Pull Request has been successfully merged by CANN-robot.
(Thank you for rk's contribution.)变更摘要
本 PR 针对 norm 目录下的算子实施编码规范整改,核心目标是把硬编码的魔法数字替换为具名常量、为 lambda 补充显式捕获列表,并用更安全的类型转换替代 reinterpret_cast。改动覆盖 add_rms_norm_hi_f4_quant、rms_norm_hi_f4_quant、batch_norm_grad_v3、rms_norm_dynamic_mx_quant 等算子的 tiling 与 kernel 代码,整体不改变算子对外接口与业务行为,属于可读性与规范一致性维护。
主要改动
- HiF4 量化公共常量集中定义: 在
add_rms_norm_hi_f4_quant_common.h与rms_norm_hi_f4_quant_common.h中新增HIF4_MAX_PER4_ELEMENTS_PER_GROUP、HIF4_MAX_PER8_ELEMENTS_PER_GROUP、HIF4_TMP_SCALE_ELEMENTS_PER_GROUP、HIF4_DEINTERLEAVE_REGISTERS_PER_LOAD以及VREG_SIZE_BYTES,供多个 kernel 文件共享。 - kernel 缓冲区与寄存器步长改用常量:
hif4_r_full_load.h和hif4_split_d.h中的InitBuffer分组元素数由16/8/32改为具名常量,sysRegLen_由硬编码256改为VREG_SIZE_BYTES,gmCopyInRegStep_中的乘数2改为HIF4_DEINTERLEAVE_REGISTERS_PER_LOAD。 - tiling 校验与平台信息获取规范化: 两个 HiF4 tiling base 文件在
CheckOutputShapeValue中引入SCALE_BLOCK_DIM_OFFSET_FROM_END与scaleBlockDimIndex统一索引计算,GetPlatformInfo将reinterpret_cast改为static_cast,并清理多余空行。 batch_norm_grad_v3与rms_norm_dynamic_mx_quant的 lambda 捕获与常量替换:batch_norm_grad_v3_tiling_infer.cpp和rms_norm_dynamic_mx_quant_full_load_tiling.cpp的ubUsed、getUbSize由隐式捕获改为显式捕获依赖变量;batch_norm_grad_v3相关 host/kernel 文件将分组增长因子与折半基数等字面量替换为PACKED_GROUP_GROWTH_FACTOR、FINAL_REDUCTION_ROWS、TWO。
(注:本次仅做只读分析,未执行任何仓库操作。)
🧩 本次 Review 已使用自定义 Skill:




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 |
|---|---|---|
| norm | ✅ 唐超, 钱泽洪 (2/2) | ✅ 钱泽洪 (1/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
wangpengbo26, thanks for your pull request. All authors of the commits have signed the CLA. 👍


compile


| 🚀 CI 流水线已启动 |
|---|
| 📋 执行详情: 点击查看流水线 |


您好,pr合入已准备就绪,请尽快联系committer进行检视,谢谢!
@renruhai @east_yang @fanqirui @yourealize @wang-xing001 @wangrui_ @su-yueming @pingchuantang


/lgtm
/approve


/lgtm
/approve


The following users do not have permission to comment /lgtm or /approve on any module in this PR:
/approve: 唐超


描述
对 norm 系算子(
add_rms_norm_hi_f4_quant、rms_norm_hi_f4_quant、batch_norm_grad_v3、rms_norm_dynamic_mx_quant)的编码规范问题进行整改,无功能逻辑变更。改动原因
编码规范检查发现以下问题:
2、16、8、32及硬编码sysRegLen_ = 256等);[&]隐式捕获,不符合显式捕获要求;GetCompileInfo()返回值转换使用reinterpret_cast,应使用更安全的static_cast。改动方法
add_rms_norm_hi_f4_quant/rms_norm_hi_f4_quant(arch92,两算子对称整改):新增具名常量HIF4_MAX_PER4_ELEMENTS_PER_GROUP(16)、HIF4_MAX_PER8_ELEMENTS_PER_GROUP(8)、HIF4_TMP_SCALE_ELEMENTS_PER_GROUP(32)、HIF4_DEINTERLEAVE_REGISTERS_PER_LOAD(2) 替代InitBuffer中的裸数字;sysRegLen_ = 256硬编码改为VECTOR_LENGTH(由platform::GetVRegSize()派生);scaleRank - 2改为scaleRank - NUM_TWO。batch_norm_grad_v3:k / 2、copies *= 2、rows /= 2等裸2改用TWO常量;rowsPerGroup_增长因子、终止行数改用具名常量(PACKED_GROUP_GROWTH_FACTOR、FINAL_REDUCTION_ROWS)。batch_norm_grad_v3_tiling_infer.cpp的ubUsed与rms_norm_dynamic_mx_quant_full_load_tiling.cpp的getUbSize由[&]隐式捕获改为显式捕获列表。GetCompileInfo()的reinterpret_cast改为static_cast。涉及文件(14 个,+60 -45):
norm/add_rms_norm_hi_f4_quant/op_host/op_tiling/arch92/add_rms_norm_hi_f4_quant_tiling_base.cpp(+5 -5)norm/add_rms_norm_hi_f4_quant/op_kernel/arch92/add_rms_norm_hi_f4_quant_common.h(+7 -1)norm/add_rms_norm_hi_f4_quant/op_kernel/arch92/add_rms_norm_hi_f4_quant_hif4_r_full_load.h(+6 -6)norm/add_rms_norm_hi_f4_quant/op_kernel/arch92/add_rms_norm_hi_f4_quant_hif4_split_d.h(+6 -6)norm/rms_norm_hi_f4_quant/对称 4 个文件(同模式整改)norm/batch_norm_grad_v3/op_host/batch_norm_grad_v3_tiling_infer.cpp(+1 -1)norm/batch_norm_grad_v3/op_host/batch_norm_grad_v3_tiling_ra_split_r.cpp(+3 -2)norm/batch_norm_grad_v3/op_kernel/arch35/3 个 kernel 文件norm/rms_norm_dynamic_mx_quant/op_host/arch35/rms_norm_dynamic_mx_quant_full_load_tiling.cpp(+2 -1)关联的Issue
-https://gitcode.com/cann/ops-nn/issues/7767
测试
编码规范整改,不涉及功能逻辑变更,无需测试。
文档更新
不涉及
类型标签
AI/Agent生成声明