[docs]v2.7.1文档产品名称更新-pytorch-AtomGit

已合并

[docs]v2.7.1文档产品名称更新 #39928

chenzhenxi创建于 7月2日从zxchen152/pytorch： v2.7.1 合入到 Ascend/pytorch： v2.7.1

[docs]v2.7.1文档产品名称更新 #39928

已合并

chenzhenxi创建于 7月2日

从zxchen152/pytorch v2.7.1合入到Ascend/pytorch v2.7.1

共 9 个文件变更+129-101

MREADME.md+47-47

	@@ -1,8 +1,8 @@
1	-# Ascend Extension for PyTorch	1	+# TorchNPU
2		2
3	## Overview	3	## Overview
4		4
5	-This repository develops the Ascend Extension for PyTorch named torch_npu to adapt Ascend NPU to PyTorch so that developers who use the PyTorch can obtain powerful compute capabilities of Ascend AI Processors.	5	+TorchNPU is a PyTorch adapter plugin that brings Ascend NPU support into the PyTorch ecosystem, enabling developers to harness the high-performance computing power of Ascend AI processors.
6		6
7	Ascend is a full-stack AI computing infrastructure for industry applications and services based on Huawei Ascend processors and software. For more information about Ascend, see [Ascend Community](https://www.hiascend.com/en/).	7	Ascend is a full-stack AI computing infrastructure for industry applications and services based on Huawei Ascend processors and software. For more information about Ascend, see [Ascend Community](https://www.hiascend.com/en/).
8		8
	@@ -10,76 +10,75 @@ Ascend is a full-stack AI computing infrastructure for industry applications and
10		10
11	### From Binary	11	### From Binary
12		12
13	-Provide users with wheel package to quickly install torch_npu. Before installing torch_npu, complete the installation of CANN according to [Ascend Auxiliary Software](#ascend-auxiliary-software). To obtain the CANN installation package, refer to the [CANN Installation](https://www.hiascend.com/cann).	13	+Provide users with wheel package to quickly install TorchNPU. Before installing TorchNPU, complete the installation of CANN according to [Ascend Auxiliary Software](#ascend-auxiliary-software). To obtain the CANN installation package, refer to the [CANN Installation](https://www.hiascend.com/cann).
14		14
15	1. Install PyTorch	15	1. Install PyTorch
16		16
17	-Install PyTorch through pip.	17	+ Install PyTorch through pip.
18		18
19	-For Aarch64:	19	+ For Aarch64:
20		20
21	-```bash	21	+ ```bash
22	-pip3 install torch==2.7.1	22	+ pip3 install torch==2.7.1
23	-```	23	+ ```
24		24
25	-For x86:	25	+ For x86:
26		26
27	-```bash	27	+ ```bash
28	-pip3 install torch==2.7.1+cpu --index-url https://download.pytorch.org/whl/cpu	28	+ pip3 install torch==2.7.1+cpu --index-url https://download.pytorch.org/whl/cpu
29	-```	29	+ ```
30		30
31	-2. Install torch-npu dependencies	31	+2. Install TorchNPU dependencies
32		32
33	-Run the following command to install dependencies.	33	+ Run the following command to install dependencies.
34		34
35	-```bash	35	+ ```bash
36	-pip3 install pyyaml	36	+ pip3 install pyyaml
37	-pip3 install setuptools	37	+ pip3 install setuptools
38	-```	38	+ ```
39		39
40	-If the installation fails, use the download link or visit the [PyTorch official website](https://pytorch.org/) to download the installation package of the corresponding version.	40	+ If the installation fails, use the download link or visit the [PyTorch official website](https://pytorch.org/) to download the installation package of the corresponding version.
41		41
42	-\| OS arch \| Python version \| link \|	42	+ \| OS arch \| Python version \| link \|
43	-\|---------\|----------------\|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|	43	+ \|---------\|----------------\|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|
44	-\| x86 \| Python3.9 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp39-cp39-manylinux_2_28_x86_64.whl#sha256=d205cac087d60bc176bdc0b63a1d00dc7a4ee5ac76fd20a2ca318ac65674167e) \|	44	+ \| x86 \| Python3.9 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp39-cp39-manylinux_2_28_x86_64.whl#sha256=d205cac087d60bc176bdc0b63a1d00dc7a4ee5ac76fd20a2ca318ac65674167e) \|
45	-\| x86 \| Python3.10 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp310-cp310-manylinux_2_28_x86_64.whl#sha256=1f04a373a3f643821f721da9898ef77dce73b5b6bfc64486f0976f7fb5f90e83) \|	45	+ \| x86 \| Python3.10 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp310-cp310-manylinux_2_28_x86_64.whl#sha256=1f04a373a3f643821f721da9898ef77dce73b5b6bfc64486f0976f7fb5f90e83) \|
46	-\| x86 \| Python3.11 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl#sha256=a1684793e352f03fa14f78857e55d65de4ada8405ded1da2bf4f452179c4b779) \|	46	+ \| x86 \| Python3.11 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp311-cp311-manylinux_2_28_x86_64.whl#sha256=a1684793e352f03fa14f78857e55d65de4ada8405ded1da2bf4f452179c4b779) \|
47	-\| aarch64 \| Python3.9 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp39-cp39-manylinux_2_28_aarch64.whl#sha256=a4551cb97b83df5f93fc0d7538332535828581e1db2f179afc287027afbdd6e8) \|	47	+ \| aarch64 \| Python3.9 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp39-cp39-manylinux_2_28_aarch64.whl#sha256=a4551cb97b83df5f93fc0d7538332535828581e1db2f179afc287027afbdd6e8) \|
48	-\| aarch64 \| Python3.10 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp310-cp310-manylinux_2_28_aarch64.whl#sha256=c0df17cee97653d09a4e84488a33d21217f9b24208583c55cf28f0045aab0766) \|	48	+ \| aarch64 \| Python3.10 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp310-cp310-manylinux_2_28_aarch64.whl#sha256=c0df17cee97653d09a4e84488a33d21217f9b24208583c55cf28f0045aab0766) \|
