Hewlett Packard Enterprise recently launched the HPE ProLiant Compute XD685, a cutting-edge system engineered to improve the training of large and complex AI models. By leveraging the latest from AMD, including 5th Gen AMD EPYC processors and Instinct MI325X accelerators, this solution aims to meet the growing demands of AI model builders.
Designed for high-performance AI clusters, this new system excels in tasks like large language model training, natural language processing, and multi-modal training. HPE is targeting AI service providers, government bodies, and large model builders, offering a high-performance solution that ensures quick deployment. Trish Damkroger of HPE emphasized the need for strong-scaling and parallel computing capabilities. She stated, “As a leader in delivering the world’s most powerful and energy-efficient systems, together with AMD, we are extending this innovation to the new HPE ProLiant Compute XD685 to support a growing market of AI model builders and accelerate scientific and engineering breakthroughs across industries.”
Forrest Norrod from AMD echoed this sentiment, highlighting the collaboration’s focus on accelerating time-to-market and maximizing AI performance efficiency. The system’s novel 5U chassis speeds up delivery times and offers hardware and software flexibility. Various cooling options, including air and direct liquid cooling, enhance energy efficiency, a benefit drawn from HPE’s experience with liquid-cooled environments.
What sets this system apart is its ability to run large AI models using fewer GPUs, thanks to AMD’s CDNA 3 architecture. Notably, the Instinct MI325X accelerators feature HBM3e memory with a bandwidth of 6 TB/s, which boosts performance while lowering the total cost of ownership. The direct liquid cooling option helps manage increasing power needs and supports sustainability efforts while reducing operational expenses.
Adding to the appeal of the ProLiant Compute XD685 is its comprehensive suite of services, provided by HPE Services, for setting up large AI clusters globally. This includes configuration, validation, and testing, aimed at reducing deployment times. Security is a focus too, with HPE iLO providing advanced security management from the production level, embedded into the silicon.
Maximizing rack density, the XD685 supports configurations with either the MI325X or MI300X accelerators, alongside the latest AMD EPYC 9005 series processors. HPE’s Performance Cluster Manager simplifies system management by automating setup and maintaining cluster health.
Additionally, HPE announced 30 new world records achieved on its ProLiant Gen11 servers featuring AMD EPYC 9005 Series processors. These achievements highlight the server’s superior performance and power efficiency in medium-to-large enterprise workloads.
