Commit a18c2291 authored by Fabio Roberto Vitello's avatar Fabio Roberto Vitello
Browse files

Code redesigned for OmpSs

parent 692132e7

aprod.c

+681 −94

File changed.

Preview size limit exceeded, changes collapsed.

cblas.h

+2 −2
Original line number Diff line number Diff line
@@ -50,7 +50,7 @@ cblas_daxpy(const int N, const double alpha, const double *X, 


void

cblas_dcopy(const long int N, const double *X, const int incX, 

            double *Y, const int incY);

            double *Y, const int incY, int ntasks=0);





double
@@ -58,7 +58,7 @@ cblas_ddot(const int N, const double *X, const int incX, 
           const double *Y, const int incY);



double

cblas_dnrm2(const long int N, const double *X, const int incX);

cblas_dnrm2(const long int N, const double *X, const int incX, int ntasks=0);



void

cblas_dscal(const long int N, const double alpha, double *X, const int incX);

lsqr.c

+291 −143

File changed.

Preview size limit exceeded, changes collapsed.

lsqrblas.c

+40 −24
Original line number Diff line number Diff line
@@ -26,6 +26,10 @@ 
*/

#include <stdio.h>

#include <stdlib.h>

#define MIN(a,b) (((a)<(b))?(a):(b))

#define MAX(a,b) (((a)>(b))?(a):(b))





/*!

   \file
@@ -95,7 +99,7 @@ cblas_daxpy( const int N, const double alpha, const double *X, 
*/

void

cblas_dcopy( const long int N, const double *X,

             const int incX, double *Y, const int incY)

             const int incX, double *Y, const int incY, int ntasks=0)

{

  long int i,ix,iy;

//  int ix = OFFSET(N, incX);
@@ -107,17 +111,24 @@ cblas_dcopy( const long int N, const double *X, 
    else iy=((N) - 1) * (-(incY));





#pragma omp parallel shared(ix, iy)

//#pragma omp parallel shared(ix, iy)

//#pragma omp task shared(ix, iy)

{

    

    

   long ixThread,iyThread;

   long ixThreadOffset=ix, iyThreadOffset=iy;

#pragma omp for

//#pragma omp for

  for (i = 0; i < N; i++) {

      //printf("Inside task %d\n", i);

      ixThread=ixThreadOffset+incX*i;

      iyThread=iyThreadOffset+incY*i;

      Y[iyThread]  = X[ixThread];

  }

   



}//end pragma

//#pragma omp taskwait

}



/*!
@@ -157,7 +168,7 @@ cblas_ddot( const int N, const double *X, 
*/



double

cblas_dnrm2( const long int N, const double *X, const int incX) 

cblas_dnrm2( const long int N, const double *X, const int incX, int ntasks=0) 

{

  double

      scale = 0.0,
@@ -190,9 +201,9 @@ cblas_dnrm2( const long int N, const double *X, const int incX) 
  return scale * sqrt(ssq);

}





//FV_ not used ??

double

cblas_dnrm21( const long int N, const double *X, const int incX) 

cblas_dnrm21( const long int N, const double *X, const int incX, int ntasks=0) 

{

  double

       ssq[100], ssqFinal=0;
@@ -204,21 +215,25 @@ cblas_dnrm21( const long int N, const double *X, const int incX) 


  if (N <= 0 || incX <= 0) return 0;

  else if (N == 1)         return fabs(X[0]);

#pragma omp parallel private(tid,nthreads,ix) 



////#pragma omp parallel private(tid,nthreads,ix) 

{

    

    /*

#ifdef OMP

        tid = omp_get_thread_num();

        nthreads = omp_get_num_threads();

        if(nthreads>100) exit(1);

#endif

    */

  

  

#pragma omp for

////#pragma omp for

  for (long i = 0; i < N; i++) {

       ix=incX*i;

       ssq[tid] += X[ix]*X[ix];

  }

}  // pragma

}

// pragma



for(int j=0;j<100;j++)

	ssqFinal+=ssq[j];
@@ -226,7 +241,7 @@ for(int j=0;j<100;j++) 
  return sqrt(ssqFinal);

}





//FV_ not used ??

double

cblas_dnrm22( const long int N, const double *X, const int incX) 

{
@@ -243,14 +258,14 @@ cblas_dnrm22( const long int N, const double *X, const int incX) 


  if (N <= 0 || incX <= 0) return 0;

  else if (N == 1)         return fabs(X[0]);

#pragma omp parallel private(ix)

{

#ifdef OMP

	tid = omp_get_thread_num();

        nthreads = omp_get_num_threads();

        if(nthreads>100) exit(1);

#endif

#pragma omp for

///#pragma omp parallel private(ix)

////{

////#ifdef OMP

////	tid = omp_get_thread_num();

////        nthreads = omp_get_num_threads();

////        if(nthreads>100) exit(1);

////#endif

////#pragma omp for

  for (i = 0; i < N; i++) {

  	  ix=incX*i;

      const double x = X[ix];
@@ -269,7 +284,7 @@ cblas_dnrm22( const long int N, const double *X, const int incX) 
  }

  

  resultSqr[tid]=(scale[tid] * sqrt(ssq[tid]))*(scale[tid] * sqrt(ssq[tid]));

}// pragma

////}// pragma



double resultFinal=0.0;

for(int j=0;j<100;j++) resultFinal+=resultSqr[j];
@@ -301,6 +316,7 @@ void cblas_dscal(const long int N, const double alpha, double *X, const int incX 
    }

}



//FV_ not used ??

void

cblas_dscal1(const long int N, const double alpha, double *X, const int incX)

{
@@ -312,14 +328,14 @@ cblas_dscal1(const long int N, const double alpha, double *X, const int incX) 
    if (incX > 0) ix=0;

    else ix=((N) - 1) * (-(incX));

    

#pragma omp parallel shared(ix)

{

////#pragma omp parallel shared(ix)

/////{

   long ixThread;

   long ixThreadOffset=ix;

#pragma omp for

/////#pragma omp for

    for (i = 0; i < N; i++) {

    	ixThread=ixThreadOffset+incX*i;

        X[ixThread] *= alpha;

    }

}// end pragma    

/////}// end pragma    

}

solvergaiaSim.c

+929 −498

File changed.

Preview size limit exceeded, changes collapsed.