Pull Request已成功合入, 合并人@CANN-robot
（感谢 liyuanqiang 的贡献）
变更摘要
此 PR 修复了 quant_batch_matmul_v3 算子在 transA 且权重为 FRACTAL_NZ 格式时的精度问题。核心改动集中在三个方面：在 Tiling 阶段引入 CapBaseNForATransWeightNz 将 baseN 限制在 32 以内，以匹配 Cube 指令的 N 维度获取步长；在多个 Kernel 实现中修正了 NZ 格式下 B 矩阵的偏移量计算（移除不必要的 Align(nOffset, K0_INT8) 对齐），并新增 SetMatmulOrgShape 函数为不同格式组合正确设置矩阵乘法的原始形状参数。
主要改动
-
新增
CapBaseNForATransWeightNzTiling 约束：在QuantBatchMatmulV3BasicTiling中，当transA && !transB且权重为FORMAT_FRACTAL_NZ时，若baseN > 32则将其截断为 32（ONE_BLK_SIZE），并重新计算usedCoreNum和baseK，防止 Cube 指令因 N 步长过大导致精度异常。 -
修正 B 矩阵 NZ 格式偏移量计算：在
quant_batch_matmul_v3_bf16.h、quant_batch_matmul_v3_bf16_opt.h、quant_batch_matmul_v3_pertoken.h和quant_batch_matmul_v3_pertoken_opt.h中，将非转置 NZ 格式的列偏移从Align(nOffset, K0_INT8) * Align(k_, BMM_BLOCK_NUM)统一修正为nOffset * Align(k_, BMM_BLOCK_NUM)，因为 n1,k1,k0,n0 布局的列步幅应基于 k 对齐而非 n 对齐。 -
新增
SetMatmulOrgShape函数：在quant_batch_matmul_v3_pertoken.h、quant_batch_matmul_v3_pertoken_basic.h和quant_batch_matmul_v3_pertoken_opt.h中，根据特征图/权重是否均为 NZ 格式、仅一侧为 NZ 格式、或均为普通 ND 格式，为mm.SetOrgShape传入正确的对齐后维度参数，确保矩阵乘法指令准确理解数据布局。 -
区分 NZ 格式的
alignedKbSize计算：在quant_batch_matmul_v3_update.h的Init中，为非转置且x2Format为 NZ 格式的情况显式添加分支，保持其alignedKbSize基于BMM_BLOCK_NUM对齐，与普通 ND 格式逻辑一致但代码结构更清晰。
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
| module | lgtm status | approve status |
|---|---|---|
| ** | ✅ gcw_kUomxQ2l, 陈琦, 刘波 (3/2) | ✅ gcw_kUomxQ2l (1/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
liyuanqiang, thanks for your pull request. All authors of the commits have signed the CLA. 👍
compile
CANN-robot流水线任务触发成功
任务链接 [5a84fd02c2e94ec48b2dcb1d83a69ae4][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| Compile_Ascend_X86_mobile_station | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_experimental | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Pre | ✅ SUCCESS | >>>>> | |
| pre_comment | ✅ SUCCESS | >>>>> | |
| Compile_Ascend_X86 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910b | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910c | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_harmony-infer-chs-nn | ✅ SUCCESS | >>>>> | |
| Compile_harmony-infer-chs-nn-1 | ✅ SUCCESS | >>>>> | |
| UT_Test_ophost | ❌ FAILED | ||
| UT_Test_opapi | ✅ SUCCESS | ||
| UT_Test_kernel | ✅ SUCCESS | ||
| UT_Test_opgraph | ✅ SUCCESS | ||
| PreSmoke_A900 | ✅ SUCCESS | >>>>> | |
| PreSmoke_ATK_Test_A2 | ✅ SUCCESS | >>>>> |
流水线任务触发成功
任务链接 [61b36ffffc7c44c1b4f0cd85f141cf5d][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| codecheck | ✅ SUCCESS | >>>>> | |
| antipoison | ✅ SUCCESS | >>>>> | |
| codecheck_checkpr | ✅ SUCCESS | ||
| StaticCheck_codespell_check | ✅ SUCCESS | ||
| StaticCheck_link_validity_check | ✅ SUCCESS | ||
| StaticCheck_resource_existence_check | ✅ SUCCESS | ||
| StaticCheck_tag_closed_check | ✅ SUCCESS | ||
| StaticCheck_markdownlint | ✅ SUCCESS | ||
| codecheck_style | ✅ SUCCESS | >>>>> | |
| codecheck_precommit | ⚠️ WARNING | >>>>> | >>>>> |
| SCA | ✅ SUCCESS | >>>>> |
[2026-07-09 18:43:37] CI执行结束
The following users do not have permission to comment /lgtm or /approve on any module in this PR:
唐超
The following users do not have permission to comment /lgtm or /approve on any module in this PR:
唐超
compile
流水线任务触发成功
任务链接 [f0058f1a9c454a6fbb71f668259caeec][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| Compile_Ascend_X86_mobile_station | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_experimental | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Pre | ✅ SUCCESS | >>>>> | |
| pre_comment | ✅ SUCCESS | >>>>> | |
| Compile_Ascend_X86 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910b | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910c | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_harmony-infer-chs-nn | ✅ SUCCESS | >>>>> | |
| Compile_harmony-infer-chs-nn-1 | ✅ SUCCESS | >>>>> | |
| UT_Test_ophost | ❌ FAILED | ||
