HPE S5E91A - NVIDIA GB200 NVL72 HPE Compute Tray

Do you need a highly performant solution for training or inferencing of AI models over one trillion parameters? The NVIDIA GB200 NVL72 by HPE offers seamless integration of NVIDIA CPUs, GPUs, compute, switch trays, networking, and software. This powerful solution features a high-bandwidth, low-latency NVIDIA NVLink™ interconnect to create a 72 GPU single shared-memory space. Combined with HPE advanced direct liquid cooling technology and its industry-leading services and support, this offering delivers extreme performance, scalability, and efficiency for the largest AI models. HPE Services provides expertise for global installation and quick deployment of large AI clusters, helping enable stability and operational excellence, including vast experience with large direct-liquid cooled environments.

What is new supporting point:
72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs interconnected through high-speed NVIDIA NVLink™
Up to 13.5 TB total HBM3E memory with 576 TB/s bandwidth
48 RU MGX rack
NVIDIA BlueField-3 data processing units
HPE direct liquid cooling technology
HPE Services