Shulou(Shulou.com)11/24 Report--

Thanks to CTOnews.com netizens West window past, Mr. Aviation, South China Daniel Wu for the delivery of clues! According to CTOnews.com news on September 20, during Huawei full connection Conference 2023 held today, Wang Tao, Managing Director of Huawei, Chairman of ICT Infrastructure Business Management Committee and President of Enterprise BG, officially released the new structure of Atlas AI computing cluster-- Atlas 900SuperCluster, which can support large model training with over one trillion parameters.

According to reports, the new cluster uses the brand-new Huawei Xinghe AI smart switch CloudEngine XH16800, and with its high-density 800GE port capacity, the two-layer switching network can achieve 2250 nodes (equivalent to 18000 cards) super large-scale non-convergent cluster network.

The new cluster also uses an innovative super-node architecture, which greatly improves the training ability of large models. In addition, give full play to Huawei's comprehensive advantages in computing, network, storage, energy and other fields, comprehensively improve system reliability from device level, node level, cluster level and business level, and upgrade the stability of large model training from sky level to month level.

In addition, Huawei has released a more open and easy-to-use CANN 7.0heterogeneous computing architecture, which is not only fully compatible with the industry's AI framework, accelerator libraries and mainstream models, but also deeply opens up the underlying capabilities, so that the AI framework and accelerator libraries can more directly invoke and manage computing resources, allowing developers to customize high-performance operators, so that large models have differentiated competitiveness.

Huawei has also upgraded the Ascend C programming language to simplify operator implementation logic in a more efficient programming way, greatly shorten the development cycle of fusion operators, and enable the rapid development of AI models and applications.

For enterprises and developers around the world, Huawei Yunguan.com today officially launched the "hundred models" zone of Teng AI cloud service, which includes the mainstream open source models in the industry, and is fully adapted and optimized based on Penton AI cloud services; provides a tool chain for application development, and development tools have all been realized in the cloud, eliminating tedious configuration processes and realizing one-click access and instant use.

According to the CTOnews.com query of ▲ Teng AI Cloud Service, as of July this year, Penteng AI cluster has supported the construction of artificial intelligence computing centers in 25 cities across the country, of which 7 urban public computing platforms have been selected as the first batch of national "new generation artificial intelligence public computing open innovation platform".

At the same time, Teng AI has developed more than 30 hardware partners and more than 1200 ISV, and jointly launched more than 2500 industry AI solutions, serving operators, Internet, finance and other industries on a large scale.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.