Huawei Uncovers Full Open-Source Release of Ascend Chip Software
(MENAFN) Chinese tech powerhouse Huawei has unveiled the complete open-source release of its Ascend chip software ecosystem, aiming to empower users to tap into its advanced capabilities and independently pursue tailored development.
The software, known as CANN (Compute Architecture for Neural Networks), functions as a critical interface linking high-level AI training frameworks with Ascend chips. This allows users to harness the chip’s computing power without navigating intricate chip-level details.
The announcement came during an Ascend computing industry summit held Tuesday in Beijing. Attendees included representatives from top AI companies, academic institutions, research organizations, and Huawei partners, who collectively launched an initiative to cultivate the CANN open-source ecosystem. Their goal is to combine industry expertise, push AI boundaries, and foster a robust chip ecosystem.
Huawei’s Rotating Chairman Xu Zhijun stated that computing power continues to be central to the company’s AI strategy.
While most major computing power providers offer comparable software, Huawei’s approach diverges sharply from Nvidia’s CUDA—a proprietary ecosystem developed over nearly two decades by the U.S. chipmaker. Instead, Huawei’s CANN embraces full open source.
Speaking to media, Huawei revealed that after seven years of development, CANN has delivered breakthroughs in computing optimization, communication efficiency, and memory management. The platform now supports the entire AI model training and deployment pipeline.
This release marks the latest effort by Huawei to share its proprietary technologies publicly, following the open-sourcing of its HarmonyOS operating system and MindSpore AI framework.
Legal Disclaimer:
MENAFN provides the
information “as is” without warranty of any kind. We do not accept
any responsibility or liability for the accuracy, content, images,
videos, licenses, completeness, legality, or reliability of the information
contained in this article. If you have any complaints or copyright
issues related to this article, kindly contact the provider above.
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
