* Copyright (c) 2025 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
#include "reduce_max.h"
#include "opdev/aicpu/aicpu_task.h"
#include "opdev/make_op_executor.h"
#include "opdev/op_def.h"
#include "opdev/op_dfx.h"
#include "opdev/op_executor.h"
#include "opdev/op_log.h"
#include "opdev/shape_utils.h"
#include "opdev/platform.h"
#include "op_api/aclnn_check.h"
#include "aclnn_kernels/common/op_error_check.h"
using namespace op;
namespace l0op {
OP_TYPE_REGISTER(ReduceMax);
static const std::initializer_list<op::DataType> AICORE_DTYPE_SUPPORT_LIST_910 = {
op::DataType::DT_FLOAT16, op::DataType::DT_FLOAT, op::DataType::DT_UINT8, op::DataType::DT_INT32};
static const std::initializer_list<op::DataType> AICORE_DTYPE_SUPPORT_LIST_GE910B = {
op::DataType::DT_FLOAT16, op::DataType::DT_FLOAT, op::DataType::DT_UINT8, op::DataType::DT_INT32,
op::DataType::DT_INT64, op::DataType::DT_BF16, op::DataType::DT_INT8};
static inline bool CheckSocVersionGe910B(void)
{
return (GetCurrentPlatformInfo().GetSocVersion() >= SocVersion::ASCEND910B &&
GetCurrentPlatformInfo().GetSocVersion() <= SocVersion::ASCEND910E) ||
IsRegBase();
}
static bool IsAiCoreSupport(const aclTensor* self)
{
if (CheckSocVersionGe910B()) {
return CheckType(self->GetDataType(), AICORE_DTYPE_SUPPORT_LIST_GE910B);
}
return CheckType(self->GetDataType(), AICORE_DTYPE_SUPPORT_LIST_910);
}
static const aclTensor* ReduceMaxAiCore(
const aclTensor* self, const aclTensor* dimList, bool keepDim, bool noopWithEmptyDims, const aclTensor* maxOut,
aclOpExecutor* executor)
{
L0_DFX(ReduceMaxAiCore, self, dimList, keepDim, noopWithEmptyDims, maxOut);
auto retAicore = ADD_TO_LAUNCHER_LIST_AICORE(
ReduceMax, OP_INPUT(self, dimList), OP_OUTPUT(maxOut), OP_ATTR(keepDim, noopWithEmptyDims));
OP_CHECK_ADD_TO_LAUNCHER_LIST_AICORE(
retAicore != ACLNN_SUCCESS, return nullptr, "ReduceMax ADD_TO_LAUNCHER_LIST_AICORE failed.");
return maxOut;
}
static const aclTensor* ReduceMaxAiCpu(
const aclTensor* self, const aclTensor* dimList, bool keepDim, const aclTensor* maxOut, aclOpExecutor* executor)
{
L0_DFX(ReduceMaxAiCpu, self, dimList, keepDim, maxOut);
static internal::AicpuTaskSpace space("Max", ge::DEPEND_IN_SHAPE, true);
auto ret = ADD_TO_LAUNCHER_LIST_AICPU(
ReduceMax, OP_ATTR_NAMES({"keep_dims", "Tidx"}), OP_INPUT(self, dimList), OP_OUTPUT(maxOut),
OP_ATTR(keepDim, dimList->GetDataType()));
CHECK_RET(ret == ACLNN_SUCCESS, nullptr);
return maxOut;
}
const aclTensor* ReduceMax(
const aclTensor* self, const aclIntArray* dim, bool keepDim, bool noopWithEmptyDims, aclOpExecutor* executor)
{
auto dimList = executor->ConvertToTensor(dim, op::DataType::DT_INT64);
auto maxOut = executor->AllocTensor(self->GetViewShape(), self->GetDataType());
auto ret = INFER_SHAPE(ReduceMax, OP_INPUT(self, dimList), OP_OUTPUT(maxOut), OP_ATTR(keepDim, noopWithEmptyDims));
if (ret != ACLNN_SUCCESS) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID, "ReduceMax infer shape faild.");
return nullptr;
}
if (IsAiCoreSupport(self)) {
return ReduceMaxAiCore(self, dimList, keepDim, noopWithEmptyDims, maxOut, executor);
} else {
return ReduceMaxAiCpu(self, dimList, keepDim, maxOut, executor);
}
}
}