portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-10-31 03:54:33 +00:00
Code Releases Activity

Eigen implementation and SYCL implementation

Browse Source
This commit is contained in:
Peter Boyle
2024-07-25 18:02:56 +00:00
parent 6ae52da571
commit 18d2d7da4a
1 changed files with 336 additions and 176 deletions
Download Patch File Download Diff File Expand all files Collapse all files

494
Grid/algorithms/blas/BatchedBlas.h
View File

@@ -89,9 +89,10 @@ public:
gridblasHandle = theGridAccelerator; gridblasHandle = theGridAccelerator;
#endif #endif
#ifdef GRID_ONE_MKL #ifdef GRID_ONE_MKL
cl::sycl::cpu_selector selector; cl::sycl::gpu_selector selector;
cl::sycl::device selectedDevice { selector }; cl::sycl::device selectedDevice { selector };
gridblasHandle =new sycl::queue (selectedDevice); cl::sycl::property_list q_prop{cl::sycl::property::queue::in_order()};
gridblasHandle =new sycl::queue (selectedDevice,q_prop);
#endif #endif
gridblasInit=1; gridblasInit=1;
} }
@@ -207,6 +208,9 @@ public:
assert(Bkn.size()==batchCount); assert(Bkn.size()==batchCount);
assert(Cmn.size()==batchCount); assert(Cmn.size()==batchCount);
assert(OpA!=GridBLAS_OP_T); // Complex case expect no transpose
assert(OpB!=GridBLAS_OP_T);
int lda = m; // m x k column major int lda = m; // m x k column major
int ldb = k; // k x n column major int ldb = k; // k x n column major
int ldc = m; // m x b column major int ldc = m; // m x b column major
@@ -266,26 +270,131 @@ public:
assert(err==CUBLAS_STATUS_SUCCESS); assert(err==CUBLAS_STATUS_SUCCESS);
#endif #endif
#ifdef GRID_SYCL #ifdef GRID_SYCL
//MKL’s cblas_<T>gemm_batch & OneAPI // std::cerr << " Calling SYCL batched ZGEMM "<<std::endl;
#warning "oneMKL implementation not built " int64_t m64=m;
#endif int64_t n64=n;
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP) int64_t k64=k;
// Need a default/reference implementation int64_t lda64=lda;
int sda = lda*k; int64_t ldb64=ldb;
int sdb = ldb*k; int64_t ldc64=ldc;
int sdc = ldc*n; int64_t batchCount64=batchCount;
for (int p = 0; p < batchCount; ++p) {
oneapi::mkl::transpose iOpA;
oneapi::mkl::transpose iOpB;
if ( OpA == GridBLAS_OP_N ) iOpA = oneapi::mkl::transpose::N;
if ( OpA == GridBLAS_OP_T ) iOpA = oneapi::mkl::transpose::T;
if ( OpA == GridBLAS_OP_C ) iOpA = oneapi::mkl::transpose::C;
if ( OpB == GridBLAS_OP_N ) iOpB = oneapi::mkl::transpose::N;
if ( OpB == GridBLAS_OP_T ) iOpB = oneapi::mkl::transpose::T;
if ( OpB == GridBLAS_OP_C ) iOpB = oneapi::mkl::transpose::C;
oneapi::mkl::blas::column_major::gemm_batch(*gridblasHandle,
&iOpA,
&iOpB,
&m64,&n64,&k64,
(ComplexD *) &alpha_p[0],
(const ComplexD **)&Amk[0], (const int64_t *)&lda64,
(const ComplexD **)&Bkn[0], (const int64_t *)&ldb64,
(ComplexD *) &beta_p[0],
(ComplexD **)&Cmn[0], (const int64_t *)&ldc64,
(int64_t)1,&batchCount64,std::vector<sycl::event>());
synchronise();
#if 0
// This code was used to check the mat mul on Sunspot/OneMKL
std::cerr << " Called SYCL batched ZGEMM OpA "<< OpA << " OpB "<<OpB <<std::endl;
std::vector<ComplexD> A(m*k); // pointer list to matrices
std::vector<ComplexD> B(k*n);
std::vector<ComplexD> C(m*n);
// int sda = lda*k;
// int sdb = ldb*k;
// int sdc = ldc*n;