49	-\| aarch64 \| Python3.11 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp311-cp311-manylinux_2_28_aarch64.whl#sha256=5fe6045b8f426bf2d0426e4fe009f1667a954ec2aeb82f1bd0bf60c6d7a85445) \|	49	+ \| aarch64 \| Python3.11 \| [link](https://download.pytorch.org/whl/cpu/torch-2.7.1%2Bcpu-cp311-cp311-manylinux_2_28_aarch64.whl#sha256=5fe6045b8f426bf2d0426e4fe009f1667a954ec2aeb82f1bd0bf60c6d7a85445) \|
50		50
51	-3. Install torch-npu	51	+3. Install TorchNPU
52		52
53	-```	53	+ ```bash
54	-pip3 install torch-npu==2.7.1	54	+ pip3 install torch-npu==2.7.1
55	-	55	+ ```
56	-```
57		56
58	### From Source	57	### From Source
59		58
60	-In some special scenarios, users may need to compile torch-npu by themselves.Select a branch in table [Ascend Auxiliary Software](#ascend-auxiliary-software) and a Python version in table [PyTorch and Python Version Matching Table](#pytorch-and-python-version-matching-table) first. The docker image is recommended for compiling torch-npu through the following steps(It is recommended to mount the working path only and avoid the system path to reduce security risks.), the generated .whl file path is ./dist/. Note that gcc version has the following constraints if you try to compile without using docker image: we recommend to use gcc 11.2 for both ARM and X86.	59	+In some special scenarios, users may need to compile TorchNPU by themselves. Select a branch in table [Ascend Auxiliary Software](#ascend-auxiliary-software) and a Python version in table [PyTorch and Python Version Matching Table](#pytorch-and-python-version-matching-table) first. The docker image is recommended for compiling TorchNPU through the following steps (It is recommended to mount the working path only and avoid the system path to reduce security risks.), the generated .whl file path is ./dist/. Note that gcc version has the following constraints if you try to compile without using docker image: we recommend to use gcc 11.2 for both ARM and X86.
61		60
62	-1. Clone torch-npu	61	+1. Clone TorchNPU
63		62
64	- ```	63	+ ```bash
65	git clone https://github.com/ascend/pytorch.git -b 2.7.1-7.2.0 --depth 1	64	git clone https://github.com/ascend/pytorch.git -b 2.7.1-7.2.0 --depth 1
66	```	65	```
67		66
68	2. Build Docker Image	67	2. Build Docker Image
69		68
70	- ```	69	+ ```bash
71	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM	70	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM
72	docker build -t manylinux-builder:v1 .	71	docker build -t manylinux-builder:v1 .
73	```	72	```
74		73
75	3. Enter Docker Container	74	3. Enter Docker Container
76		75
77	- ```	76	+ ```bash
78	docker run -it -v /{code_path}/pytorch:/home/pytorch manylinux-builder:v1 bash	77	docker run -it -v /{code_path}/pytorch:/home/pytorch manylinux-builder:v1 bash
79	- # {code_path} is the torch_npu source code path	78	+ # {code_path} is the TorchNPU source code path
80	```	79	```
81		80
82	-4. Compile torch-npu	81	+4. Compile TorchNPU
83		82
84	Take Python 3.9 as an example.	83	Take Python 3.9 as an example.
85		84
	@@ -92,13 +91,13 @@ In some special scenarios, users may need to compile torch-npu by themselves
92		91
93	If you would like to compile with new C++ ABI, then first run this command, at this point, the recommended compilation environment is same to community torch package: glibc 2.28, gcc 11.2.1	92	If you would like to compile with new C++ ABI, then first run this command, at this point, the recommended compilation environment is same to community torch package: glibc 2.28, gcc 11.2.1
94		93
95	- ```	94	+ ```bash
96	export _GLIBCXX_USE_CXX11_ABI=1	95	export _GLIBCXX_USE_CXX11_ABI=1
97	```	96	```
98		97
99	Meanwhile, we support configuring -fabi-version using the following variables，require consistency with the community torch package	98	Meanwhile, we support configuring -fabi-version using the following variables，require consistency with the community torch package
100		99
101	- ```	100	+ ```bash
102	export _ABI_VERSION=16	101	export _ABI_VERSION=16
103	```	102	```
104		103
	@@ -130,7 +129,7 @@ print(z)
130		129
131	## User Manual	130	## User Manual
132		131
133	-Refer to [API of Ascend Extension for PyTorch](docs/api/torch_npu_apis.md) for more detailed information.	132	+Refer to [API of TorchNPU](docs/zh/native_apis/menu_pt_native_apis.md) for more detailed information.
134		133
135	## PyTorch and Python Version Matching Table	134	## PyTorch and Python Version Matching Table
136		135
	@@ -146,7 +145,6 @@ Refer to [API of Ascend Extension for PyTorch](docs/api/torch_npu_apis.md) for m
146	\| PyTorch2.7.1 \| Python3.9.x,Python3.10.x,Python3.11.x \|	145	\| PyTorch2.7.1 \| Python3.9.x,Python3.10.x,Python3.11.x \|
147	\| PyTorch2.8.0 \| Python3.9.x,Python3.10.x,Python3.11.x \|	146	\| PyTorch2.8.0 \| Python3.9.x,Python3.10.x,Python3.11.x \|
148		147
149	-
150	## Ascend Auxiliary Software	148	## Ascend Auxiliary Software
151		149
152	PyTorch Extension versions follow the naming convention `{PyTorch version}-{Ascend version}`, where the former represents the PyTorch version compatible with the PyTorch Extension, and the latter is used to match the CANN version. The detailed matching is as follows:	150	PyTorch Extension versions follow the naming convention `{PyTorch version}-{Ascend version}`, where the former represents the PyTorch version compatible with the PyTorch Extension, and the latter is used to match the CANN version. The detailed matching is as follows:
	@@ -212,6 +210,7 @@ Refer to [API of Ascend Extension for PyTorch](docs/api/torch_npu_apis.md) for m
212	## Hardware support	210	## Hardware support
213		211
