Documentation Center
Directory Structure
The Docs directory structure is described as follows:
├── zh
├── context # Public documents, such as terminology, basic concepts, and so on
├── debug # Operator debugging guidance documents
│ ├── cann_sim.md
│ ├── op_debug_prof.md
│ └── ...
├── develop # Operator development guidance documents
│ ├── aicore_develop_guide.md
│ ├── aicpu_develop_guide.md
│ ├── cross_platform_migration_guide.md
│ ├── graph_develop_guide.md
│ └── ...
├── figures # Image directory
├── install # Environment installation and compilation guidance documents
│ ├── build.md
│ ├── compile.md
│ ├── dir_structure.md
│ ├── quick_install.md
│ └── ...
├── invocation # Operator invocation guidance documents (including aclnn invocation, graph mode invocation, and so on)
│ ├── quick_op_invocation.md
│ ├── op_invocation.md
│ └── ...
├── op_api_list.md # Complete operator interface list (aclnn)
├── op_list.md # Complete operator list
├── CONTRIBUTING_DOCS.md # Documentation contribution guide
├── QUICKSTART.md # Quick start
└── README.md
Advanced Tutorials
Guide Documents
| Document |
Description |
| Source Code Build Guide |
Introduces different source code build methods and verification methods in online and offline scenarios. |
| Operator Invocation Guide |
Introduces the method of invoking operator samples and different operator invocation methods (such as aclnn/graph, and so on). |
| Standard Operator Development Guide |
Introduces how to define operator prototypes and implement Tiling and Kernel based on standard engineering. Such operators are called "standard operators". Standard operators support aclnn and graph mode invocation. |
| Simple Operator Development Guide |
Introduces how to implement fast_kernel_launch based on simple engineering, that is, the <<<>>> method. Such operators are called "simple operators". Simple operators only support PyTorch invocation. |
| Operator Debugging and Tuning |
Introduces common operator function debugging and performance tuning methods (such as data collection and simulation pipeline, and so on). |
API Documents
| Document |
Description |
| Operator List |
Introduces the list of all operators included in the project. |
| aclnn List |
Introduces the list of all operator aclnn APIs included in the project. To facilitate users to invoke operators on the Host side, C language APIs are provided, that is, APIs with the aclnn prefix. |
| Document |
Description |
| Simulator Tool |
A SoC-level simulation tool for operator development scenarios, used to analyze the precision and performance data of AI tasks running on the AI simulator at each stage. |
More Documents
- Sample Documents: Refer to the operator samples in the cann-samples repository.
Appendix
| Document |
Description |
| Operator Basic Concepts |
Introduces basic concepts and terminology in the operator domain, such as quantization/sparse, data types, data formats, and so on. |
| build Parameter Description |
Introduces the functions and parameter values of the build.sh script in this project, including source code compilation, operator invocation, debugging, and so on. |