* 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.
*/
* \file aclnn_div_v3.cpp
* \brief DivV3 L2 API (aclnn) implementation with broadcast support
*/
#include "aclnn_div_v3.h"
#include "div_v3.h"
#include "aclnn_kernels/contiguous.h"
#include "aclnn_kernels/common/op_error_check.h"
#include "opdev/op_log.h"
#include "opdev/op_dfx.h"
#include "opdev/common_types.h"
#include "opdev/data_type_utils.h"
#include "opdev/make_op_executor.h"
#include "opdev/platform.h"
#include "opdev/shape_utils.h"
#include "op_api/aclnn_check.h"
using namespace op;
#ifdef __cplusplus
extern "C" {
#endif
#define ACLNN_MAX_SHAPE_RANK 8
static const std::initializer_list<op::DataType> DTYPE_SUPPORT_LIST = {
DataType::DT_FLOAT, DataType::DT_FLOAT16,
DataType::DT_BF16, DataType::DT_INT32, DataType::DT_INT16};
static bool CheckNotNull(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
OP_CHECK_NULL(self, return false);
OP_CHECK_NULL(other, return false);
OP_CHECK_NULL(out, return false);
return true;
}
static bool CheckDtypeValid(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
if (!CheckType(self->GetDataType(), DTYPE_SUPPORT_LIST)) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID,
"self dtype %s not in support list.", op::ToString(self->GetDataType()).GetString());
return false;
}
if (self->GetDataType() != other->GetDataType()) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID,
"self and other must have same dtype.");
return false;
}
if (self->GetDataType() != out->GetDataType()) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID,
"input and output must have same dtype.");
return false;
}
return true;
}
static bool CheckModeValid(int64_t mode)
{
if (mode < 0 || mode > 2) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID,
"mode must be 0 (RealDiv), 1 (TruncDiv), or 2 (FloorDiv), but got %ld.", mode);
return false;
}
return true;
}
static bool CheckBroadcastShape(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
OP_CHECK_MAX_DIM(self, ACLNN_MAX_SHAPE_RANK, return false);
OP_CHECK_MAX_DIM(other, ACLNN_MAX_SHAPE_RANK, return false);
OP_CHECK_MAX_DIM(out, ACLNN_MAX_SHAPE_RANK, return false);
OP_CHECK_BROADCAST(self, other, return false);
Shape broadcastShape;
BroadcastInferShape(self->GetViewShape(), other->GetViewShape(), broadcastShape);
if (broadcastShape != out->GetViewShape()) {
OP_LOGE(ACLNN_ERR_PARAM_INVALID,
"broadcast shape %s != out shape %s.",
op::ToString(broadcastShape).GetString(),
op::ToString(out->GetViewShape()).GetString());
return false;
}
return true;
}
static aclnnStatus CheckParams(const aclTensor* self, const aclTensor* other,
int64_t mode, const aclTensor* out)
{
CHECK_RET(CheckNotNull(self, other, out), ACLNN_ERR_PARAM_NULLPTR);
CHECK_RET(CheckDtypeValid(self, other, out), ACLNN_ERR_PARAM_INVALID);
CHECK_RET(CheckModeValid(mode), ACLNN_ERR_PARAM_INVALID);
CHECK_RET(CheckBroadcastShape(self, other, out), ACLNN_ERR_PARAM_INVALID);
return ACLNN_SUCCESS;
}
static aclIntArray* GetShapeAsIntArray(const aclTensor* tensor, aclOpExecutor* executor)
{
int64_t dimNum = static_cast<int64_t>(tensor->GetViewShape().GetDimNum());
if (dimNum == 0) {
int64_t shape[1] = {1};
return executor->AllocIntArray(shape, 1);
}
std::vector<int64_t> shape(dimNum);
for (int64_t i = 0; i < dimNum; i++) {
shape[i] = tensor->GetViewShape()[i];
}
return executor->AllocIntArray(shape.data(), dimNum);
}
aclnnStatus aclnnDivV3GetWorkspaceSize(
const aclTensor* self, const aclTensor* other, int64_t mode,
aclTensor* out, uint64_t* workspaceSize, aclOpExecutor** executor)
{
L2_DFX_PHASE_1(aclnnDivV3, DFX_IN(self, other), DFX_OUT(out));
auto uniqueExecutor = CREATE_EXECUTOR();
CHECK_RET(uniqueExecutor.get() != nullptr, ACLNN_ERR_INNER_CREATE_EXECUTOR);
auto ret = CheckParams(self, other, mode, out);
CHECK_RET(ret == ACLNN_SUCCESS, ret);
if (self->IsEmpty() || other->IsEmpty()) {
*workspaceSize = 0;
uniqueExecutor.ReleaseTo(executor);
return ACLNN_SUCCESS;
}
auto selfContiguous = l0op::Contiguous(self, uniqueExecutor.get());
CHECK_RET(selfContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
auto otherContiguous = l0op::Contiguous(other, uniqueExecutor.get());
CHECK_RET(otherContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
auto outShape = GetShapeAsIntArray(out, uniqueExecutor.get());
CHECK_RET(outShape != nullptr, ACLNN_ERR_INNER_NULLPTR);
if (selfContiguous->GetViewShape() != out->GetViewShape()) {
selfContiguous = l0op::BroadcastTo(selfContiguous, outShape, uniqueExecutor.get());
CHECK_RET(selfContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
}
if (otherContiguous->GetViewShape() != out->GetViewShape()) {
otherContiguous = l0op::BroadcastTo(otherContiguous, outShape, uniqueExecutor.get());
CHECK_RET(otherContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
}
auto opResult = l0op::DivV3(selfContiguous, otherContiguous, mode, uniqueExecutor.get());
CHECK_RET(opResult != nullptr, ACLNN_ERR_INNER_NULLPTR);
auto viewCopyResult = l0op::ViewCopy(opResult, out, uniqueExecutor.get());
CHECK_RET(viewCopyResult != nullptr, ACLNN_ERR_INNER_NULLPTR);
*workspaceSize = uniqueExecutor->GetWorkspaceSize();
uniqueExecutor.ReleaseTo(executor);
return ACLNN_SUCCESS;
}
aclnnStatus aclnnDivV3(
void* workspace, uint64_t workspaceSize, aclOpExecutor* executor, aclrtStream stream)
{
L2_DFX_PHASE_2(aclnnDivV3);
return CommonOpExecutorRun(workspace, workspaceSize, executor, stream);
}
#ifdef __cplusplus
}
#endif