Pull Request已成功合入, 合并人@ascend-robot
(感谢 zhucehw 的贡献)变更摘要
本 PR 主要对 torch_npu 的 Inductor 相关代码进行重构优化:改进了 aot_compile/aot_load 在 NPU 设备上的加载与绑定逻辑,调整了 aot_load_npu 的设备判断流程,并将 AOTI runner 的 Python 绑定初始化接口从接收 PyObject* module 参数改为无参、内部自行导入 torch._C 模块。PR 描述还提到为 aot_compile 和 aot_load 补充文档(相关文档文件无内嵌补丁)。
主要改动
aot_load_npu加载逻辑重构(torch_npu/_inductor/export.py):由原来的先调用原生origin_aot_load、捕获RuntimeError后再回退的 try/except 方式,改为根据device参数预先判断——当设备为"npu"或以"npu:"开头时直接使用torch._C._aoti.AOTIModelContainerRunnerNpu(so_path, 1, device)创建 runner,否则回退到origin_aot_load,并顺带删除了函数 docstring。initAOTIRunnerBindingsNpu接口变更(pybind.cpp/pybind.h):函数签名去掉PyObject* module参数,改为无参并在函数体内通过py::module::import("torch._C")导入模块;同时将挂载方式由def_submodule("_aoti")改为取用已存在的rootModule.attr("_aoti")子模块。- 调用点同步更新(
torch_npu/csrc/npu/Module.cpp):RegisterNPUDeviceProperties中对torch::inductor::initAOTIRunnerBindingsNpu的调用改为无参形式,以匹配新的函数签名;相关头文件补上了文件末尾缺失的换行符。


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.
PR Approval Progress
✅ Congratulations! All modules have met the lgtm and approve requirements.
Module Approval Details
| module | lgtm status | approve status |
|---|---|---|
| **/*.md | ✅ li_jing_hw, liangsongwei (2/2) | ✅ li_jing_hw (1/1) |
| repo-Ascend/pytorch | ✅ liangsongwei, zichun_ye, li_jing_hw (3/2) | ✅ zichun_ye, li_jing_hw (2/1) |
| torch_npu/_inductor | ✅ weizhan4, zichun_ye, li_jing_hw, HinPeng, liangsongwei (5/2) | ✅ HinPeng, weizhan4, zichun_ye, li_jing_hw (4/1) |
| torch_npu/csrc/inductor | ✅ HinPeng, liangsongwei, weizhan4, zichun_ye, li_jing_hw (5/2) | ✅ weizhan4, zichun_ye, li_jing_hw, HinPeng (4/1) |
💡 Tip:
- Committer can comment
/approveor/lgtm- Commenting
/approveimplies both code review (lgtm) and intent to merge (approve)
CLA Signature Pass
zhucehw, thanks for your pull request. All authors of the commits have signed the CLA. 👍


当前仓库存在以下 保护分支 :
| Protected Branch | Version | Release |
|---|---|---|
| master | ||
| v2.11.0 | ||
| v2.7.1 | ||
| v2.9.0 | ||
| v2.12.0 | ||
| v2.10.0 | ||
| v2.12.0-26.1.0 | ||
| v2.11.0-26.1.0 | ||
| v2.10.0-26.1.0 | ||
| v2.9.0-26.1.0 | ||
| v2.7.1-26.1.0 | ||
| ci-test |
评论 /sync <branch1> <branch2> ... 可将当前 PR 修改同步到其它分支(创建同步 PR):
a) 如果当前 PR 是 Open 状态,同步操作将延迟到 PR 被合并时执行
b) 如果当前 PR 已经 Merged,将立即执行同步操作
注意:
- /sync 命令可以指定同步到多个分支,仅最后一个 /sync 命令生效
- 如果创建的同步 PR 不正确,可通过向同步 PR 的源分支提交轻量级 PR 完善,或使用 /close 命令关闭


ascend docs pipeline is running...


| 阶段 | 任务名 | 状态 | 详情 |
|---|---|---|---|
| 编译构建 | Build_X86 | ✅ | >>> |
| Build_ARM | ✅ | >>> | |
| Build_X86_torchair | 🛑 | >>> | |
| Build_ARM_torchair | 🛑 | >>> | |
| patch_test | 🛑 | >>> | |
| Build_X86_213 | ✅ | >>> | |
| Build_ARM_213 | ✅ | >>> | |
| 恶意代码检查 | Antipoison | ✅ | >>> |
| 编码安全与规范检查 | codecheck_pre-commit | ✅ | >>> |
| check_error | ✅ | >>> | |
| lintrunner | ✅ | >>> | |
| 开源片段检查 | SCA | ✅ | >>> |
| 开发者测试 | UT_ARM_A3_Part_01 | 🛑 | >>> |
| UT_ARM_A3_Part_02 | 🛑 | >>> | |
| UT_ARM_A2_Part_01 | ✅ | >>> | |
| UT_ARM_A2_Part_02 | ✅ | >>> | |
| UT_ARM_A2_Part_03 | ✅ | >>> | |
| UT_inductor_Part_01 | 🛑 | >>> | |
| UT_inductor_Part_02 | 🛑 | >>> | |
| UT_inductor_Part_03 | 🛑 | >>> | |
| UT_inductor_Part_04 | 🛑 | >>> | |
| UT_DIST_ARM_Part_01 | 🛑 | >>> | |
| UT_DIST_ARM_Part_02 | 🛑 | >>> | |
| UT_DIST_ARM_Part_03 | 🛑 | >>> | |
| UT_DIST_ARM_Part_04 | 🛑 | >>> | |
| UT_ARM_A2_Select_Part_01 | ✅ | >>> | |
| UT_ARM_A2_Select_Part_02 | ✅ | >>> | |
| UT_ARM_A2_Part_01_213 | ✅ | >>> | |
| UT_ARM_A2_Part_02_213 | ✅ | >>> | |
| UT_ARM_A2_Part_03_213 | ✅ | >>> | |
| UT_inductor_Part_01_213 | 🛑 | >>> | |
| UT_inductor_Part_02_213 | 🛑 | >>> | |
| UT_inductor_Part_03_213 | 🛑 | >>> | |
| UT_inductor_Part_04_213 | 🛑 | >>> | |
| UT_DIST_ARM_Part_01_213 | 🛑 | >>> | |
| UT_DIST_ARM_Part_02_213 | 🛑 | >>> | |
| UT_DIST_ARM_Part_03_213 | 🛑 | >>> | |
| UT_DIST_ARM_Part_04_213 | 🛑 | >>> | |
| UT_ARM_A2_Select_Part_01_213 | ✅ | >>> | |
| UT_ARM_A2_Select_Part_02_213 | ✅ | >>> | |
| 流水线 | PR-pipeline_pytorch | ✅ | >>> |
- compile、compile_inductor、compile_torchair : 运行流水线
- retry : 重试流水线所有失败子任务
- retry <任务名> : 仅重试指定失败子任务
- stop : 停止流水线


/approve


/approve


/lgtm




【合入来源】
https://gitcode.com/Ascend/pytorch/issues/4097?ref=&did=4257078#tid-4257078
【修改方案】
【资料变更】
不涉及
【接口变更】
不涉及
【功能验证】
CI PASS
【CheckList】