* Copyright (c) 2026 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.
*/
#ifndef ADD_CUSTOM_H
#define ADD_CUSTOM_H
#include <algorithm>
#include "acl/acl.h"
#include "kernel_operator.h"
#include "host_device/shmem_common_types.h"
#include "tiling/platform/platform_ascendc.h"
template <typename T>
ACLSHMEM_GLOBAL_VECTOR void add_custom(
__gm__ void *x, __gm__ void *y, __gm__ void *z, int64_t totalLength, int64_t blockLength, uint32_t tileSize)
{
constexpr static int64_t PIPELINE_DEPTH = 2;
AscendC::TPipe pipe;
AscendC::GlobalTensor<T> xGm, yGm, zGm;
AscendC::TQue<AscendC::TPosition::VECIN, PIPELINE_DEPTH> inQueueX;
AscendC::TQue<AscendC::TPosition::VECIN, PIPELINE_DEPTH> inQueueY;
AscendC::TQue<AscendC::TPosition::VECOUT, PIPELINE_DEPTH> outQueueZ;
pipe.InitBuffer(inQueueX, PIPELINE_DEPTH, tileSize);
pipe.InitBuffer(inQueueY, PIPELINE_DEPTH, tileSize);
pipe.InitBuffer(outQueueZ, PIPELINE_DEPTH, tileSize);
xGm.SetGlobalBuffer((__gm__ T *)x + blockLength * AscendC::GetBlockIdx());
yGm.SetGlobalBuffer((__gm__ T *)y + blockLength * AscendC::GetBlockIdx());
zGm.SetGlobalBuffer((__gm__ T *)z + blockLength * AscendC::GetBlockIdx());
int64_t currentBlockLength = totalLength - AscendC::GetBlockIdx() * blockLength;
if (currentBlockLength > blockLength) {
currentBlockLength = blockLength;
}
int64_t elementNumPerTile = tileSize / sizeof(T);
int64_t tileNum = currentBlockLength / elementNumPerTile;
int64_t tailTileElementNum = currentBlockLength - tileNum * elementNumPerTile;
for (int64_t i = 0; i < tileNum; ++i) {
int64_t offset = i * elementNumPerTile;
AscendC::DataCopyExtParams copyParams;
copyParams.blockCount = 1;
copyParams.blockLen = elementNumPerTile * sizeof(T);
copyParams.srcStride = 0;
copyParams.dstStride = 0;
AscendC::DataCopyPadExtParams<T> padParams{false, 0, 0, 0};
AscendC::LocalTensor<T> xLocal = inQueueX.AllocTensor<T>();
AscendC::LocalTensor<T> yLocal = inQueueY.AllocTensor<T>();
AscendC::DataCopyPad(xLocal, xGm[offset], copyParams, padParams);
AscendC::DataCopyPad(yLocal, yGm[offset], copyParams, padParams);
inQueueX.EnQue(xLocal);
inQueueY.EnQue(yLocal);
xLocal = inQueueX.DeQue<T>();
yLocal = inQueueY.DeQue<T>();
AscendC::LocalTensor<T> zLocal = outQueueZ.AllocTensor<T>();
AscendC::Add(zLocal, xLocal, yLocal, elementNumPerTile);
outQueueZ.EnQue(zLocal);
inQueueX.FreeTensor(xLocal);
inQueueY.FreeTensor(yLocal);
zLocal = outQueueZ.DeQue<T>();
AscendC::DataCopyPad(zGm[offset], zLocal, copyParams);
outQueueZ.FreeTensor(zLocal);
}
if (tailTileElementNum > 0) {
int64_t offset = tileNum * elementNumPerTile;
AscendC::DataCopyExtParams copyParams;
copyParams.blockCount = 1;
copyParams.blockLen = tailTileElementNum * sizeof(T);
copyParams.srcStride = 0;
copyParams.dstStride = 0;
AscendC::DataCopyPadExtParams<T> padParams{false, 0, 0, 0};
AscendC::LocalTensor<T> xLocal = inQueueX.AllocTensor<T>();
AscendC::LocalTensor<T> yLocal = inQueueY.AllocTensor<T>();
AscendC::DataCopyPad(xLocal, xGm[offset], copyParams, padParams);
AscendC::DataCopyPad(yLocal, yGm[offset], copyParams, padParams);
inQueueX.EnQue(xLocal);
inQueueY.EnQue(yLocal);
xLocal = inQueueX.DeQue<T>();
yLocal = inQueueY.DeQue<T>();
AscendC::LocalTensor<T> zLocal = outQueueZ.AllocTensor<T>();
AscendC::Add(zLocal, xLocal, yLocal, tailTileElementNum);
outQueueZ.EnQue(zLocal);
inQueueX.FreeTensor(xLocal);
inQueueY.FreeTensor(yLocal);
zLocal = outQueueZ.DeQue<T>();
AscendC::DataCopyPad(zGm[offset], zLocal, copyParams);
outQueueZ.FreeTensor(zLocal);
}
}
template <class T>
void run_vector_add(int64_t numElements, void *a, void *b, void *c, void *stream)
{
constexpr static int64_t MIN_ELEMS_PER_CORE = 1024;
constexpr static int64_t PIPELINE_DEPTH = 2;
constexpr static int64_t BUFFER_NUM = 3;
auto ascendcPlatform = platform_ascendc::PlatformAscendCManager::GetInstance();
uint64_t ubSize;
ascendcPlatform->GetCoreMemSize(platform_ascendc::CoreMemType::UB, ubSize);
int64_t coreNum = ascendcPlatform->GetCoreNumAiv();
int64_t numBlocks = std::min(coreNum, (numElements + MIN_ELEMS_PER_CORE - 1) / MIN_ELEMS_PER_CORE);
numBlocks = std::max(numBlocks, static_cast<int64_t>(1));
int64_t blockLength = (numElements + numBlocks - 1) / numBlocks;
int64_t tileSize = ubSize / PIPELINE_DEPTH / BUFFER_NUM;
add_custom<T><<<numBlocks, nullptr, stream>>>(a, b, c, numElements, blockLength, tileSize);
}
#endif