214	The Ascend training device includes the following models, all of which can be used as training environments for PyTorch models	212	The Ascend training device includes the following models, all of which can be used as training environments for PyTorch models
		213	+
215	\| Product series \| Product model \|	214	\| Product series \| Product model \|
216	\|-----------------------\|----------------------------------\|	215	\|-----------------------\|----------------------------------\|
217	\| Atlas Training series products \| Atlas 800（model: 9000） \|	216	\| Atlas Training series products \| Atlas 800（model: 9000） \|
	@@ -225,13 +224,14 @@ The Ascend training device includes the following models, all of which can be us
225	\| \| Atlas 300T A2 \|	224	\| \| Atlas 300T A2 \|
226		225
227	The Ascend inference device includes the following models, all of which can be used as inference environments for large models	226	The Ascend inference device includes the following models, all of which can be used as inference environments for large models
		227	+
228	\| Product series \| Product model \|	228	\| Product series \| Product model \|
229	\|-----------------------\|----------------------------------\|	229	\|-----------------------\|----------------------------------\|
230	\| Atlas 800I A2 Inference product \| Atlas 800I A2 \|	230	\| Atlas 800I A2 Inference product \| Atlas 800I A2 \|
231		231
232	## Pipeline Status	232	## Pipeline Status
233		233
234	-Due to the asynchronous development mechanism of upstream and downstream, incompatible modifications in upstream may cause some functions of torch_npu to be unavailable (only upstream and downstream development branches are involved, excluding stable branches). Therefore, we built a set of daily tasks that make it easy to detect relevant issues in time and fix them within 48 hours (under normal circumstances), providing users with the latest features and stable quality.	234	+Due to the asynchronous development mechanism of upstream and downstream, incompatible modifications in upstream may cause some functions of TorchNPU to be unavailable (only upstream and downstream development branches are involved, excluding stable branches). Therefore, we built a set of daily tasks that make it easy to detect relevant issues in time and fix them within 48 hours (under normal circumstances), providing users with the latest features and stable quality.
235		235
236	\| OS \| CANN Version(Docker Image) \| Upstream Branch \| Downstream Branch \| Period \| Status \|	236	\| OS \| CANN Version(Docker Image) \| Upstream Branch \| Downstream Branch \| Period \| Status \|
237	\| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \|	237	\| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \|
	@@ -252,7 +252,7 @@ The version branches of AscendPyTorch have the following maintenance phases:
252	\| Maintained \| 1 year/3.5 years \| Regular Release branch for 1 year, Long Term Support branch maintenance for 3.5 years. Fix major issues, do not incorporate new features, and release patch versions based on the impact of fixed bugs. \|	252	\| Maintained \| 1 year/3.5 years \| Regular Release branch for 1 year, Long Term Support branch maintenance for 3.5 years. Fix major issues, do not incorporate new features, and release patch versions based on the impact of fixed bugs. \|
253	\| End Of Life (EOL) \| N/A \| Do not accept any modification to a branch. \|	253	\| End Of Life (EOL) \| N/A \| Do not accept any modification to a branch. \|
254		254
255	-## PyTorch Maintenance Policies	255	+## PyTorch Maintenance Policies
256		256
257	\| PyTorch \| Maintenance Policies \| Status \| Launch Date \| Subsequent Status \| EOL Date \|	257	\| PyTorch \| Maintenance Policies \| Status \| Launch Date \| Subsequent Status \| EOL Date \|
258	\|-------------\|--------------------------\|-------------\|-----------------\|-------------------------------------------------------------------\|--------------\|	258	\|-------------\|--------------------------\|-------------\|-----------------\|-------------------------------------------------------------------\|--------------\|
	@@ -272,7 +272,7 @@ The version branches of AscendPyTorch have the following maintenance phases:
272		272
273	## Reference Documents	273	## Reference Documents
274		274
275	-For more detailed information on installation guides, model migration, training/inference tutorials, and API lists, please refer to the [Ascend Extension for PyTorch on the HiAI Community](https://www.hiascend.com/software/ai-frameworks?framework=pytorch).	275	+For more detailed information on installation guides, model migration, training/inference tutorials, and API lists, please refer to the [TorchNPU on the HiAI Community](https://www.hiascend.com/software/ai-frameworks?framework=pytorch).
276		276
277	\| Document Name \| Document Link \|	277	\| Document Name \| Document Link \|
278	\|------------------------------------------\|-------------------------------------------------------------------------------------------------------------------------\|	278	\|------------------------------------------\|-------------------------------------------------------------------------------------------------------------------------\|
	@@ -283,4 +283,4 @@ For more detailed information on installation guides, model migration, training/
283		283
284	## License	284	## License
285		285
286	-Ascend Extension for PyTorch has a BSD-style license, as found in the [LICENSE](LICENSE) file.	286	+TorchNPU has a BSD-style license, as found in the [LICENSE](LICENSE) file.

