#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/OpApiInterface.h"
#include "op_plugin/utils/op_api_common.h"
namespace op_api {
using npu_preparation = at_npu::native::OpPreparation;
at::Tensor& ge_out(const at::Tensor& self, const at::Scalar& other, at::Tensor& result) {
DO_COMPATIBILITY(aclnnGeScalar, acl_op::ge_out(self, other, result));
auto output_size = self.sizes();
npu_preparation::check_tensor({self}, result, output_size);
EXEC_NPU_CMD(aclnnGeScalar, self, other, result);
return result;
}
at::Tensor ge(const at::Tensor& self, const at::Scalar& other) {
DO_COMPATIBILITY(aclnnGeScalar, acl_op::ge(self, other));
auto output_size = op_infer::input_same_output_size(self);
at::Tensor result = npu_preparation::apply_tensor_without_format(output_size, self.options().dtype(at::kBool));
EXEC_NPU_CMD(aclnnGeScalar, self, other, result);
return result;
}
at::Tensor& ge_(at::Tensor& self, const at::Scalar& other) {
DO_COMPATIBILITY(aclnnInplaceGeScalar, acl_op::ge_(self, other));
EXEC_NPU_CMD(aclnnInplaceGeScalar, self, other);
return self;
}
at::Tensor& ge_out(const at::Tensor& self, const at::Tensor& other, at::Tensor& result) {
DO_COMPATIBILITY(aclnnGeTensor, acl_op::ge_out(self, other, result));
auto output_size = op_infer::broadcast_ops_npu_output_size(self, other);
npu_preparation::check_tensor({self, other}, result, output_size);
EXEC_NPU_CMD(aclnnGeTensor, self, other, result);
return result;
}
at::Tensor ge(const at::Tensor& self, const at::Tensor& other) {
DO_COMPATIBILITY(aclnnGeTensor, acl_op::ge(self, other));
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
DO_COMPATIBILITY(aclnnGeScalar, acl_op::ge(self, other));
at::Tensor result = npu_preparation::apply_tensor_without_format(self.sizes(), self.options().dtype(at::kBool));
const at::Scalar tmpItem = other.item();
EXEC_NPU_CMD(aclnnGeScalar, self, tmpItem, result);
return result;
} else if (self.dim() == 0 && !torch_npu::utils::is_npu(self)) {
DO_COMPATIBILITY(aclnnLessScalar, acl_op::ge(self, other));
at::Tensor result = npu_preparation::apply_tensor_without_format(other.sizes(), other.options().dtype(at::kBool));
const at::Scalar tmpItem = self.item();
EXEC_NPU_CMD(aclnnLessScalar, other, tmpItem, result);
return result;
} else {
auto output_size = op_infer::broadcast_ops_npu_output_size(self, other);
at::Tensor result = npu_preparation::apply_tensor_without_format(output_size, self.options().dtype(at::kBool));
EXEC_NPU_CMD(aclnnGeTensor, self, other, result);
return result;
}
}
at::Tensor& ge_(at::Tensor &self, const at::Tensor &other)
{
DO_COMPATIBILITY(aclnnInplaceGeTensor, acl_op::ge_(self, other));
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
return op_api::ge_(self, other.item());
} else {
TORCH_CHECK(self.device() == other.device(),
"Expected all tensors to be on the same device, but found at least two devices", OPS_ERROR(ErrCode::INTERNAL));
npu_preparation::CheckMemory({self, other}, {self});
EXEC_NPU_CMD(aclnnInplaceGeTensor, self, other);
return self;
}
}
}