Open Architecture
Diverse interface types and samples that are compatible with mainstream AI frameworks
Superior Performance
High-performance operators, communication algorithms, and hardware-software synergy to maximize computing power
Easy to Use
Development system and operator template library, compatible with third-party operator frameworks

CANN Helps Build a Powerful Ascend AI Computing Platform

CANN is at the core of Huawei Ascend AI's basic software and hardware platform, designed to build an all-scenario Ascend AI platform with ultimate performance, intuitive development, and an open ecosystem.

AI Framework

Framework Adapter

Operator Library
Fused Operators for LLMs NN/CV/Math Basic Operators
Communication Library
Collective Communication Algorithms Distributed Communication
Graph Engine
Graph Compilation and Optimization Graph Execution Acceleration
Domain-specific Library
ATB, SiP, etc.
Programming Language
Ascend C (API & CATLASS)
BiSheng Compiler
Heterogeneous Compilation and Optimization | Instruction Affinity | Third-Party Languages (Triton, etc.)
Runtime
Control Flow | Memory Management | Job Scheduling
Driver
Board Driver | Accelerator Driver | Device Management

MindStudio

E2E development toolchain

Provides visualized debugging and tuning for training, inference, and operator development.

Ascend AI Processors

Learning Resources

Learning Map
All-in-one learning plan to accelerate developer growth
Online Courses
Structured courses to learn and use CANN with ease
Micro Certification
Comprehensive certification programs to improve coding capabilities
Experiment
Sample Code