Merged
docs: 修复注释与文档不一致（circular_pad、clip_by_value、coalesce_sparse 文件注释） #6597
TangPCAdd on 19 days ago
docs: 修复注释与文档不一致（circular_pad、clip_by_value、coalesce_sparse 文件注释） #6597
Merged
Total: 5 files changed+12-12
| @@ -54,4 +54,4 @@ ACLNN_API aclnnStatus aclnnCircularPad2d(void* workspace, uint64_t workspaceSize | |||
| 54 | } | 54 | } |
| 55 | #endif | 55 | #endif |
| 56 | 56 | ||
| 57 | -#endif // OP_API_INC_CIRCULAR_PAD_H_ | 57 | +#endif // OP_API_INC_CIRCULAR_PAD2D_H_ |
| @@ -55,4 +55,4 @@ ACLNN_API aclnnStatus aclnnCircularPad3d(void* workspace, uint64_t workspaceSize | |||
| 55 | } | 55 | } |
| 56 | #endif | 56 | #endif |
| 57 | 57 | ||
| 58 | -#endif // OP_API_INC_CIRCULAR_PAD_H_ | 58 | +#endif // OP_API_INC_CIRCULAR_PAD3D_H_ |
| @@ -9,7 +9,7 @@ | |||
| 9 | */ | 9 | */ |
| 10 | 10 | ||
| 11 | /*! | 11 | /*! |
| 12 | - * \file test_geir_add_n.cpp | 12 | + * \file test_geir_clip_by_value.cpp |
| 13 | * \brief | 13 | * \brief |
| 14 | */ | 14 | */ |
| 15 | 15 | ||
| @@ -25,9 +25,8 @@ template <typename uIdxType, typename idxType, typename dataType> | |||
| 25 | class KernelCoalesceSparse { | 25 | class KernelCoalesceSparse { |
| 26 | public: | 26 | public: |
| 27 | __aicore__ inline KernelCoalesceSparse() = default; | 27 | __aicore__ inline KernelCoalesceSparse() = default; |
| 28 | - __aicore__ inline void Init( | 28 | + __aicore__ inline void Init(GM_ADDR uniqueIndices, GM_ADDR indices, GM_ADDR values, GM_ADDR newIndices, |
| 29 | - GM_ADDR uniqueIndices, GM_ADDR indices, GM_ADDR values, GM_ADDR newIndices, GM_ADDR newValue, | 29 | + GM_ADDR newValue, const CoalesceSparseTilingData* __restrict tilingData); |
| 30 | - const CoalesceSparseTilingData* __restrict tilingData); | ||
| 31 | __aicore__ inline void Process(); | 30 | __aicore__ inline void Process(); |
| 32 | 31 | ||
| 33 | private: | 32 | private: |
| @@ -182,8 +181,8 @@ __aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::CopyIn | |||
| 182 | } | 181 | } |
| 183 | 182 | ||
| 184 | template <typename uIdxType, typename idxType, typename dataType> | 183 | template <typename uIdxType, typename idxType, typename dataType> |
| 185 | -__aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::ComputeAndCopyOut( | 184 | +__aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::ComputeAndCopyOut(uint64_t repeatTime, |
| 186 | - uint64_t repeatTime, uint64_t taskLen) | 185 | + uint64_t taskLen) |
| 187 | { | 186 | { |
| 188 | // maybe the value_size too big and should move many times ---> moveValueTimes and moveValueTail | 187 | // maybe the value_size too big and should move many times ---> moveValueTimes and moveValueTail |
| 189 | // need moveValueTimes, moveValueLen and moveValueTail tiling | 188 | // need moveValueTimes, moveValueLen and moveValueTail tiling |
| @@ -222,8 +221,9 @@ __aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::Comput | |||
| 222 | } | 221 | } |
| 223 | 222 | ||
| 224 | template <typename uIdxType, typename idxType, typename dataType> | 223 | template <typename uIdxType, typename idxType, typename dataType> |
| 225 | -__aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::valueMove( | 224 | +__aicore__ inline void KernelCoalesceSparse<uIdxType, idxType, dataType>::valueMove(uint64_t valueOffset, |
| 226 | - uint64_t valueOffset, uint64_t ubValueOffset, uint64_t moveValueLen) | 225 | + uint64_t ubValueOffset, |
| 226 | + uint64_t moveValueLen) | ||
| 227 | { | 227 | { |
| 228 | uint64_t valueByte = moveValueLen * sizeof(dataType); | 228 | uint64_t valueByte = moveValueLen * sizeof(dataType); |
| 229 | LocalTensor<dataType> valueLocal = valueQueue.AllocTensor<dataType>(); | 229 | LocalTensor<dataType> valueLocal = valueQueue.AllocTensor<dataType>(); |
| @@ -246,4 +246,4 @@ __aicore__ inline uint64_t KernelCoalesceSparse<uIdxType, idxType, dataType>::Ce | |||
| 246 | return y == 0 ? x : (x + y - 1) / y; | 246 | return y == 0 ? x : (x + y - 1) / y; |
| 247 | } | 247 | } |
| 248 | 248 | ||
| 249 | -#endif // COALESCE_SPARSE | 249 | +#endif // COALESCE_SPARSE_H |
| @@ -172,4 +172,4 @@ __aicore__ inline void KernelCoalesceSparseSimt<uIdxType, idxType, dataType>::Pr | |||
| 172 | } | 172 | } |
| 173 | } | 173 | } |
| 174 | 174 | ||
| 175 | -#endif // COALESCE_SPARSE_SIMT | 175 | +#endif // COALESCE_SPARSE_SIMT_H |