Line data    Source code

       1              : /**
       2              :  * Copyright (c) 2025 Huawei Technologies Co., Ltd.
       3              :  * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
       4              :  * CANN Open Software License Agreement Version 2.0 (the "License").
       5              :  * Please refer to the License for details. You may not use this file except in compliance with the License.
       6              :  * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
       7              :  * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
       8              :  * See LICENSE in the root of the software repository for the full text of the License.
       9              :  */
      10              : 
      11              : /*!
      12              :  * \file kernel_vec_scatter_check.cpp
      13              :  * \brief
      14              :  */
      15              : 
      16              : #include "kernel_check_params.h"
      17              : #include "kernel_vec_scatter_check.h"
      18              : 
      19              : namespace AscendC {
      20              : namespace check {
      21              : 
      22            0 : bool TikcppVecScatterCheck::CheckTensorSize(std::vector<uint64_t> maskArray)
      23              : {
      24              :     // max element calculated in a repeat
      25            0 :     uint64_t maskVal = (maskArray.size() == 1) ? maskArray[0] : GetMaskLength(maskArray, param_.dstDtypeBytes);
      26              : 
      27              :     // dstLocal:
      28            0 :     uint64_t blockLen = ONE_BLK_SIZE / param_.dstDtypeBytes; // ele num in one block
      29            0 :     uint64_t blkNumLastRep = DivCeil(maskVal, blockLen);     // last repeat needs x blocks for maskLen elements
      30            0 :     uint64_t eleNumLastBlk = ((maskVal % blockLen) != 0) ? (maskVal % blockLen) : blockLen;
      31            0 :     uint64_t maxOffset =
      32            0 :         ((param_.repeatTimes - 1) * DEFAULT_REPEAT_STRIDE + (blkNumLastRep - 1) * DEFAULT_BLK_STRIDE) * blockLen +
      33              :         eleNumLastBlk;
      34            0 :     maxOffset = maxOffset * param_.dstDtypeBytes + param_.dstBaseAddr;
      35            0 :     ASCENDC_CHECK(CheckTensorSizeOverflow(maxOffset, param_.dstSize, "dstLocal", "Scatter"));
      36            0 :     return true;
      37              : }
      38              : 
      39            0 : bool TikcppVecScatterCheck::CommonCheck()
      40              : {
      41            0 :     const std::string supportPos = "VECIN/VECOUT/VECCALC";
      42            0 :     ASCENDC_CHECK(CheckTensorScope(param_.dstLogicPos, static_cast<uint8_t>(HardWareIndex::UB), "dst", supportPos));
      43            0 :     ASCENDC_CHECK(CheckTensorScope(param_.srcLogicPos, static_cast<uint8_t>(HardWareIndex::UB), "src", supportPos));
      44            0 :     ASCENDC_CHECK(
      45              :         CheckTensorScope(param_.dstOffsetLogicPos, static_cast<uint8_t>(HardWareIndex::UB), "dstOffset", supportPos));
      46              : 
      47            0 :     ASCENDC_CHECK(CheckAddrAlign());
      48            0 :     ASCENDC_CHECK(CheckBufferSizeOverFlow(
      49              :         param_.dstSize, GlobalParams::Instance().bufferSizeMap.at(param_.dstPos),
      50              :         "check dst tensor buffersize failed"));
      51            0 :     ASCENDC_CHECK(CheckBufferSizeOverFlow(
      52              :         param_.srcSize, GlobalParams::Instance().bufferSizeMap.at(param_.srcPos),
      53              :         "check src tensor buffersize failed"));
      54            0 :     ASCENDC_CHECK(CheckBufferSizeOverFlow(
      55              :         param_.dstOffsetSize, GlobalParams::Instance().bufferSizeMap.at(param_.dstOffsetPos),
      56              :         "check dstOffset tensor buffersize failed"));
      57            0 :     return true;
      58            0 : }
      59              : 
      60            0 : bool TikcppVecScatterCheck::CheckAllLowLevel(std::vector<uint64_t> maskArray)
      61              : {
      62            0 :     uint32_t maxByteLen = param_.dstDtypeBytes;
      63            0 :     ASCENDC_CHECK(UpdateMaskArrayAndCheck(maskArray, maxByteLen));
      64            0 :     ASCENDC_CHECK(CommonCheck());
      65            0 :     ASCENDC_CHECK(CheckTensorSize(maskArray));
      66              :     TensorOverflowParams params = {
      67            0 :         param_.srcSize,
      68            0 :         param_.srcDtypeBytes,
      69            0 :         static_cast<uint64_t>(param_.repeatTimes),
      70              :         static_cast<uint64_t>(DEFAULT_BLK_STRIDE),
      71            0 :         static_cast<uint64_t>(param_.srcRepStride),
      72            0 :         false};
      73            0 :     ASCENDC_CHECK(CheckTensorOverflowLow(maskArray, params, "srcLocal"));
      74            0 :     return true;
      75              : }
      76            0 : bool TikcppVecScatterCheck::CheckAddrAlign()
      77              : {
      78            0 :     bool dstRes = CheckTensorAddrAlign(param_.dstAddr, param_.dstPos, ONE_BLK_SIZE, "dst");
      79            0 :     bool srcRes = CheckTensorAddrAlign(param_.srcAddr, param_.srcPos, ONE_BLK_SIZE, "src");
      80            0 :     bool dstOffsetRes = CheckTensorAddrAlign(param_.dstOffsetAddr, param_.dstOffsetPos, ONE_BLK_SIZE, "dstOffset");
      81            0 :     return dstRes && srcRes && dstOffsetRes;
      82              : }
      83              : 
      84            0 : bool TikcppVecScatterCheck::CheckAllHighLevel()
      85              : {
      86            0 :     ASCENDC_CHECK(CommonCheck());
      87            0 :     ASCENDC_CHECK(CheckTensorOverflowHigh(param_.dstDtypeBytes, param_.dstSize, param_.count, "dstLocal"));
      88            0 :     ASCENDC_CHECK(CheckTensorOverflowHigh(param_.srcDtypeBytes, param_.srcSize, param_.count, "srcLocal"));
      89            0 :     ASCENDC_CHECK(
      90              :         CheckTensorOverflowHigh(param_.dstOffsetDtypeBytes, param_.dstOffsetSize, param_.count, "dstOffsetLocal"));
      91            0 :     return true;
      92              : }
      93              : } // namespace check
      94              : } // namespace AscendC