std::cerr << " Checking the GEMM results "<<std::endl;
for (int p = 0; p < 1; ++p) {
ComplexD * Amk_p; // pointer list to matrices
ComplexD * Bkn_p; // pointer list to matrices
ComplexD * Cmn_p; // pointer list to matrices
acceleratorCopyFromDevice((void *)&Amk[p],(void *)&Amk_p,sizeof(ComplexD*));
acceleratorCopyFromDevice((void *)&Bkn[p],(void *)&Bkn_p,sizeof(ComplexD*));
acceleratorCopyFromDevice((void *)&Cmn[p],(void *)&Cmn_p,sizeof(ComplexD*));
std::cerr << " p " << p << " copied pointers "<<std::endl;
acceleratorCopyFromDevice((void *)Amk_p,(void *)&A[0],m*k*sizeof(ComplexD));
acceleratorCopyFromDevice((void *)Bkn_p,(void *)&B[0],k*n*sizeof(ComplexD));
acceleratorCopyFromDevice((void *)Cmn_p,(void *)&C[0],m*n*sizeof(ComplexD));
std::cerr << " p " << p << " copied matrices "<<std::endl;
std::cerr << " C[0] "<<C[0]<<std::endl;
std::cerr << " A[0] "<<A[0]<<std::endl;
std::cerr << " B[0] "<<B[0]<<std::endl;
std::cerr << " m "<<m<<std::endl;
std::cerr << " n "<<n<<std::endl;
std::cerr << " k "<<k<<std::endl;
for (int mm = 0; mm < m; ++mm) { for (int mm = 0; mm < m; ++mm) {
for (int nn = 0; nn < n; ++nn) { for (int nn = 0; nn < n; ++nn) {
ComplexD c_mn(0.0); ComplexD c_mn(0.0);
for (int kk = 0; kk < k; ++kk) for (int kk = 0; kk < k; ++kk) {
c_mn += Amk[p][mm + kk*lda ] * Bkn[p][kk + nn*ldb]; int idx_a, idx_b;
Cmn[p][mm + nn*ldc] = (alpha)*c_mn + (beta)*Cmn[p][mm + nn*ldc ]; // int lda = m; // m x k column major
// int ldb = k; // k x n column major
// int ldc = m; // m x b column major
if(OpA!=GridBLAS_OP_N) {
idx_a =kk + mm*lda;
} else {
idx_a =mm + kk*lda;
}
if(OpB!=GridBLAS_OP_N) {
idx_b =nn + kk*ldb;
} else {
idx_b =kk + nn*ldb;
}
// std::cerr << " idx_a "<<idx_a<<" idx_b "<<idx_b<<std::endl;
ComplexD Ac = A[idx_a];
ComplexD Bc = B[idx_b];
if(OpA==GridBLAS_OP_C) Ac = conjugate(Ac);
if(OpB==GridBLAS_OP_C) Bc = conjugate(Bc);
c_mn += Ac*Bc;
}
std::cerr << " beta "<<beta<<" alpha "<<alpha<<" C_"<<mm<<","<<nn<<" "<<c_mn<<" "<<C[mm + nn*ldc]<<std::endl;
} }
} }
} }
#endif #endif
// synchronise(); #endif
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP)
// Need a default/reference implementation; use Eigen
if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_N) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcd> eAmk(Amk[p],m,k);
Eigen::Map<Eigen::MatrixXcd> eBkn(Bkn[p],k,n);
Eigen::Map<Eigen::MatrixXcd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk * eBkn ;
});
} else if ( (OpA == GridBLAS_OP_C ) && (OpB == GridBLAS_OP_N) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcd> eAmk(Amk[p],k,m);
Eigen::Map<Eigen::MatrixXcd> eBkn(Bkn[p],k,n);
Eigen::Map<Eigen::MatrixXcd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.adjoint() * eBkn ;
});
} else if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_C) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcd> eAmk(Amk[p],m,k);
Eigen::Map<Eigen::MatrixXcd> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXcd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk * eBkn.adjoint() ;
});
} else if ( (OpA == GridBLAS_OP_C ) && (OpB == GridBLAS_OP_C) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcd> eAmk(Amk[p],k,m);
Eigen::Map<Eigen::MatrixXcd> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXcd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.adjoint() * eBkn.adjoint() ;
} );
} else {
assert(0);
}
#endif
RealD t1=usecond(); RealD t1=usecond();
RealD flops = 8.0*m*n*k*batchCount; RealD flops = 8.0*m*n*k*batchCount;
RealD bytes = 1.0*sizeof(ComplexD)*(m*k+k*n+m*n)*batchCount; RealD bytes = 1.0*sizeof(ComplexD)*(m*k+k*n+m*n)*batchCount;
@@ -306,6 +415,9 @@ public:
RealD t2=usecond(); RealD t2=usecond();
int32_t batchCount = Amk.size(); int32_t batchCount = Amk.size();
assert(OpA!=GridBLAS_OP_T); // Complex case expect no transpose
assert(OpB!=GridBLAS_OP_T);
int lda = m; // m x k column major int lda = m; // m x k column major
int ldb = k; // k x n column major int ldb = k; // k x n column major
int ldc = m; // m x b column major int ldc = m; // m x b column major
@@ -366,25 +478,68 @@ public:
assert(err==CUBLAS_STATUS_SUCCESS); assert(err==CUBLAS_STATUS_SUCCESS);
#endif #endif
#ifdef GRID_SYCL #ifdef GRID_SYCL
//MKL’s cblas_<T>gemm_batch & OneAPI int64_t m64=m;
#warning "oneMKL implementation not built " int64_t n64=n;
int64_t k64=k;
int64_t lda64=lda;
int64_t ldb64=ldb;
int64_t ldc64=ldc;
int64_t batchCount64=batchCount;
oneapi::mkl::transpose iOpA;
oneapi::mkl::transpose iOpB;
if ( OpA == GridBLAS_OP_N ) iOpA = oneapi::mkl::transpose::N;
if ( OpA == GridBLAS_OP_T ) iOpA = oneapi::mkl::transpose::T;
if ( OpA == GridBLAS_OP_C ) iOpA = oneapi::mkl::transpose::C;
if ( OpB == GridBLAS_OP_N ) iOpB = oneapi::mkl::transpose::N;
if ( OpB == GridBLAS_OP_T ) iOpB = oneapi::mkl::transpose::T;
if ( OpB == GridBLAS_OP_C ) iOpB = oneapi::mkl::transpose::C;
oneapi::mkl::blas::column_major::gemm_batch(*gridblasHandle,
&iOpA,
&iOpB,
&m64,&n64,&k64,
(ComplexF *) &alpha_p[0],
(const ComplexF **)&Amk[0], (const int64_t *)&lda64,
(const ComplexF **)&Bkn[0], (const int64_t *)&ldb64,
(ComplexF *) &beta_p[0],
(ComplexF **)&Cmn[0], (const int64_t *)&ldc64,
(int64_t)1,&batchCount64,std::vector<sycl::event>());
synchronise();
#endif #endif
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP) #if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP)
int sda = lda*k; // Need a default/reference implementation; use Eigen
int sdb = ldb*k; if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_N) ) {
int sdc = ldc*n; thread_for (p, batchCount, {
ComplexF alphaf(real(alpha),imag(alpha)); Eigen::Map<Eigen::MatrixXcf> eAmk(Amk[p],m,k);
ComplexF betaf(real(beta),imag(beta)); Eigen::Map<Eigen::MatrixXcf> eBkn(Bkn[p],k,n);
// Need a default/reference implementation Eigen::Map<Eigen::MatrixXcf> eCmn(Cmn[p],m,n);
for (int p = 0; p < batchCount; ++p) { eCmn = beta * eCmn + alpha * eAmk * eBkn ;
for (int mm = 0; mm < m; ++mm) { });
for (int nn = 0; nn < n; ++nn) { } else if ( (OpA == GridBLAS_OP_C ) && (OpB == GridBLAS_OP_N) ) {
ComplexF c_mn(0.0); thread_for (p, batchCount, {
for (int kk = 0; kk < k; ++kk) Eigen::Map<Eigen::MatrixXcf> eAmk(Amk[p],k,m);
c_mn += Amk[p][mm + kk*lda ] * Bkn[p][kk + nn*ldb]; Eigen::Map<Eigen::MatrixXcf> eBkn(Bkn[p],k,n);
Cmn[p][mm + nn*ldc] = (alphaf)*c_mn + (betaf)*Cmn[p][mm + nn*ldc ]; Eigen::Map<Eigen::MatrixXcf> eCmn(Cmn[p],m,n);
} eCmn = beta * eCmn + alpha * eAmk.adjoint() * eBkn ;
} });
} else if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_C) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcf> eAmk(Amk[p],m,k);
Eigen::Map<Eigen::MatrixXcf> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXcf> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk * eBkn.adjoint() ;
});
} else if ( (OpA == GridBLAS_OP_C ) && (OpB == GridBLAS_OP_C) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXcf> eAmk(Amk[p],k,m);
Eigen::Map<Eigen::MatrixXcf> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXcf> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.adjoint() * eBkn.adjoint() ;
} );
} else {
assert(0);
} }
#endif #endif
RealD t1=usecond(); RealD t1=usecond();
@@ -408,6 +563,9 @@ public:
RealD t2=usecond(); RealD t2=usecond();
int32_t batchCount = Amk.size(); int32_t batchCount = Amk.size();
assert(OpA!=GridBLAS_OP_C); // Real case no conjugate
assert(OpB!=GridBLAS_OP_C);
int lda = m; // m x k column major int lda = m; // m x k column major
int ldb = k; // k x n column major int ldb = k; // k x n column major
int ldc = m; // m x b column major int ldc = m; // m x b column major
@@ -467,23 +625,68 @@ public:
assert(err==CUBLAS_STATUS_SUCCESS); assert(err==CUBLAS_STATUS_SUCCESS);
#endif #endif
#ifdef GRID_SYCL #ifdef GRID_SYCL
//MKL’s cblas_<T>gemm_batch & OneAPI int64_t m64=m;
#warning "oneMKL implementation not built " int64_t n64=n;
int64_t k64=k;
int64_t lda64=lda;
int64_t ldb64=ldb;
int64_t ldc64=ldc;
int64_t batchCount64=batchCount;
oneapi::mkl::transpose iOpA;
oneapi::mkl::transpose iOpB;
if ( OpA == GridBLAS_OP_N ) iOpA = oneapi::mkl::transpose::N;
if ( OpA == GridBLAS_OP_T ) iOpA = oneapi::mkl::transpose::T;
if ( OpA == GridBLAS_OP_C ) iOpA = oneapi::mkl::transpose::C;
if ( OpB == GridBLAS_OP_N ) iOpB = oneapi::mkl::transpose::N;
if ( OpB == GridBLAS_OP_T ) iOpB = oneapi::mkl::transpose::T;
if ( OpB == GridBLAS_OP_C ) iOpB = oneapi::mkl::transpose::C;
oneapi::mkl::blas::column_major::gemm_batch(*gridblasHandle,
&iOpA,
&iOpB,
&m64,&n64,&k64,
(float *) &alpha_p[0],
(const float **)&Amk[0], (const int64_t *)&lda64,
(const float **)&Bkn[0], (const int64_t *)&ldb64,
(float *) &beta_p[0],
(float **)&Cmn[0], (const int64_t *)&ldc64,
(int64_t)1,&batchCount64,std::vector<sycl::event>());
synchronise();
#endif #endif
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP) #if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP)
int sda = lda*k; // Need a default/reference implementation; use Eigen
int sdb = ldb*k; if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_N) ) {
int sdc = ldc*n; thread_for (p, batchCount, {
// Need a default/reference implementation Eigen::Map<Eigen::MatrixXf> eAmk(Amk[p],m,k);
for (int p = 0; p < batchCount; ++p) { Eigen::Map<Eigen::MatrixXf> eBkn(Bkn[p],k,n);
for (int mm = 0; mm < m; ++mm) { Eigen::Map<Eigen::MatrixXf> eCmn(Cmn[p],m,n);
for (int nn = 0; nn < n; ++nn) { eCmn = beta * eCmn + alpha * eAmk * eBkn ;
RealD c_mn(0.0); });
for (int kk = 0; kk < k; ++kk) } else if ( (OpA == GridBLAS_OP_T ) && (OpB == GridBLAS_OP_N) ) {
c_mn += Amk[p][mm + kk*lda ] * Bkn[p][kk + nn*ldb]; thread_for (p, batchCount, {
Cmn[p][mm + nn*ldc] = (alpha)*c_mn + (beta)*Cmn[p][mm + nn*ldc ]; Eigen::Map<Eigen::MatrixXf> eAmk(Amk[p],k,m);
} Eigen::Map<Eigen::MatrixXf> eBkn(Bkn[p],k,n);
} Eigen::Map<Eigen::MatrixXf> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.transpose() * eBkn ;
});
} else if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_T) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXf> eAmk(Amk[p],m,k);
Eigen::Map<Eigen::MatrixXf> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXf> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk * eBkn.transpose() ;
});
} else if ( (OpA == GridBLAS_OP_T ) && (OpB == GridBLAS_OP_T) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXf> eAmk(Amk[p],k,m);
Eigen::Map<Eigen::MatrixXf> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXf> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.transpose() * eBkn.transpose() ;
} );
} else {
assert(0);
} }
#endif #endif
RealD t1=usecond(); RealD t1=usecond();
@@ -495,7 +698,6 @@ public:
/////////////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////////////
// Double precision real GEMM // Double precision real GEMM
/////////////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////////////
void gemmBatched(GridBLASOperation_t OpA, void gemmBatched(GridBLASOperation_t OpA,
GridBLASOperation_t OpB, GridBLASOperation_t OpB,
int m,int n, int k, int m,int n, int k,
@@ -508,6 +710,9 @@ public:
RealD t2=usecond(); RealD t2=usecond();
int32_t batchCount = Amk.size(); int32_t batchCount = Amk.size();
assert(OpA!=GridBLAS_OP_C); // Real case no conjugate
assert(OpB!=GridBLAS_OP_C);
int lda = m; // m x k column major int lda = m; // m x k column major
int ldb = k; // k x n column major int ldb = k; // k x n column major
int ldc = m; // m x b column major int ldc = m; // m x b column major
@@ -568,39 +773,68 @@ public:
assert(err==CUBLAS_STATUS_SUCCESS); assert(err==CUBLAS_STATUS_SUCCESS);
#endif #endif
#ifdef GRID_SYCL #ifdef GRID_SYCL
/*
int64_t m64=m; int64_t m64=m;
int64_t n64=n; int64_t n64=n;
int64_t k64=k; int64_t k64=k;
int64_t lda64=lda;
int64_t ldb64=ldb;
int64_t ldc64=ldc;
int64_t batchCount64=batchCount; int64_t batchCount64=batchCount;
oneapi::mkl::blas::column_major::gemm_batch(*theGridAccelerator,
onemkl::transpose::N, oneapi::mkl::transpose iOpA;
onemkl::transpose::N, oneapi::mkl::transpose iOpB;
if ( OpA == GridBLAS_OP_N ) iOpA = oneapi::mkl::transpose::N;
if ( OpA == GridBLAS_OP_T ) iOpA = oneapi::mkl::transpose::T;
if ( OpA == GridBLAS_OP_C ) iOpA = oneapi::mkl::transpose::C;
if ( OpB == GridBLAS_OP_N ) iOpB = oneapi::mkl::transpose::N;
if ( OpB == GridBLAS_OP_T ) iOpB = oneapi::mkl::transpose::T;
if ( OpB == GridBLAS_OP_C ) iOpB = oneapi::mkl::transpose::C;
oneapi::mkl::blas::column_major::gemm_batch(*gridblasHandle,
&iOpA,
&iOpB,
&m64,&n64,&k64, &m64,&n64,&k64,
(double *) &alpha_p[0], (double *) &alpha_p[0],
(double **)&Amk[0], lda, (const double **)&Amk[0], (const int64_t *)&lda64,
(double **)&Bkn[0], ldb, (const double **)&Bkn[0], (const int64_t *)&ldb64,
(double *) &beta_p[0], (double *) &beta_p[0],
(double **)&Cmn[0], ldc, (double **)&Cmn[0], (const int64_t *)&ldc64,
1,&batchCount64); (int64_t)1,&batchCount64,std::vector<sycl::event>());
*/ synchronise();
//MKL’s cblas_<T>gemm_batch & OneAPI
#warning "oneMKL implementation not built "
#endif #endif
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP) #if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP)
int sda = lda*k; // Need a default/reference implementation; use Eigen
int sdb = ldb*k; if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_N) ) {
int sdc = ldc*n; thread_for (p, batchCount, {
// Need a default/reference implementation Eigen::Map<Eigen::MatrixXd> eAmk(Amk[p],m,k);
for (int p = 0; p < batchCount; ++p) { Eigen::Map<Eigen::MatrixXd> eBkn(Bkn[p],k,n);
for (int mm = 0; mm < m; ++mm) { Eigen::Map<Eigen::MatrixXd> eCmn(Cmn[p],m,n);
for (int nn = 0; nn < n; ++nn) { eCmn = beta * eCmn + alpha * eAmk * eBkn ;
RealD c_mn(0.0); });
for (int kk = 0; kk < k; ++kk) } else if ( (OpA == GridBLAS_OP_T ) && (OpB == GridBLAS_OP_N) ) {
c_mn += Amk[p][mm + kk*lda ] * Bkn[p][kk + nn*ldb]; thread_for (p, batchCount, {
Cmn[p][mm + nn*ldc] = (alpha)*c_mn + (beta)*Cmn[p][mm + nn*ldc ]; Eigen::Map<Eigen::MatrixXd> eAmk(Amk[p],k,m);
} Eigen::Map<Eigen::MatrixXd> eBkn(Bkn[p],k,n);
} Eigen::Map<Eigen::MatrixXd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.transpose() * eBkn ;
});
} else if ( (OpA == GridBLAS_OP_N ) && (OpB == GridBLAS_OP_T) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXd> eAmk(Amk[p],m,k);
Eigen::Map<Eigen::MatrixXd> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk * eBkn.transpose() ;
});
} else if ( (OpA == GridBLAS_OP_T ) && (OpB == GridBLAS_OP_T) ) {
thread_for (p, batchCount, {
Eigen::Map<Eigen::MatrixXd> eAmk(Amk[p],k,m);
Eigen::Map<Eigen::MatrixXd> eBkn(Bkn[p],n,k);
Eigen::Map<Eigen::MatrixXd> eCmn(Cmn[p],m,n);
eCmn = beta * eCmn + alpha * eAmk.transpose() * eBkn.transpose() ;
});
} else {
assert(0);
} }
#endif #endif
RealD t1=usecond(); RealD t1=usecond();
@@ -608,120 +842,46 @@ public:
RealD bytes = 1.0*sizeof(RealD)*(m*k+k*n+m*n)*batchCount; RealD bytes = 1.0*sizeof(RealD)*(m*k+k*n+m*n)*batchCount;
} }
template<class CComplex>
////////////////////////////////////////////////////////////////////////////////////////////////
// Strided case used by benchmark, but generally unused in Grid
// Keep a code example in double complex, but don't generate the single and real variants for now
////////////////////////////////////////////////////////////////////////////////////////////////
void gemmStridedBatched(int m,int n, int k,
ComplexD alpha,
ComplexD* Amk, // pointer list to matrices
ComplexD* Bkn,
ComplexD beta,
ComplexD* Cmn,
int batchCount)
{
// Use C-row major storage, so transpose calls
int lda = m; // m x k column major
int ldb = k; // k x n column major
int ldc = m; // m x b column major
int sda = m*k;
int sdb = k*n;
int sdc = m*n;
deviceVector<ComplexD> alpha_p(1);
deviceVector<ComplexD> beta_p(1);
acceleratorCopyToDevice((void *)&alpha,(void *)&alpha_p[0],sizeof(ComplexD));
acceleratorCopyToDevice((void *)&beta ,(void *)&beta_p[0],sizeof(ComplexD));
// std::cout << "blasZgemmStridedBatched mnk "<<m<<","<<n<<","<<k<<" count "<<batchCount<<std::endl;
// std::cout << "blasZgemmStridedBatched ld "<<lda<<","<<ldb<<","<<ldc<<std::endl;
// std::cout << "blasZgemmStridedBatched sd "<<sda<<","<<sdb<<","<<sdc<<std::endl;
#ifdef GRID_HIP
auto err = hipblasZgemmStridedBatched(gridblasHandle,