MREADME.zh.md+13-13

	@@ -1,8 +1,8 @@
1	-# Ascend Extension for PyTorch插件	1	+# TorchNPU
2		2
3	## 简介	3	## 简介
4		4
5	-本项目开发了名为torch_npu的Ascend Extension for PyTorch插件，使昇腾NPU可以适配PyTorch框架，为使用PyTorch框架的开发者提供昇腾AI处理器的超强算力。	5	+TorchNPU是为PyTorch框架开发的适配插件，旨在将昇腾NPU接入PyTorch生态，为使用PyTorch框架的开发者提供昇腾AI处理器的超强算力。
6		6
7	昇腾为基于华为昇腾处理器和软件的行业应用及服务提供全栈AI计算基础设施。您可以通过访问[昇腾社区](https://www.hiascend.com/zh/)，了解关于昇腾的更多信息。	7	昇腾为基于华为昇腾处理器和软件的行业应用及服务提供全栈AI计算基础设施。您可以通过访问[昇腾社区](https://www.hiascend.com/zh/)，了解关于昇腾的更多信息。
8		8
	@@ -26,28 +26,28 @@
26		26
27	## 版本说明	27	## 版本说明
28		28
29	-Ascend Extension for PyTorch的版本说明包含版本配套说明、版本兼容性说明和更新说明等，具体请参见《[Ascend Extension for PyTorch 版本说明](docs/zh/release_notes/release_notes.md)》。	29	+TorchNPU的版本说明包含版本配套说明、版本兼容性说明和更新说明等，具体请参见《[TorchNPU 版本说明](docs/zh/release_notes/release_notes.md)》。
30		30
31	## 环境部署	31	## 环境部署
32		32
33	-Ascend Extension for PyTorch插件的安装操作，具体请参见《[Ascend Extension for PyTorch 软件安装](docs/zh/installation_guide/menu_installation_guide.md)》。	33	+TorchNPU插件的安装操作，具体请参见《[TorchNPU 软件安装](docs/zh/installation_guide/menu_installation_guide.md)》。
34		34
35	## 快速入门	35	## 快速入门
36		36
37	-以CNN模型为例，介绍将其迁移至昇腾NPU上进行训练的方法，具体操作请参见《[Ascend Extension for PyTorch 快速入门](docs/zh/quick_start/quick_start.md)》。	37	+以CNN模型为例，介绍将其迁移至昇腾NPU上进行训练的方法，具体操作请参见《[TorchNPU 快速入门](docs/zh/quick_start/quick_start.md)》。
38		38
39	## 特性介绍	39	## 特性介绍
40		40
41	-Ascend Extension for PyTorch插件从内存资源优化、通信性能优化、计算性能优化、辅助报错定位等方面精心打造了一系列独特的特性，具体特性指导请参见《[PyTorch 框架特性指南](docs/zh/framework_feature_guide_pytorch/menu_framework_feature.md)》。	41	+TorchNPU插件从内存资源优化、通信性能优化、计算性能优化、辅助报错定位等方面精心打造了一系列独特的特性，具体特性指导请参见《[PyTorch 框架特性指南](docs/zh/framework_feature_guide_pytorch/menu_framework_feature.md)》。
42		42
43	## API参考	43	## API参考
44		44
45	- 原生PyTorch API在昇腾NPU设备上的支持情况请参见《[PyTorch 原生API支持度](docs/zh/native_apis/menu_pt_native_apis.md)》。	45	- 原生PyTorch API在昇腾NPU设备上的支持情况请参见《[PyTorch 原生API支持度](docs/zh/native_apis/menu_pt_native_apis.md)》。
46	-- Ascend Extension for PyTorch插件提供了部分自定义API接口，具体使用请参见《[Ascend Extension for PyTorch自定义API](https://gitcode.com/Ascend/op-plugin/blob/master/docs/zh/custom_APIs/menu_Pytorch_API.md)》。	46	+- TorchNPU插件提供了部分自定义API接口，具体使用请参见《[TorchNPU自定义API](https://gitcode.com/Ascend/op-plugin/blob/master/docs/zh/custom_APIs/menu_Pytorch_API.md)》。
47		47
48	## 分支维护策略	48	## 分支维护策略
49		49
50	-Ascend Extension for PyTorch版本分支的维护阶段如下：	50	+TorchNPU版本分支的维护阶段如下：
51		51
52	\| 状态 \| 时间 \| 说明 \|	52	\| 状态 \| 时间 \| 说明 \|
53	\| ------------------- \| -------- \| ------------------------------------------------ \|	53	\| ------------------- \| -------- \| ------------------------------------------------ \|
	@@ -77,7 +77,7 @@ Ascend Extension for PyTorch版本分支的维护阶段如下：
77		77
78	## 贡献指导	78	## 贡献指导
79		79
80	-介绍如何向Ascend Extension for PyTorch插件库贡献代码，具体请参见[Ascend Extension for PyTorch插件贡献指南](docs/zh/CONTRIBUTING.md)。	80	+介绍如何向TorchNPU插件库贡献代码，具体请参见[TorchNPU插件贡献指南](docs/zh/CONTRIBUTING.md)。
81		81
82	## 联系我们	82	## 联系我们
83		83
	@@ -85,11 +85,11 @@ Ascend Extension for PyTorch版本分支的维护阶段如下：
85		85
86	## 安全声明	86	## 安全声明
87		87
88	-Ascend Extension for PyTorch的系统安全加固、运行用户建议和文件权限控制等内容，请参见[Ascend Extension for PyTorch插件安全声明](docs/zh/SECURITYNOTE.md)。	88	+TorchNPU的系统安全加固、运行用户建议和文件权限控制等内容，请参见[TorchNPU插件安全声明](docs/zh/SECURITYNOTE.md)。
89		89
90	## 免责声明	90	## 免责声明
91		91
92	-致Ascend Extension for PyTorch插件使用者	92	+致TorchNPU插件使用者
93		93
94	- 本插件仅供调试和开发使用，使用者需自行承担使用风险，并理解以下内容：	94	- 本插件仅供调试和开发使用，使用者需自行承担使用风险，并理解以下内容：
95	- 数据处理及删除：用户在使用本插件过程中产生的数据属于用户责任范畴。建议用户在使用完毕后及时删除相关数据，以防信息泄露。	95	- 数据处理及删除：用户在使用本插件过程中产生的数据属于用户责任范畴。建议用户在使用完毕后及时删除相关数据，以防信息泄露。
	@@ -100,8 +100,8 @@ Ascend Extension for PyTorch的系统安全加固、运行用户建议和文件
100		100
101	## License	101	## License
102		102
103	-Ascend Extension for PyTorch插件的使用许可证，具体请参见[LICENSE](LICENSE)文件。	103	+TorchNPU插件的使用许可证，具体请参见[LICENSE](LICENSE)文件。
104		104
105	## 致谢	105	## 致谢
106		106
107	-感谢来自社区的每一个PR，欢迎贡献Ascend Extension for PyTorch插件！	107	+感谢来自社区的每一个PR，欢迎贡献TorchNPU插件！

Mbenchmarks/v1sparsedrive/README.md+1-1

	@@ -1,4 +1,4 @@
1	-# SimpleV1SparseDrive (torch-npu / Ascend)	1	+# SimpleV1SparseDrive (TorchNPU / Ascend)
2		2
3	---	3	---
4		4

Mci/docker/README.md+8-4

	@@ -1,10 +1,10 @@
1	-## Ascend Pytorch Dockerfile Repository	1	+# Ascend Pytorch Dockerfile Repository
2		2
3	This folder hosts the `Dockerfile` to build docker images with various platforms.	3	This folder hosts the `Dockerfile` to build docker images with various platforms.
4		4
5	-### Build torch_npu from Docker container	5	+## Build TorchNPU from Docker container
6		6
7	-Clone torch-npu	7	+Clone TorchNPU
8		8
9	```Shell	9	```Shell
10	git clone https://gitcode.com/ascend/pytorch.git --depth 1	10	git clone https://gitcode.com/ascend/pytorch.git --depth 1
	@@ -16,7 +16,9 @@ git clone https://gitcode.com/ascend/pytorch.git --depth 1
16	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM	16	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM
17	docker build -t manylinux-builder:v1 .	17	docker build -t manylinux-builder:v1 .
18	```	18	```
		19	+
19	If you want to configure the environment of LCOV, please build docker image like this:	20	If you want to configure the environment of LCOV, please build docker image like this:
		21	+
20	```Shell	22	```Shell
21	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM	23	cd pytorch/ci/docker/{arch} # {arch} for X86 or ARM
22	docker build -t manylinux-builder:v1 --build-arg CONFIG_FOR_LCOV=1 .	24	docker build -t manylinux-builder:v1 --build-arg CONFIG_FOR_LCOV=1 .
	@@ -28,9 +30,11 @@ docker build -t manylinux-builder:v1 --build-arg CONFIG_FOR_LCOV=1 .
28	docker run -it -v /{code_path}/pytorch:/home/pytorch manylinux-builder:v1 bash	30	docker run -it -v /{code_path}/pytorch:/home/pytorch manylinux-builder:v1 bash
29	# {code_path} is the torch_npu source code path	31	# {code_path} is the torch_npu source code path
30	```	32	```
31	-Compile torch_npu	33	+
		34	+Compile TorchNPU
32		35
33	Take Python 3.9 as an example	36	Take Python 3.9 as an example
		37	+
34	```Shell	38	```Shell
35	cd /home/pytorch	39	cd /home/pytorch
36	bash ci/build.sh --python=3.9	40	bash ci/build.sh --python=3.9

