Pull Request已成功合入, 合并人@CANN-robot
(感谢 esok11 的贡献)变更摘要
本 PR 主要包含两类改动:一是对 AscendC 算子 kernel 的寄存器级(AscendC::Reg)与 ND DMA 相关 API 进行更名适配,涉及 broadcast_to、pad_v3、pad_v3_grad、pad_v3_grad_replication、truncate_div 等算子的 op_kernel 代码;二是日志清理,包括示例程序中失败路径的日志级别从 INFO 修正为 ERROR、日志措辞与变量命名(如 finded→found)的规范化,以及 transpose tiling 接口的拼写修正。整体不改变业务逻辑,仅做接口适配与日志/命名清理。
主要改动
- AscendC 寄存器级 API 更名:在
pad_v3/pad_v3_grad/pad_v3_grad_replication/truncate_div/broadcast_to等 kernel 中,将Reg::DataCopy按读写语义拆分为Reg::LoadAlign/Reg::StoreAlign,DataCopyUnAlign*更名为LoadUnAlign*/StoreUnAlign*,DataCopyGather→Gather、DataCopyScatter→Scatter、CompareScalar→Compares、Copy→Move、MaskNot/MaskAnd→Not/And、MaskSel/GatherMask→Select/Squeeze,并将非对齐辅助类型UnalignReg拆分为UnalignRegForLoad/UnalignRegForStore。 - ND DMA 类型更名与指针标注调整:
broadcast_tokernel 中MultiCopyConfig→NdDmaConfig、MultiCopyLoopInfo→NdDmaLoopInfo、MultiCopyParams→NdDmaParams;同时各 kernel 中 UB 内存指针标注由__local_mem__统一改为__ubuf__。 - Transpose tiling 接口拼写修正:
TransposeNddmaTiling的TilingForReleatedTranspose更名为TilingForRelatedTranspose,成员isReleatedTranspsoe_更名为isRelatedTranspose_,并同步更新transpose、space_to_depth、depth_to_space、confusion_transpose_d的 tiling 调用点。 - 日志与命名清理:各
test_geir_*.cpp示例中初始化/建图/运行/终结失败路径的日志级别由 INFO 改为 ERROR;日志措辞规范化(如 "is/are"、"expect"→"expected"、multiples's→multiples'、"Format only support ND"→"supports ND"),pad_v3与pad_v3_gradtiling 中防死循环标志变量finded统一改为found,并对math/tile示例代码做了格式化整理。


Thanks for your pull-request.
The full list of commands accepted by me can be found at here.
You can get sig-info at here.
You can self-configure the PR merge rules for this repository. For more details, please refer to Here.
For more, you also can visit HICANN.
PR Approval Progress
✅ Congratulations! All modules have met the lgtm and approve requirements.
Module Approval Details
| module | lgtm status | approve status |
|---|---|---|
| conversion/broadcast_to | ✅ zhanw_coding, 汤平川 (2/2) | ✅ zhanw_coding, 汤平川 (2/1) |
| conversion/transpose | ✅ zhanw_coding, 汤平川 (2/2) | ✅ zhanw_coding, 汤平川 (2/1) |
| math/erfc | ✅ 汤平川, zhanw_coding (2/2) | ✅ zhanw_coding, 汤平川 (2/1) |
| math/tile | ✅ zhanw_coding, 汤平川 (2/2) | ✅ 汤平川, zhanw_coding (2/1) |
| repo-cann/ops-math | ✅ zhanw_coding, 汤平川 (2/2) | ✅ zhanw_coding, 汤平川 (2/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)


流水线任务触发成功
任务链接 [89c1e88a6b3a478e9bc8e77c6f31c3f0][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| pre_comment | ✅ SUCCESS | >>>>> | |
| Compile_Ascend_X86_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_mobile_station_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_experimental_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_experimental_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910b | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910c | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_A5_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_A5_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_mobile_station_9030_ubuntu24 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single | ✅ SUCCESS | >>>>> | >>>>> |
| UT_Test | ✅ SUCCESS | ||
| UT_Test_experimental | ✅ SUCCESS | ||
| UT_Test_kernel | ✅ SUCCESS | ||
| PreSmoke_A900 | ✅ SUCCESS | >>>>> | |
| API_Check | ⚠️ WARNING | >>>>> | |
| PreSmoke_ATK_Test_A2 | ✅ SUCCESS | >>>>> |
[2026-08-27 20:21:06] CI执行结束


流水线任务触发成功
任务链接 [e3834f82cf434d7683dc289872a251e1][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| SCA | ✅ SUCCESS | >>>>> | |
| antipoison | ✅ SUCCESS | >>>>> | |
| codecheck_Pr | ✅ SUCCESS | ||
| codecheck_precommit | ✅ SUCCESS | >>>>> | |
| StaticCheck_md | ✅ SUCCESS |
[2026-08-27 18:51:47] CI执行结束


The following labels are not ready.
lgtm: Please wait for reviewers to review the code.
approved: Please wait for committers to review the code.


/lgtm
/approve


/lgtm
/approve


描述
对 arch35 架构下多个算子进行 AscendC Reg API 迁移、命名拼写修正及日志文案规范化,共 46 个文件,涉及 conversion/ 与 math/ 下 15 个算子。
主要改动:
__local_mem__→__ubuf__AscendC::Reg::LoadAlign(原Reg::DataCopy加载语义),存储用AscendC::Reg::StoreAlign(原Reg::DataCopy存储语义)AscendC::Reg::MaskNot→AscendC::Reg::Not、AscendC::Reg::MaskAnd→AscendC::Reg::Andusing AscendC::Reg::DataCopy;声明TilingForReleatedTranspose→TilingForRelatedTranspose(Releated→Related)isReleatedTranspsoe_→isRelatedTranspose_(Releated→Related, Transpsoe→Transpose)INFO→ERROR(Initialize ge failed、Run graph failed、Finalize ir graph session failed 等)only support→only supports、is: %s and %s→are: %s and %s、expect %f→expected %f、multiples's→multiples'改动原因
配合 AscendC Reg API 演进,将 arch35 算子迁移至新的地址空间修饰符与 LoadAlign/StoreAlign/Not/And API;修正 transpose 相关命名中的拼写错误;统一日志级别与文案规范,提升日志可读性与问题定位效率。
改动方法
__local_mem__→__ubuf__、Reg::DataCopy(加载)→Reg::LoadAlign、Reg::DataCopy(存储)→Reg::StoreAlign、Reg::MaskNot→Reg::Not、Reg::MaskAnd→Reg::And,不改变控制流。TilingForReleatedTranspose→TilingForRelatedTranspose、isReleatedTranspsoe_→isRelatedTranspose_,同步所有调用点。关联的Issue
测试
本 PR 为 API 迁移、命名修正与日志清理,通过已有测试用例(含 arch35 tiling UT 与 examples 二级冒烟)回归验证。
文档更新
本 PR 未包含文档更新。
类型标签