Commit 4832b0ed authored by D'azevedo, Ed's avatar D'azevedo, Ed
Browse files

Merge branch 'master' into use_magma_timing

parents a590de66 06a10210

src/BatchedGemm.h

+1 −0
Original line number Diff line number Diff line
@@ -3,6 +3,7 @@ 
#include <stdexcept>



#include "dmrg_types.h"

#include "estimate_work.h"



template<typename T>

class BatchedGemm {

src/Makefile

+1 −0
Original line number Diff line number Diff line
@@ -12,6 +12,7 @@ OBJS=\ 
	setup_sparse_batch.o \

        unsetup_sparse_batch.o \

        apply_Htarget_sparse.o \

        estimate_work.o \

        BatchedGemm.o



libdmrgppPluginSc.a: $(OBJS)

src/apply_Htarget_sparse.c

+8 −2
Original line number Diff line number Diff line
@@ -103,7 +103,10 @@ void apply_Htarget_sparse( 
  if (need_allocate_X) {

    X_ = (FpType *) dmrg_malloc( sizeof(FpType) * xy_size_dim );

    assert( X_ != NULL );

    memcpy( &(X_[0]), &(Xin_[0]), sizeof(FpType) * xy_size );

    void *dest = (void *) &(X_[0]);

    void *src = (void *) &(Xin_[0]);

    size_t count = sizeof(FpType) * xy_size;

    dmrg_memcpy( dest, src, count );

    };



  if (need_allocate_Y) {
@@ -501,7 +504,10 @@ void apply_Htarget_sparse( 
    dmrg_free( X_ ); X_ = NULL;

    };

 if (need_allocate_Y) {

   memcpy( &(Yout_[0]), &(Y_[0]), sizeof(FpType) * xy_size );

   void *dest = &(Yout_[0]);

   void *src = &(Y_[0]);

   size_t count = sizeof(FpType) * xy_size;

   dmrg_memcpy( dest, src, count );

   dmrg_free( Y_ ); Y_ = NULL;

   };

#endif

src/cal_kron_flops.c

+2 −1
Original line number Diff line number Diff line
@@ -23,7 +23,8 @@ void cal_kron_flops( 
    double flops_BX = (2.0*nrowB) * ncolB * ncolX;

    

    int ncolBX = ncolX;

    double flops_BX_At =  (2.0*nrowY) * ncolY * ncolBX;

    int nrowBX = nrowB;

    double flops_BX_At =  (2.0*nrowBX) * ncolBX * nrowA;

    double flops_method1 = flops_BX + flops_BX_At;

    

    /*

src/dmrg_malloc.c

+12 −0
Original line number Diff line number Diff line
@@ -20,6 +20,18 @@ int dmrg_is_managed( const void *ptr ) 
  return( is_managed );

}

  

void dmrg_memcpy(void *dest, const void *src, size_t count)

{

#ifdef USE_MAGMA

  cudaError_t istat = cudaMemcpy( dest, src, count, cudaMemcpyDefault );

  if (istat != cudaSuccess) {

    fprintf(stderr,"dmrg_memcpy: %s\n", cudaGetErrorString(istat));

    };

  assert( istat == cudaSuccess );

#else

  memcpy( dest, src, count );

#endif

}



void *dmrg_malloc(const size_t alloc_size )

{