变更摘要

该 PR 按 resources/design_template.md 模板提交 aclblasCherk（单精度复数厄米特秩-k 更新，BLAS-3）在 Atlas A2/A3（arch22 / DAV_2201，Ascend 910B3）上的实现设计文档（04_tasks/01_community-task-2026/tasklist/08-29-aclblasCherk-A2A3/m0_56676577/docs/design.md，新增 193 行）。文档规划了三段式流水线（AIV deinterleave → AIC GEMM → AIV combine）结合 4M 分解的算子实现方案、GEMM 双路径选择、交错 4M 分解的虚部精度优化、uplo 三角 tile 裁剪与多核 round-robin 调度，并给出精度与性能实测数据，确认满足任务硬指标。

主要改动

• 三段式流水线与 4M 分解：设计 AIV deinterleave → AIC GEMM → AIV combine 的算子流水结构，并结合 4M 分解组织整体计算流程。
• GEMM 双路径：提供 KFC 高吞吐路径（47–60 TFLOPS）与 MatmulImpl gemm4 通用回退路径，兼顾峰值性能与通用性。
• 虚部精度方案：采用交错 4M 分解方案，将 k=2048 时虚部绝对误差由 1.0e-3 降至 2.2e-5，matched_ratio 达到 0.9998+。
• 三角 tile 裁剪与调度：利用 Hermitian 对称性对 uplo 三角 tile 进行裁剪（计算量减半），并采用多核 round-robin 调度。
• 精度与性能验证：记录 ST 1220/1220、UT 37/37、白盒 1044/1044、PyTorch ST 76/76，以及 Atlas 800T A2 / CANN 9.1.0 下三个 case 的性能实测，均满足硬指标；合入目标为 blas/herk/ 与 test/herk/cherk/arch22/。

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

💡 Tip:

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

CLA Signature Guide

@m0_56676577 , thanks for your pull request.

The following commit(s) have not signed Contributor License Agreement (CLA).

If you need to sign CLA, you can click here to sign the CLA.

If you need to check if the email is set up correctly, you can click here to do it in the FAQs.

After signing the CLA or updating the email, you must comment /check-cla to check the CLA status again.

/compile

评审通过
@m0_56676577

/approve

/approve