| UT_Test_opapi | ✅ SUCCESS | ||
| UT_Test_kernel | ✅ SUCCESS | ||
| UT_Test_opgraph | ✅ SUCCESS | ||
| PreSmoke_A900 | ✅ SUCCESS | >>>>> | |
| PreSmoke_ATK_Test_A2 | ✅ SUCCESS | >>>>> |
流水线任务触发成功
任务链接 [b2ccfc2dcd6047f99202dba21391f3a6][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| codecheck | ✅ SUCCESS | >>>>> | |
| antipoison | ✅ SUCCESS | >>>>> | |
| codecheck_checkpr | ✅ SUCCESS | ||
| StaticCheck_codespell_check | ✅ SUCCESS | ||
| StaticCheck_link_validity_check | ✅ SUCCESS | ||
| StaticCheck_resource_existence_check | ✅ SUCCESS | ||
| StaticCheck_tag_closed_check | ✅ SUCCESS | ||
| StaticCheck_markdownlint | ✅ SUCCESS | ||
| codecheck_style | ✅ SUCCESS | >>>>> | |
| codecheck_precommit | ⚠️ WARNING | >>>>> | >>>>> |
| SCA | ✅ SUCCESS | >>>>> |
[2026-07-09 19:39:23] CI执行结束
compile
流水线任务触发成功
任务链接 [26e7efc1d8d441768a9bd5652df2d9c1][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| Compile_Ascend_X86_mobile_station | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_experimental | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Pre | ✅ SUCCESS | >>>>> | |
| pre_comment | ✅ SUCCESS | >>>>> | |
| Compile_Ascend_X86 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910b | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910c | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_harmony-infer-chs-nn | ✅ SUCCESS | >>>>> | |
| Compile_harmony-infer-chs-nn-1 | ✅ SUCCESS | >>>>> |
[2026-07-09 20:00:08] CI执行结束
流水线任务触发成功
任务链接 [f3b1d1c56b2a474285241bf0ad79954c][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| codecheck | ✅ SUCCESS | >>>>> | |
| antipoison | ✅ SUCCESS | >>>>> | |
| codecheck_checkpr | ✅ SUCCESS | ||
| StaticCheck_codespell_check | ✅ SUCCESS | ||
| StaticCheck_link_validity_check | ✅ SUCCESS | ||
| StaticCheck_resource_existence_check | ✅ SUCCESS | ||
| StaticCheck_tag_closed_check | ✅ SUCCESS | ||
| StaticCheck_markdownlint | ✅ SUCCESS | ||
| codecheck_style | ✅ SUCCESS | >>>>> | |
| codecheck_precommit | ⚠️ WARNING | >>>>> | >>>>> |
| SCA | ✅ SUCCESS | >>>>> |
[2026-07-09 19:55:05] CI执行结束
compile
流水线任务触发成功
任务链接 [d03bc32491b247fa92a5e92546553fe4][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| Compile_Ascend_X86_mobile_station | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_single | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_experimental | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_X86_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Ascend_ARM_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_Pre | ✅ SUCCESS | >>>>> | |
| pre_comment | ✅ SUCCESS | >>>>> | |
| Compile_Ascend_X86 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910b | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_910c | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_X86_monitor_950 | ✅ SUCCESS | >>>>> | >>>>> |
| Compile_harmony-infer-chs-nn | ✅ SUCCESS | >>>>> | |
| Compile_harmony-infer-chs-nn-1 | ✅ SUCCESS | >>>>> | |
| UT_Test_ophost | ✅ SUCCESS | ||
| UT_Test_opapi | ✅ SUCCESS | ||
| UT_Test_kernel | ✅ SUCCESS | ||
| UT_Test_opgraph | ✅ SUCCESS | ||
| PreSmoke_A900 | ✅ SUCCESS | >>>>> | |
| PreSmoke_ATK_Test_A2 | ✅ SUCCESS | >>>>> |
[2026-07-09 20:10:15] CI执行结束
流水线任务触发成功
任务链接 [882b3e721c5b43438c38ce1bd35bd178][流水线指导]
| 任务名称 | 状态 | 日志 | 下载链接 |
|---|---|---|---|
| codecheck | ✅ SUCCESS | >>>>> | |
| antipoison | ✅ SUCCESS | >>>>> | |
| codecheck_checkpr | ✅ SUCCESS | ||
| StaticCheck_codespell_check | ✅ SUCCESS | ||
| StaticCheck_link_validity_check | ✅ SUCCESS | ||
| StaticCheck_resource_existence_check | ✅ SUCCESS | ||
| StaticCheck_tag_closed_check | ✅ SUCCESS | ||
| StaticCheck_markdownlint | ✅ SUCCESS | ||
| codecheck_style | ✅ SUCCESS | >>>>> | |
| codecheck_precommit | ⚠️ WARNING | >>>>> | >>>>> |
| SCA | ✅ SUCCESS | >>>>> |
[2026-07-09 20:05:02] CI执行结束
/approve
描述
关联的Issue
#issue: #3847
测试
文档更新
类型标签
AI/Agent生成声明