Mdocker/devel/README.md+3-3

	@@ -1,6 +1,6 @@
1	# TorchNPU 开发镜像	1	# TorchNPU 开发镜像
2		2
3	-本目录提供 Dockerfile 及构建脚本，用于生成 TorchNPU 的构建与开发镜像：`builder` 镜像提供 torch_npu 编译环境（不含 CANN），`dev` 镜像在此基础上叠加 CANN 运行环境。	3	+本目录提供 Dockerfile 及构建脚本，用于生成 TorchNPU 的构建与开发镜像：`builder` 镜像提供 TorchNPU 编译环境（不含 CANN），`dev` 镜像在此基础上叠加 CANN 运行环境。
4		4
5	## 1 镜像介绍	5	## 1 镜像介绍
6		6
	@@ -12,7 +12,7 @@ base manylinux + Python 软链接 + pip 源 + 基础系统包（curl/wget
12	└── dev builder + CANN（Toolkit + Ops + NNAL 可选） ← 运行环境	12	└── dev builder + CANN（Toolkit + Ops + NNAL 可选） ← 运行环境
13	```	13	```
14		14
15	-- builder（默认）：用于编译 torch_npu wheel，不含 CANN	15	+- builder（默认）：用于编译 TorchNPU wheel，不含 CANN
16	- dev：基于 builder，叠加 CANN 运行环境；继承全部编译工具链，可在容器内直接重新编译	16	- dev：基于 builder，叠加 CANN 运行环境；继承全部编译工具链，可在容器内直接重新编译
17		17
18	> Dockerfile会自动根据当前架构（ARM/X86）拉取对应镜像。	18	> Dockerfile会自动根据当前架构（ARM/X86）拉取对应镜像。
	@@ -111,7 +111,7 @@ bash builder.sh --cann --no-cache # 含 CANN 且不使用
111		111
112	## 3 启动容器	112	## 3 启动容器
113		113
114	-> torch_npu 编译无需 CANN/驱动；CANN 与 NPU 驱动仅在运行时（`import torch_npu`、调用 NPU 算子）需要。因此根据镜像类型选择不同的启动方式。	114	+> TorchNPU 编译无需 CANN/驱动；CANN 与 NPU 驱动仅在运行时（`import torch_npu`、调用 NPU 算子）需要。因此根据镜像类型选择不同的启动方式。
115	>	115	>
116	> 若已通过 `builder.sh` 构建，脚本会自动构建镜像并启动容器，可直接执行下方的 `docker exec` 命令进入容器；本节其余命令仅适用于手动执行 `docker build` 后需自行启动容器的场景。	116	> 若已通过 `builder.sh` 构建，脚本会自动构建镜像并启动容器，可直接执行下方的 `docker exec` 命令进入容器；本节其余命令仅适用于手动执行 `docker build` 后需自行启动容器的场景。
117		117

Mexamples/libtorch_triton_ascend/README.md+2-2

	@@ -2,9 +2,9 @@
2		2
3	## 环境约束	3	## 环境约束
4		4
5	-1）torch_npu 和 CANN 环境	5	+1）TorchNPU 和 CANN 环境
6		6
7	-本样例依赖 torch_npu 运行环境，推荐 torch_npu 26.1.0 及配套 CANN 9.1.0 版本。	7	+本样例依赖 TorchNPU 运行环境，推荐 TorchNPU 26.1.0 及配套 CANN 9.1.0 版本。
8		8
9	2）triton-ascend >= 3.2.2	9	2）triton-ascend >= 3.2.2
10		10

Mtest/README.md+19-7

	@@ -1,29 +1,38 @@
1	-## 简介	1	+# 简介
2	-test目录为PTA相关的测试用例。	2	+
		3	+test目录为TorchNPU相关的测试用例。
		4	+
3	## 安装依赖	5	## 安装依赖
4		6
5	`pip3 install -r requirements.txt`	7	`pip3 install -r requirements.txt`
6		8
7	## 补全脚本	9	## 补全脚本
		10	+
8	该操作需要联网	11	该操作需要联网
9	-```	12	+
		13	+```bash
10	cd test	14	cd test
11	bash get_synchronized_files.sh	15	bash get_synchronized_files.sh
12	```	16	```
		17	+
13	通过以上操作，会自动补齐testfiles_synchronized.txt和testfolder_synchronized.txt中的文件或文件夹。	18	通过以上操作，会自动补齐testfiles_synchronized.txt和testfolder_synchronized.txt中的文件或文件夹。
		19	+
14	## 跳过失败用例	20	## 跳过失败用例
		21	+
15	`export DISABLED_TESTS_FILE=./unsupported_test_cases/.pytorch-disabled-tests.json`	22	`export DISABLED_TESTS_FILE=./unsupported_test_cases/.pytorch-disabled-tests.json`
16		23
17	如果不是在test目录下运行测试用例，需要传入.pytorch-disabled-tests.json的绝对路径。	24	如果不是在test目录下运行测试用例，需要传入.pytorch-disabled-tests.json的绝对路径。
		25	+
18	## 执行方式	26	## 执行方式
		27	+
19	### 执行单个测试脚本	28	### 执行单个测试脚本
		29	+
20	运行以test开头的文件。以test_autocast.py为例：	30	运行以test开头的文件。以test_autocast.py为例：
21		31
22	方式一：	32	方式一：
23		33
24	`python test_autocast.py`	34	`python test_autocast.py`
25		35
26	-
27	方式二：	36	方式二：
28		37
29	`python run_test.py -i test_autocast`	38	`python run_test.py -i test_autocast`
	@@ -31,6 +40,7 @@ bash get_synchronized_files.sh
31	说明：部分以test开头的脚本不是直接运行的脚本，比如jit中的测试脚本是通过test_jit.py执行的。	40	说明：部分以test开头的脚本不是直接运行的脚本，比如jit中的测试脚本是通过test_jit.py执行的。
32		41
33	### 执行具体的用例	42	### 执行具体的用例
		43	+
34	通过-k参数传入具体的用例名。以test_autocast.py为例：	44	通过-k参数传入具体的用例名。以test_autocast.py为例：
35		45
36	方式一：	46	方式一：
	@@ -42,24 +52,26 @@ bash get_synchronized_files.sh
42	`python run_test.py -v -i test_autocast -- -k test_autocast_nn_fp32`	52	`python run_test.py -v -i test_autocast -- -k test_autocast_nn_fp32`
43		53
44	### 执行全量UT的方式	54	### 执行全量UT的方式
45	-```	55	+
		56	+```bash
46	# 进入到test目录的上一级	57	# 进入到test目录的上一级
47	cd ../	58	cd ../
48	```	59	```
49		60
50	运行非分布式全量用例：	61	运行非分布式全量用例：
51		62
52	-```	63	+```bash
53	python ci/access_control_test.py --all	64	python ci/access_control_test.py --all
54	```	65	```
55		66
56	运行分布式全量用例：	67	运行分布式全量用例：
57		68
58	-```	69	+```bash
59	python ci/access_control_test.py --distributed	70	python ci/access_control_test.py --distributed
60	```	71	```
61		72
62	## FAQ	73	## FAQ
		74	+
63	1. 报错："dictionary changed size during interation".	75	1. 报错："dictionary changed size during interation".
64		76
65	如果python 环境是3.8.1版本，报错在unitest/case.py中，可考虑是sys.modules被修改导致的。第三方包可能会有对sys.modules的修改，比如beartype。	77	如果python 环境是3.8.1版本，报错在unitest/case.py中，可考虑是sys.modules被修改导致的。第三方包可能会有对sys.modules的修改，比如beartype。

Mtorch_npu/_inductor/ascend_npu_ir/README.md+34-22

	@@ -1,41 +1,49 @@
1	# 简介	1	# 简介
		2	+
2	Torch 的 Inductor-MLIR 编译优化是基于 torch.compile 接口、融合 MLIR（Multi-Level Intermediate Representation）技术的新一代后端优化方案。Inductor-MLIR 是 PyTorch 编译体系对原生 Inductor 后端的扩展升级，在保留 torch.compile 动态图捕获 + 静态图优化 + 高效代码生成的核心逻辑的基础上，借助 MLIR 的多层级、跨架构编译能力，适配多类硬件架构的深度优化需求，显著提升不同算力平台下模型训练与推理的编译灵活性和性能上限。	3	Torch 的 Inductor-MLIR 编译优化是基于 torch.compile 接口、融合 MLIR（Multi-Level Intermediate Representation）技术的新一代后端优化方案。Inductor-MLIR 是 PyTorch 编译体系对原生 Inductor 后端的扩展升级，在保留 torch.compile 动态图捕获 + 静态图优化 + 高效代码生成的核心逻辑的基础上，借助 MLIR 的多层级、跨架构编译能力，适配多类硬件架构的深度优化需求，显著提升不同算力平台下模型训练与推理的编译灵活性和性能上限。
3	Inductor-MLIR 包含如下核心组件：	4	Inductor-MLIR 包含如下核心组件：
4	-\|组件\| 定位 \|作用\|
5	-\|-\|-\|-\|
6	-\|Dynamo \|前端编译器（代码转换器） \|延续 TorchDynamo 的核心能力，JIT 将用户的 eager（动态图）代码编译为 FX Graph，完成动态图到静态中间表示的捕获，为 Inductor-MLIR 后端提供统一的输入，保证前端编译逻辑的一致性\|
7	-\|Inductor-MLIR \|核心编译器（多层级 IR 优化器） \|作为 Inductor-MLIR 的核心层，将 FX Graph 转换为 MLIR 多层级中间表示（IR）；通过跨层级的优化策略（如算子拆分与融合、数据类型优化、计算图化简），兼顾编译灵活性与优化深度，适配不同硬件的编译特性\|
8	-\|MLIR CodeGen \|代码生成器（硬件感知代码生成器） \|接收优化后的 MLIR IR，针对目标硬件生成专属的高性能底层代码\|
9		5
10	-# 使用指导	6	+\| 组件 \| 定位 \| 作用 \|
		7	+\| -------------- \| ------------------------------ \| ------------------------------------------------------------------------------------------------------------------------ \|
		8	+\| Dynamo \| 前端编译器（代码转换器） \| 延续 TorchDynamo 的核心能力，JIT 将用户的 eager（动态图）代码编译为 FX Graph，完成动态图到静态中间表示的捕获，为 Inductor-MLIR 后端提供统一的输入，保证前端编译逻辑的一致性 \|
		9	+\| Inductor-MLIR \| 核心编译器（多层级 IR 优化器） \| 作为 Inductor-MLIR 的核心层，将 FX Graph 转换为 MLIR 多层级中间表示（IR）；通过跨层级的优化策略（如算子拆分与融合、数据类型优化、计算图化简），兼顾编译灵活性与优化深度，适配不同硬件的编译特性 \|
		10	+\| MLIR CodeGen \| 代码生成器（硬件感知代码生成器） \| 接收优化后的 MLIR IR，针对目标硬件生成专属的高性能底层代码 \|
		11	+
		12	+## 使用指导
		13	+
11	## 环境配置	14	## 环境配置
12		15
13	-* torch_npu：2.6.0或之后版本的torch_npu支持torch.compile接口；	16	+* TorchNPU：2.6.0或之后版本的TorchNPU支持torch.compile接口；
14	* torch-mlir：Inductor后端的MLIR模式依赖torch-mlir。用户可以从[LINK](https://repo.oepkgs.net/ascend/pytorch/vllm/torch/)获取torch-mlir软件包;	17	* torch-mlir：Inductor后端的MLIR模式依赖torch-mlir。用户可以从[LINK](https://repo.oepkgs.net/ascend/pytorch/vllm/torch/)获取torch-mlir软件包;
15		18
16	## 使用方式	19	## 使用方式
		20	+
17	MLIR使用方式有三种, 任选一种即可调用	21	MLIR使用方式有三种, 任选一种即可调用
18		22
19	1. config导入（torch.compile之前导入）	23	1. config导入（torch.compile之前导入）
20	-```	24	+
21	-torch._inductor.config.npu_backend == "mlir"	25	+ ```python
22	-torch.compile(op_calc)(x)	26	+ torch._inductor.config.npu_backend == "mlir"
23	-```	27	+ torch.compile(op_calc)(x)
		28	+ ```
24		29
25	2. 环境变量导入（npu初始化之前导入）	30	2. 环境变量导入（npu初始化之前导入）
26		31
