Huawei launches two new commercial AI large model storage products, supporting 12 million IOPS performance

IT House News on July 14th, Huawei recently released new commercial AI storage products “OceanStor A310 Deep Learning Data Lake Storage” and “FusionCube A3000 Training/Push Hyper-converged All-in-one Machine”. Officials said that “these two products can It provides new momentum for AI basic model training, industry model training, and segmented scene model training and reasoning."

▲ Picture source Huawei

IT House compiled and summarized:

OceanStor A310 deep learning data lake storage is mainly oriented to basic/industry large model data lake scenarios, realizing full-process massive data management of AI from data collection and preprocessing to model training and inference applications.

Officially stated that OceanStor A310 single frame 5U supports the industry’s highest bandwidth of 400GB/s and the highest performance of 12 million IOPS, can be linearly expanded to 4096 nodes, and can achieve multi-protocol lossless interoperability. The global file system GFS realizes cross-regional intelligent data weaving, simplifies the data collection process, realizes near-data preprocessing through near-memory computing, reduces data movement, and improves preprocessing efficiency by 30%.

FusionCube A3000 training/pushing hyper-converged all-in-one machine is mainly oriented to industry large model training/inference scenarios. For tens of billions of model applications, it integrates OceanStor A300 high-performance storage nodes, training/pushing nodes, switching equipment, AI platform software and Management and operation software provides large model partners with a turn-key deployment experience, enabling one-stop "out-of-the-box" delivery and deployment can be completed within 2 hours.

Officially stated that the all-in-one machine supports two flexible business models, including Huawei’s Ascend one-stop solution and third-party partner’s one-stop solution for open computing, network, and AI platform software. The training/push nodes and storage nodes of the all-in-one machine can be independently expanded horizontally to match the needs of models of different sizes. At the same time, FusionCube A3000 uses high-performance containers to share the GPU for multiple model training and inference tasks, increasing resource utilization from 40% to more than 70%.

Zhou Yuefeng, President of Huawei’s Data Storage Product Line, pointed out that in the era of large-scale models, data storage is an important infrastructure that supports large-scale artificial intelligence models. Huawei Data Storage will provide diversified solutions and products for the era of AI large models, and work with partners to overcome these difficulties and promote AI empowerment in various industries. ”

The above is the detailed content of Huawei launches two new commercial AI large model storage products, supporting 12 million IOPS performance. For more information, please follow other related articles on the PHP Chinese website!