#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/OpApiInterface.h"
#include "op_plugin/utils/op_api_common.h"
#include "op_plugin/utils/OpUtils.h"
namespace op_api {
using npu_preparation = at_npu::native::OpPreparation;
at::Tensor mm(const at::Tensor &self, const at::Tensor &mat2, const at::ScalarType output_dtype)
{
TORCH_CHECK(self.dim() == 2, "self must be a matrix");
TORCH_CHECK(mat2.dim() == 2, "mat2 must be a matrix");
TORCH_CHECK(
self.sizes()[1] == mat2.sizes()[0], "mat1 and mat2 shapes cannot be multiplied (",
self.sizes()[0], "x", self.sizes()[1], " and ", mat2.sizes()[0], "x", mat2.sizes()[1], ")"
);
auto names = at::namedinference::compute_matmul_outnames(self, mat2);
DO_MATMUL_COMPATIBILITY(aclnnMatmulWeightNz, aclnnMm, self, mat2, acl_op::mm(self, mat2));
auto output_size = {self.size(0), mat2.size(1)};
at::Tensor result = npu_preparation::apply_tensor_without_format(output_size, self.options().dtype(output_dtype));
int8_t cube_math_type = op_plugin::utils::get_cube_math_type_with_passthrough();
if (op_plugin::utils::is_nd_nz_format(self, mat2)) {
EXEC_NPU_CMD(aclnnMatmulWeightNz, self, mat2, result, cube_math_type);
} else {
int8_t cube_math_type_passthrough = npu_preparation::get_cube_math_type();
if (cube_math_type_passthrough >= 0) {
cube_math_type = cube_math_type_passthrough;
}
EXEC_NPU_CMD(aclnnMm, self, mat2, result, cube_math_type);
}
at::namedinference::propagate_names_if_nonempty(result, names);
FLOP_COUNT(FlopCounter::mm_flop, self, mat2);
return result;
}
at::Tensor mm(const at::Tensor &self, const at::Tensor &mat2)
{
TORCH_CHECK(self.dim() == 2, "self must be a matrix");
TORCH_CHECK(mat2.dim() == 2, "mat2 must be a matrix");
TORCH_CHECK(
self.sizes()[1] == mat2.sizes()[0], "mat1 and mat2 shapes cannot be multiplied (",
self.sizes()[0], "x", self.sizes()[1], " and ", mat2.sizes()[0], "x", mat2.sizes()[1], ")"
);
auto names = at::namedinference::compute_matmul_outnames(self, mat2);
DO_MATMUL_COMPATIBILITY(aclnnMatmulWeightNz, aclnnMm, self, mat2, acl_op::mm(self, mat2));
auto output_size = {self.size(0), mat2.size(1)};
at::Tensor result = npu_preparation::apply_tensor_without_format(output_size, self.options());
int8_t cube_math_type = op_plugin::utils::get_cube_math_type_with_passthrough();
if (op_plugin::utils::is_nd_nz_format(self, mat2)) {
EXEC_NPU_CMD(aclnnMatmulWeightNz, self, mat2, result, cube_math_type);
} else {
EXEC_NPU_CMD(aclnnMm, self, mat2, result, cube_math_type);
}
at::namedinference::propagate_names_if_nonempty(result, names);
FLOP_COUNT(FlopCounter::mm_flop, self, mat2);
return result;
}
at::Tensor &mm_out(
const at::Tensor &self,
const at::Tensor &mat2,
const at::ScalarType output_dtype,
at::Tensor &out)
{
TORCH_CHECK(self.dim() == 2, "self must be a matrix");
TORCH_CHECK(mat2.dim() == 2, "mat2 must be a matrix");
TORCH_CHECK(
self.sizes()[1] == mat2.sizes()[0], "mat1 and mat2 shapes cannot be multiplied (",
self.sizes()[0], "x", self.sizes()[1], " and ", mat2.sizes()[0], "x", mat2.sizes()[1], ")"
);
auto names = at::namedinference::compute_matmul_outnames(self, mat2);
DO_MATMUL_COMPATIBILITY(aclnnMatmulWeightNz, aclnnMm, self, mat2, acl_op::mm_out(self, mat2, out));
auto output_size = {self.size(0), mat2.size(1)};
npu_preparation::check_tensor({self, mat2}, out, output_dtype, output_size);
int8_t cube_math_type = op_plugin::utils::get_cube_math_type_with_passthrough();
if (op_plugin::utils::is_nd_nz_format(self, mat2)) {
EXEC_NPU_CMD(aclnnMatmulWeightNz, self, mat2, out, cube_math_type);
} else {
int8_t cube_math_type_passthrough = npu_preparation::get_cube_math_type();
if (cube_math_type_passthrough >= 0) {
cube_math_type = cube_math_type_passthrough;
}
EXEC_NPU_CMD(aclnnMm, self, mat2, out, cube_math_type);
}
at::namedinference::propagate_names_if_nonempty(out, names);
return out;
}
at::Tensor &mm_out(
const at::Tensor &self,
const at::Tensor &mat2,
at::Tensor &out)
{
TORCH_CHECK(self.dim() == 2, "self must be a matrix");
TORCH_CHECK(mat2.dim() == 2, "mat2 must be a matrix");
TORCH_CHECK(
self.sizes()[1] == mat2.sizes()[0], "mat1 and mat2 shapes cannot be multiplied (",
self.sizes()[0], "x", self.sizes()[1], " and ", mat2.sizes()[0], "x", mat2.sizes()[1], ")"
);
auto names = at::namedinference::compute_matmul_outnames(self, mat2);
DO_MATMUL_COMPATIBILITY(aclnnMatmulWeightNz, aclnnMm, self, mat2, acl_op::mm_out(self, mat2, out));
auto output_size = {self.size(0), mat2.size(1)};
npu_preparation::check_tensor({self, mat2}, out, self.scalar_type(), output_size);
int8_t cube_math_type = op_plugin::utils::get_cube_math_type_with_passthrough();
if (op_plugin::utils::is_nd_nz_format(self, mat2)) {
EXEC_NPU_CMD(aclnnMatmulWeightNz, self, mat2, out, cube_math_type);
} else {
EXEC_NPU_CMD(aclnnMm, self, mat2, out, cube_math_type);
}
at::namedinference::propagate_names_if_nonempty(out, names);
return out;
}
}