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

PR Approval Progress

✅ Congratulations! All modules have met the lgtm and approve requirements.

Module Approval Details

💡 Tip:

• Committer can comment /approve or /lgtm
• Commenting /approve implies both code review (lgtm) and intent to merge (approve)

CLA Signature Pass

Crrryyyy, thanks for your pull request. All authors of the commits have signed the CLA. 👍

变更摘要

此 PR 主要完成三项工作：将 AscendC Tensor API 的 Mmad 和 Copy 调用方式从"参数尾随"改为通过 .with() 链式传递参数的新 API 风格；全局将 GetTaskRation 重命名为 GetTaskRatio；将第三方依赖子模块从 third_party/tensor_api 迁移到 third_party/ops-tensor。此外，为 CopyL12L0MxScaleA3510 / CopyL12L0MxScaleB3510 这类 scale copy 原子添加了支持运行时坐标参数的 Copy 重载。

主要改动

• Mmad / Copy API 调用风格变更：所有示例和 recipe 中的 AscendC::Te::Mmad(MadOp, dst, srcA, srcB, para) 改为 AscendC::Te::Mmad(MadOp.with(para), dst, srcA, srcB)；AscendC::Te::Copy(copyOp, dst, src, fixpipeParams) 改为 AscendC::Te::Copy(copyOp.with(fixpipeParams), dst, src)，参数通过 .with() 绑定到操作对象上。

• GetTaskRation → GetTaskRatio 全局重命名：在调度器、epilogue、kernel 和 streamk 相关文件中，AscendC::GetTaskRation() 统一更名为 AscendC::GetTaskRatio()，涉及 block scheduler、epilogue、kernel 入口等多处调用。

• third_party/tensor_api → third_party/ops-tensor 子模块迁移：CMake 配置 tensor_api.cmake 中所有路径和 submodule 初始化命令从 third_party/tensor_api 更新为 third_party/ops-tensor，子模块 commit 也相应切换。

• 新增 scale copy 原子的 Copy 重载：在多个 copy_scale_l1_to_l0a.h / copy_scale_l1_to_l0b.h 中，于 namespace AscendC::Te 内为 CopyL12L0MxScaleA3510 和 CopyL12L0MxScaleB3510 添加接受 Coord 参数的三参数 Copy 模板重载，使调用处可直接传入 AscendC::Te::MakeCoord(...) 而无需旧式包装。

• MmadParams 命名空间修正与 include 路径调整：部分文件将 AscendC::MmadParams 修正为 AscendC::Te::MmadParams；kernel_universal.h 中 "blaze/utils/common_utils.h" 改为 "blaze/gemm/utils/common_utils.h"；copy_scale_l1_to_l0b.h（grouped_matmul_recipes）移除了不再使用的 LoadDataTrait / LoadData2BTrait 定义。

代码审查

审查总结

共审查 67 个变更文件，发现 3 个问题：

整体风险评估：低-中等。此 PR 主要完成三项工作：

1. 子模块重命名：third_party/tensor_api → third_party/ops-tensor（.gitmodules、README.md、cmake/tensor_api.cmake 均已同步更新，CMake target 名 cann_samples::tensor_api 保持不变，正确）
2. API 迁移：Mmad/Copy 调用从旧式 Op(dst, src, params) 迁移为新式 Op.with(params)(dst, src)，涉及约 50 个文件，迁移模式一致、参数顺序保留正确
3. 拼写修正：GetTaskRation() → GetTaskRatio()（约 15 处，均正确）

唯一需要关注的是 kernel_universal.h 中的 include 路径变更，建议在合并前验证新路径在更新后的依赖中存在。

各文件审查结果（均逐一审查）：

• .gitmodules — no issues
• README.md — no issues
• Samples/0_Introduction/matmul/main.asc — no issues
• Samples/1_Features/instruction_optimization/mte2_preload/main.asc — no issues
• Samples/1_Features/instruction_optimization/n_buffer/main.asc — no issues
• Samples/1_Features/instruction_optimization/unit_flag/main.asc — no issues
• Samples/1_Features/instruction_optimization/weightnz/main.asc — no issues
• Samples/1_Features/memory_optimization/full_load/main.asc — no issues
• Samples/1_Features/memory_optimization/l1_bank_conflict/main.asc — no issues
• Samples/1_Features/memory_optimization/scale_cache/main.asc — no issues
• Samples/1_Features/memory_optimization/slide_window_adaptive_template/main.asc — no issues
• Samples/1_Features/system_optimization/streamk/main.asc — no issues
• Samples/1_Features/system_optimization/streamk/README.md — P3: 残留 GetTaskRation() 引用
• Samples/1_Features/system_optimization/tail_rebalance/main.asc — no issues
• Samples/2_Performance/CMakeLists.txt — P3: simd_vf_story 被注释
• Samples/2_Performance/full_quant_fused_infer_attention_score_story/include/block/flash_attention_score_block_vec_infer.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../quant_grouped_matmul_hif8_block_mmad.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../quant_grouped_matmul_mx_block_mmad.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../quant_grouped_matmul_mx_block_scheduler_split_m.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../weight_quant_grouped_matmul_mxfp8fp4_block_mmad_split_m.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../quant_grouped_matmul_mx_kernel.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../copy_scale_l1_to_l0a.h — no issues
• Samples/2_Performance/grouped_matmul_story/.../copy_scale_l1_to_l0b.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/matmul_block_mmad_streamk.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/matmul_block_mmad_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_hifp8_block_mmad_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_hifp8_block_scheduler_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_mmad_a_full_load.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_mmad_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_mmad_swat_4_buffer.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_scheduler_a_full_load.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_scheduler_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/quant_matmul_mx_block_scheduler_swat_4_buffer.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/block/weight_quant_matmul_mxfp8fp4_block_mmad_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/epilogue/matmul_block_epilogue_streamk.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/kernel/kernel_universal.h — P2: include 路径变更
• Samples/2_Performance/matmul_story/matmul_recipes/include/kernel/matmul_kernel_streamk.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/kernel/weight_quant_matmul_mxfp8fp4_kernel_swat.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/tile/copy_scale_l1_to_l0a.h — no issues
• Samples/2_Performance/matmul_story/matmul_recipes/include/tile/copy_scale_l1_to_l0b.h — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/0_naive/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/1_pingpong/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/2_block_swat/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/3_last_round_tile_balance/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/4_unit_flag/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/5_halfl1_ping_halfl1_pong/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/6_scale_memory_access_coalescing/... (3 files) — no issues
• Samples/2_Performance/matmul_story/matmul_tutorials/7_fullload/... (3 files) — no issues
• cmake/tensor_api.cmake — no issues
• third_party/ops-tensor — no issues (submodule add)
• third_party/tensor_api — no issues (submodule delete)

⚠️ 已识别出整体风险，但无法提取行内评论，请参考整体评估。