf("PE=%d AprodTiming: mode=2.2 OmpSec=%ld\n", myid, time(NULL) - startTime);
} // else if(mode==2) } // else if(mode==2)
} }
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment