/usr/local/lib64/python3.6/site-packages/torch/include/ATen/native
Edit: /usr/local/lib64/python3.6/site-packages/torch/include/ATen/native/CPUBlas.h (4199B)
#pragma once
#include
#include
#include
#include
#include
namespace at {
namespace native {
namespace cpublas {
namespace internal {
void normalize_last_dims(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
int64_t *lda, int64_t *ldb, int64_t *ldc);
} // namespace internal
using gemm_fn = void(*)(
at::ScalarType type,
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
const Scalar& alpha,
const void *a, int64_t lda,
const void *b, int64_t ldb,
const Scalar& beta,
void *c, int64_t ldc);
DECLARE_DISPATCH(gemm_fn, gemm_stub);
template
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
scalar_t alpha,
const scalar_t *a, int64_t lda,
const scalar_t *b, int64_t ldb,
scalar_t beta,
scalar_t *c, int64_t ldc) {
internal::normalize_last_dims(transa, transb, m, n, k, &lda, &ldb, &ldc);
gemm_stub(
kCPU, c10::CppTypeToScalarType::value,
transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc);
}
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
double alpha,
const double *a, int64_t lda,
const double *b, int64_t ldb,
double beta,
double *c, int64_t ldc);
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
float alpha,
const float *a, int64_t lda,
const float *b, int64_t ldb,
float beta,
float *c, int64_t ldc);
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
c10::complex alpha,
const c10::complex *a, int64_t lda,
const c10::complex *b, int64_t ldb,
c10::complex beta,
c10::complex *c, int64_t ldc);
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
c10::complex alpha,
const c10::complex *a, int64_t lda,
const c10::complex *b, int64_t ldb,
c10::complex beta,
c10::complex *c, int64_t ldc);
void gemm(
TransposeType transa, TransposeType transb,
int64_t m, int64_t n, int64_t k,
int64_t alpha,
const int64_t *a, int64_t lda,
const int64_t *b, int64_t ldb,
int64_t beta,
int64_t *c, int64_t ldc);
using axpy_fn = void(*)(at::ScalarType type, int64_t n, const Scalar& a, const void *x, int64_t incx, void *y, int64_t incy);
DECLARE_DISPATCH(axpy_fn, axpy_stub);
template
void axpy(int64_t n, scalar_t a, const scalar_t *x, int64_t incx, scalar_t *y, int64_t incy){
if(n == 1)
{
incx = 1;
incy = 1;
}
axpy_stub(
kCPU, c10::CppTypeToScalarType::value,
n, a, x, incx, y, incy);
}
void axpy(int64_t n, double a, const double *x, int64_t incx, double *y, int64_t incy);
void axpy(int64_t n, float a, const float *x, int64_t incx, float *y, int64_t incy);
void axpy(int64_t n, c10::complex a, const c10::complex *x, int64_t incx, c10::complex *y, int64_t incy);
void axpy(int64_t n, c10::complex a, const c10::complex *x, int64_t incx, c10::complex *y, int64_t incy);
using copy_fn = void(*)(at::ScalarType type, int64_t n, const void *x, int64_t incx, void *y, int64_t incy);
DECLARE_DISPATCH(copy_fn, copy_stub);
template
void copy(int64_t n, const scalar_t *x, int64_t incx, scalar_t *y, int64_t incy) {
if(n == 1)
{
incx = 1;
incy = 1;
}
copy_stub(
kCPU, c10::CppTypeToScalarType::value,
n, x, incx, y, incy);
}
void copy(int64_t n, const double *x, int64_t incx, double *y, int64_t incy);
void copy(int64_t n, const float *x, int64_t incx, float *y, int64_t incy);
void copy(int64_t n, const c10::complex *x, int64_t incx, c10::complex *y, int64_t incy);
void copy(int64_t n, const c10::complex *x, int64_t incx, c10::complex *y, int64_t incy);
}}} // namespace at::native::cpublas