The training speed was increased by 2-3 times, and HPE joined hands with Nvidia to launch an one-stop GenAI supercomputing solution

CTOnews.com, November 15 (Xinhua)-- Hewlett-Packard Corporation (HPE) recently issued a press release, saying that it has partnered with Nvidia to launch supercomputer solutions for generative AI to help large enterprises, research institutions and government organizations use private data sets to accelerate training and adapt AI models.

Tuyuan HPEHPE said it will officially launch worldwide in December 2023, increasing the training speed by 2-3 times through a "one-stop" supercomputing solution.

CTOnews.com learned from the report that the solution launched by HPE uses Nvidia's Grace Hopper GH200 superchip and integrates HPE Cray supercomputing technology, based on the same architecture of the HPE Frontier supercomputer.

Key components of this solution include new software tools that can be used to build AI applications, customize pre-built models, and take advantage of the ability to develop and modify code.

HPE and Nvidia work together to provide organizations with the scale and performance required for large AI workloads, such as LLM and Deep Learning recommendation Model (DLRM) training, to unlock AI faster by building water-cooled supercomputers, accelerated computing, networking, storage, and services.

Justin Hotard, executive vice president and general manager of HPC, AI and laboratories at HPE, said: "the world's leading companies and research centers are training and adapting AI models to drive innovation and unlock research breakthroughs, but to do this effectively and efficiently, they need specially built solutions."