HPE P35731-B21

R6B53C NVIDIA A100 40 GB Accelerator (PCIe)

The HPE P35731-B21 R6B53C NVIDIA A100 40 GB Accelerator is a powerful PCIe-based computational card engineered for high-performance workloads in modern server environments. Featuring the NVIDIA A100 GPU with 40 GB of high-bandwidth memory (HBM2), this Accelerator delivers exceptional performance for AI training and inference, HPC simulations, data analytics, and virtualized GPU environments. Integrated as a double-width, full-height plug-in card, it brings the industry-leading NVIDIA Ampere architecture into HPE ProLiant and similar server platforms, dramatically accelerating parallel workloads and deep learning tasks.

Features

Built around NNVIDIA’s Ampere GA100 GPU core, the A100 offers unparalleled precision and speed with a memory bandwidth exceeding 1.5 TB/s and support for FP64, FP32, FP16, and Tensor Core operations. The card is passively cooled and draws up to 250 W, requiring thoughtful airflow management within server chassis. It supports Multi-Instance GPU (MIG), enabling hardware-based partitioning into up to seven isolated GPU instances for optimized workload distribution. This versatility makes the A100 ideal for virtualized infrastructures, cloud platforms, AI clusters, and HPC installations where maximizing compute density and efficiency is critical. 

Advanced Options

The A100 Accelerator is designed for continuous, high-throughput operation. Its high memory bandwidth and ECC-enabled architecture ensure reliable processing of data-intensive tasks with minimal interruption. With support for MIG, administrators can allocate GPU resources efficiently across multiple workloads—enhancing utilization and isolation.