#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;
namespace {
bool is_npu_supported(at::ScalarType dtype)
{
static const bool is_adaptive_max_pool_3d_available = check_aclnn_kernel_available("aclnnAdaptiveMaxPool3d");
if (!is_adaptive_max_pool_3d_available || dtype == at::kDouble) {
return false;
}
return true;
}
}
std::tuple<at::Tensor&, at::Tensor&> adaptive_max_pool3d_out(const at::Tensor& self, at::IntArrayRef output_size, at::Tensor& out, at::Tensor& indices)
{
if (!is_npu_supported(self.scalar_type())) {
TORCH_WARN_ONCE("adaptive_max_pool3d.out is not supported by NPU currently. Now this kernel is running on CPU.");
auto out_cpu = out.cpu();
auto indices_cpu = indices.cpu();
auto cpuout = at::adaptive_max_pool3d_out(out_cpu, indices_cpu, self.cpu(), output_size);
out.copy_(std::get<0>(cpuout));
indices.copy_(std::get<1>(cpuout));
return std::tuple<at::Tensor&, at::Tensor&>(out, indices);
}
auto out_size = op_infer::max_pool3d_output_size(self, output_size);
npu_preparation::check_tensor({self}, out, self.scalar_type(), out_size);
npu_preparation::check_tensor({self}, indices, at::kInt, out_size);
EXEC_NPU_CMD(aclnnAdaptiveMaxPool3d, self, output_size, out, indices);
return std::tuple<at::Tensor&, at::Tensor&>(out, indices);
}
std::tuple<at::Tensor, at::Tensor> adaptive_max_pool3d(const at::Tensor& self, at::IntArrayRef output_size)
{
if (!is_npu_supported(self.scalar_type())) {
TORCH_WARN_ONCE("adaptive_max_pool3d is not supported by NPU currently. Now this kernel is running on CPU.");
auto result = at::adaptive_max_pool3d(self.cpu(), output_size);
auto out_npu = std::get<0>(result).to(self.device());
auto indices_npu = std::get<1>(result).to(self.device());
return std::tuple<at::Tensor, at::Tensor>(out_npu, indices_npu);
}
auto out_size = op_infer::max_pool3d_output_size(self, output_size);
at::Tensor out = npu_preparation::apply_tensor_without_format(out_size, self.options());
at::Tensor indices = npu_preparation::apply_tensor_without_format(out_size, self.options().dtype(at::kInt));
EXEC_NPU_CMD(aclnnAdaptiveMaxPool3d, self, output_size, out, indices);
return std::tuple<at::Tensor, at::Tensor>(out, indices);
}
}