* 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 logical_and.h
* \brief
*/
#ifndef __LOGICAL_AND_H__
#define __LOGICAL_AND_H__
#include "kernel_operator.h"
#include "kernel_tiling/kernel_tiling.h"
#include "logical_and_tiling_data.h"
#include "logical_and_tiling_key.h"
namespace MyLogicalAnd {
using namespace AscendC;
template <uint64_t BUFFER_NUM>
class LogicalAnd {
public:
__aicore__ inline LogicalAnd(){};
__aicore__ inline void Init(GM_ADDR x1, GM_ADDR x2, GM_ADDR y, const LogicalAndTilingData* tilingData, TPipe* pipeIn);
__aicore__ inline void Process();
private:
__aicore__ inline void TotalStage(int32_t offset);
private:
TPipe* pipe;
TQueBind<TPosition::VECIN, TPosition::VECOUT, BUFFER_NUM> queBind;
TBuf<QuePosition::VECCALC> xCastTmp;
GlobalTensor<int8_t> x1Gm;
GlobalTensor<int8_t> x2Gm;
GlobalTensor<int8_t> yGm;
uint32_t coreDataNum;
uint32_t tileNum;
uint32_t tileDataNum;
uint32_t tailDataNum;
uint32_t processDataNum;
};
template <uint64_t BUFFER_NUM>
__aicore__ inline void LogicalAnd<BUFFER_NUM>::Init(GM_ADDR x1, GM_ADDR x2, GM_ADDR y, const LogicalAndTilingData* tilingData, TPipe* pipeIn)
{
ASSERT(GetBlockNum() != 0 && "block dim can not be zero!");
uint32_t coreNum = GetBlockIdx();
uint32_t globalBufferIndex = tilingData->bigCoreDataNum * GetBlockIdx();
this->tileDataNum = tilingData->tileDataNum;
this->pipe = pipeIn;
if (coreNum < tilingData->tailBlockNum)
{
this->coreDataNum = tilingData->bigCoreDataNum;
this->tileNum = tilingData->finalBigTileNum;
this->tailDataNum = tilingData->bigTailDataNum;
}
else
{
this->coreDataNum = tilingData->smallCoreDataNum;
this->tileNum = tilingData->finalSmallTileNum;
this->tailDataNum = tilingData->smallTailDataNum;
globalBufferIndex -= (tilingData->bigCoreDataNum - tilingData->smallCoreDataNum) * (GetBlockIdx() - tilingData->tailBlockNum);
}
x1Gm.SetGlobalBuffer((__gm__ int8_t *)x1 + globalBufferIndex, this->coreDataNum);
x2Gm.SetGlobalBuffer((__gm__ int8_t *)x2 + globalBufferIndex, this->coreDataNum);
yGm.SetGlobalBuffer((__gm__ int8_t *)y + globalBufferIndex, this->coreDataNum);
pipe->InitBuffer(queBind, BUFFER_NUM, this->tileDataNum * sizeof(int8_t) * 2);
pipe->InitBuffer(xCastTmp, this->tileDataNum * sizeof(half));
}
template <uint64_t BUFFER_NUM>
__aicore__ inline void LogicalAnd<BUFFER_NUM>::TotalStage(int32_t offset)
{
LocalTensor<int8_t> x1Local = queBind.template AllocTensor<int8_t>();
LocalTensor<int8_t> x2Local = x1Local[this->processDataNum];
int32_t eventIDMTE2ToV = static_cast<int32_t>(GetTPipePtr()->FetchEventID(HardEvent::MTE2_V));
int32_t eventIDVToMTE3 = static_cast<int32_t>(GetTPipePtr()->FetchEventID(HardEvent::V_MTE3));
LocalTensor<half> x1LocalHalf = xCastTmp.Get<half>();
LocalTensor<half> x2LocalHalf = x1Local.template ReinterpretCast<half>();
DataCopy(x1Local, x1Gm[offset], this->processDataNum);
SetFlag<HardEvent::MTE2_V>(eventIDMTE2ToV);
WaitFlag<HardEvent::MTE2_V>(eventIDMTE2ToV);
Cast(x1LocalHalf, x1Local, RoundMode::CAST_NONE, this->processDataNum);
DataCopy(x2Local, x2Gm[offset], this->processDataNum);
SetFlag<HardEvent::MTE2_V>(eventIDMTE2ToV);
WaitFlag<HardEvent::MTE2_V>(eventIDMTE2ToV);
Cast(x2LocalHalf, x2Local, RoundMode::CAST_NONE, this->processDataNum);
Mul(x1LocalHalf, x1LocalHalf, x2LocalHalf, this->processDataNum);
Cast(x1Local, x1LocalHalf, RoundMode::CAST_RINT, this->processDataNum);
SetFlag<HardEvent::V_MTE3>(eventIDVToMTE3);
WaitFlag<HardEvent::V_MTE3>(eventIDVToMTE3);
DataCopy(yGm[offset], x1Local, this->processDataNum);
queBind.template FreeTensor(x1Local);
}
template <uint64_t BUFFER_NUM>
__aicore__ inline void LogicalAnd<BUFFER_NUM>::Process()
{
int32_t loopCount = this->tileNum - 1;
this->processDataNum = this->tileDataNum;
int32_t offset = 0;
for (int32_t i = 0; i < loopCount; i++, offset+=this->tileDataNum)
{
TotalStage(offset);
}
this->processDataNum = this->tailDataNum;
TotalStage(offset);
}
}
#endif