27	-```	32	+ ```python
28	-import os	33	+ import os
29	-os.environ['TORCHINDUCTOR_NPU_BACKEND'] = 'mlir'	34	+ os.environ['TORCHINDUCTOR_NPU_BACKEND'] = 'mlir'
30	-torch.compile(op_calc)(x)	35	+ torch.compile(op_calc)(x)
31	-```	36	+ ```
		37	+
32	3. options导入	38	3. options导入
33	-```	39	+
34	-torch.compile(op_calc,options={"npu_backend":"mlir"})(x)	40	+ ```python
35	-```	41	+ torch.compile(op_calc,options={"npu_backend":"mlir"})(x)
		42	+ ```
		43	+
36	## 示例	44	## 示例
37		45
38	-```	46	+```python
39	import os	47	import os
40	# 环境变量调用需要在torch_npu初始化之前	48	# 环境变量调用需要在torch_npu初始化之前
41	os.environ['TORCHINDUCTOR_NPU_BACKEND'] = 'mlir'	49	os.environ['TORCHINDUCTOR_NPU_BACKEND'] = 'mlir'
	@@ -62,9 +70,12 @@ compile_func = torch.compile(op_calc, options={"npu_backend": "mlir"})
62	compile_out, codes = run_and_get_code(compile_func,x,y)	70	compile_out, codes = run_and_get_code(compile_func,x,y)
63	print(codes[0])	71	print(codes[0])
64	```	72	```
		73	+
65	## 输出mlir融合算子	74	## 输出mlir融合算子
		75	+
66	codes 为捕获到的 MLIR 输出代码片段，该代码中包含 call 函数与 benchmark_compiled_model 函数，其中 call 函数内部封装了 MLIR 编译生成的算子逻辑。	76	codes 为捕获到的 MLIR 输出代码片段，该代码中包含 call 函数与 benchmark_compiled_model 函数，其中 call 函数内部封装了 MLIR 编译生成的算子逻辑。
67	-```	77	+
		78	+```python
68	def call(args):	79	def call(args):
69	arg0_1, arg1_1 = args	80	arg0_1, arg1_1 = args
70	args.clear()	81	args.clear()
	@@ -77,4 +88,5 @@ def call(args):
77	del arg1_1	88	del arg1_1
78	return (buf0, )	89	return (buf0, )
79	```	90	```
80	-其中mlir_fused_mul_0.run 是 MLIR 编译生成的核心执行接口，用于在指定 NPU 设备和计算流上执行融合乘法算子计算，接收输入张量并将计算结果写入输出张量，是 MLIR 对原始乘法逻辑优化后的底层实现。	91	+
		92	+其中mlir_fused_mul_0.run 是 MLIR 编译生成的核心执行接口，用于在指定 NPU 设备和计算流上执行融合乘法算子计算，接收输入张量并将计算结果写入输出张量，是 MLIR 对原始乘法逻辑优化后的底层实现。

Mtorch_npu/_inductor/mfusion/README.md+2-2

	@@ -11,12 +11,12 @@ MFusion 包含如下核心组件：
11	\| subgraph_registry \| 子图注册表 \| 管理融合子图的元数据，包括 FX GraphModule 和动态/静态标记 \|	11	\| subgraph_registry \| 子图注册表 \| 管理融合子图的元数据，包括 FX GraphModule 和动态/静态标记 \|
12	\| decomp \| 算子分解 \| 提供 Inductor 融合前的算子预处理，确保融合路径的正确性 \|	12	\| decomp \| 算子分解 \| 提供 Inductor 融合前的算子预处理，确保融合路径的正确性 \|
13		13
14	-# 使用指导	14	+## 使用指导
15		15
16	## 环境配置	16	## 环境配置
17		17
18	* mfusion：MFusion 组件的 Python wheel 包	18	* mfusion：MFusion 组件的 Python wheel 包
19	-* torch_npu：2.7.1 和 2.9.0 支持 MFusion 功能	19	+* TorchNPU：2.7.1 和 2.9.0 支持 MFusion 功能
20	* torch-mlir：MFusion 依赖 torch-mlir 进行 MLIR 代码生成，用户可从 [LINK](https://repo.oepkgs.net/ascend/pytorch/vllm/torch/) 获取	20	* torch-mlir：MFusion 依赖 torch-mlir 进行 MLIR 代码生成，用户可从 [LINK](https://repo.oepkgs.net/ascend/pytorch/vllm/torch/) 获取
21		21
22	## 环境变量	22	## 环境变量