Huawei CANN (Compute Architecture for Neural Networks) and MindSpore open-source AI framework

PRIMARY OPEN-SOURCE SOURCE. CANN is Huawei's heterogeneous computing architecture for AI scenarios, providing multi-layer programming interfaces for the Ascend platform. It supports AscendCL (Ascend Computing Language) APIs for device management, context management, stream management, memory management, model loading and execution, operator loading and execution, and media data processing. MindSpore is Huawei's open-source deep learning training/inference framework with native support for Ascend AI processors and software-hardware co-optimization. It supports mobile, edge, and cloud scenarios. The CANN samples repository (https://github.com/Huawei-Ascend/samples) provides C++ and Python samples developed based on AscendCL APIs. Huawei announced the CANN open-source ecosystem at an Ascend computing industry summit in Beijing (reported by China Daily, August 2025). The MindSpore GitHub repository (https://github.com/mindspore-ai/mindspore) is the official open-source repository. CANN documentation is available at https://www.hiascend.com/document and https://support.huaweicloud.com/intl/en-us/ascenddocument/index.html.