HPE ProLiant Compute XD685 Direct Liquid Cooling Server

Do you need a highly performant, secure and energy-efficient AI cluster? The HPE ProLiant Compute XD685 leverages a purpose-built architecture and is designed to accelerate large, complex AI model training and tuning workloads. Powered by eight of the latest accelerators from NVIDIA, this server is refined for use in large language model training, natural language processing, and multimodal training. Optional direct liquid cooling (DLC) helps organizations meet escalating power requirements, advance sustainability goals, and lower operational costs. Managed by HPE iLO, the server offers seamless management and robust security. HPE Services provides expertise for global installation and quick deployment of large AI clusters, offering stability and operational excellence.

What is new supporting point:
Support for eight NVIDIA H200 SXM Tensor Core GPUs air-cooled or liquid-cooled, or eight NVIDIA B200 GPUs liquid-cooled.
Modular 6-U chassis for Air, 5-U chassis for DLC designed by Hewlett Packard Enterprise
DLC (direct liquid cooling) or air-cooling options
24x DDR5 6400 RDIMMs and 12 channels per CPU DDR5 w/ ECC
Seamless management and built-in security features from HPE iLO 6
2x 5th Generation AMD EPYC™ processors