HPE launches AI model training system powered by AMD Technology

Hewlett Packard Enterprise (HPE) has introduced the HPE ProLiant Compute XD685, a system designed for training large and complex artificial intelligence (AI) models. The system is powered by 5th Gen AMD EPYC processors and AMD Instinct MI325X accelerators. 

The HPE ProLiant Compute XD685 is built to support AI tasks like large language model training, natural language processing, and multi-modal training. It aims to help organisations deploy AI clusters quickly and efficiently, focusing on performance, security, and energy efficiency. 

The system uses a modular 5U chassis designed by HPE, allowing for flexibility in configuring CPUs, GPUs, components, and cooling options. The chassis also integrates direct liquid cooling technology, reducing power consumption and operational costs. This liquid cooling feature leverages HPE’s experience in deploying large-scale IT environments. 

The AMD Instinct MI325X accelerators, based on AMD's CDNA 3 architecture, provide high memory bandwidth and performance for AI training and inference tasks. These accelerators offer 6 TB/s memory bandwidth, enhancing AI performance while lowering total cost of ownership (TCO). 

HPE ProLiant Compute XD685 also comes with advanced security features through HPE’s Integrated Lights-Out (iLO) technology, which enables secure system management. 

HPE's Performance Cluster Manager is included to streamline setup and management, allowing organizations to deploy and maintain AI clusters efficiently. The system can support up to eight AMD Instinct MI325X or MI300X accelerators and two AMD EPYC 9005 Series processors. 

Forrest Norrod, executive vice president and general manager, Data Center Solutions Group, AMD, said, “Our collaboration with HPE continues to deliver flexible, highly performant solutions that help maximise AI efforts for increased demand in large language model training and advances competitive industry innovation.”