HIPBLAS_OP_N,
HIPBLAS_OP_N,
m,n,k,
(hipblasDoubleComplex *) &alpha_p[0],
(hipblasDoubleComplex *) Amk, lda, sda,
(hipblasDoubleComplex *) Bkn, ldb, sdb,
(hipblasDoubleComplex *) &beta_p[0],
(hipblasDoubleComplex *) Cmn, ldc, sdc,
batchCount);
assert(err==HIPBLAS_STATUS_SUCCESS);
#endif
#ifdef GRID_CUDA
cublasZgemmStridedBatched(gridblasHandle,
CUBLAS_OP_N,
CUBLAS_OP_N,
m,n,k,
(cuDoubleComplex *) &alpha_p[0],
(cuDoubleComplex *) Amk, lda, sda,
(cuDoubleComplex *) Bkn, ldb, sdb,
(cuDoubleComplex *) &beta_p[0],
(cuDoubleComplex *) Cmn, ldc, sdc,
batchCount);
#endif
#if defined(GRID_SYCL) || defined(GRID_ONE_MKL)
oneapi::mkl::blas::column_major::gemm_batch(*gridblasHandle,
oneapi::mkl::transpose::N,
oneapi::mkl::transpose::N,
m,n,k,
alpha,
(const ComplexD *)Amk,lda,sda,
(const ComplexD *)Bkn,ldb,sdb,
beta,
(ComplexD *)Cmn,ldc,sdc,
batchCount);
#endif
#if !defined(GRID_SYCL) && !defined(GRID_CUDA) && !defined(GRID_HIP) && !defined(GRID_ONE_MKL)
// Need a default/reference implementation
for (int p = 0; p < batchCount; ++p) {
for (int mm = 0; mm < m; ++mm) {
for (int nn = 0; nn < n; ++nn) {
ComplexD c_mn(0.0);
for (int kk = 0; kk < k; ++kk)
c_mn += Amk[mm + kk*lda + p*sda] * Bkn[kk + nn*ldb + p*sdb];
Cmn[mm + nn*ldc + p*sdc] = (alpha)*c_mn + (beta)*Cmn[mm + nn*ldc + p*sdc];
}
}
}
#endif
}
double benchmark(int M, int N, int K, int BATCH) double benchmark(int M, int N, int K, int BATCH)
{ {
int32_t N_A = M*K*BATCH; int32_t N_A = M*K*BATCH;
int32_t N_B = K*N*BATCH; int32_t N_B = K*N*BATCH;
int32_t N_C = M*N*BATCH; int32_t N_C = M*N*BATCH;
deviceVector<ComplexD> A(N_A); acceleratorMemSet(&A[0],0,N_A*sizeof(ComplexD)); deviceVector<CComplex> A(N_A); acceleratorMemSet(&A[0],0,N_A*sizeof(CComplex));
deviceVector<ComplexD> B(N_B); acceleratorMemSet(&B[0],0,N_B*sizeof(ComplexD)); deviceVector<CComplex> B(N_B); acceleratorMemSet(&B[0],0,N_B*sizeof(CComplex));
deviceVector<ComplexD> C(N_C); acceleratorMemSet(&C[0],0,N_C*sizeof(ComplexD)); deviceVector<CComplex> C(N_C); acceleratorMemSet(&C[0],0,N_C*sizeof(CComplex));
ComplexD alpha(1.0); CComplex alpha(1.0);
ComplexD beta (1.0); CComplex beta (1.0);
RealD flops = 8.0*M*N*K*BATCH; RealD flops = 8.0*M*N*K*BATCH;
int ncall=10; int ncall=10;
RealD t0 = usecond(); RealD t0 = usecond();
deviceVector<CComplex *> As(BATCH);
deviceVector<CComplex *> Bs(BATCH);
deviceVector<CComplex *> Cs(BATCH);
for(int b = 0 ; b < BATCH;b++) {
CComplex *ptr;
ptr = &A[b*M*K]; acceleratorPut(As[b],ptr);
ptr = &B[b*K*N]; acceleratorPut(Bs[b],ptr);
ptr = &C[b*M*N]; acceleratorPut(Cs[b],ptr);
}
for(int i=0;i<ncall;i++){ for(int i=0;i<ncall;i++){
gemmStridedBatched(M,N,K, gemmBatched(M,N,K,
alpha, alpha,
&A[0], // m x k As, // m x k
&B[0], // k x n Bs, // k x n
beta, beta,
&C[0], // m x n Cs);
BATCH);
} }
synchronise(); synchronise();
RealD t1 = usecond(); RealD t1 = usecond();
RealD bytes = 1.0*sizeof(ComplexD)*(M*N*2+N*K+M*K)*BATCH; RealD bytes = 1.0*sizeof(CComplex)*(M*N*2+N*K+M*K)*BATCH;
flops = 8.0*M*N*K*BATCH*ncall; flops = 8.0*M*N*K*BATCH*ncall;
flops = flops/(t1-t0)/1.e3; flops = flops/(t1-t0)/1.e3;
return flops; // Returns gigaflops return flops; // Returns gigaflops
} }
}; };
NAMESPACE_END(Grid); NAMESPACE_END(Grid);