NVIDIA A16 64GB PCIe Gen4 GPU Quad-chip for HPE ProLiant DL380

The NVIDIA A16 64GB PCIe Gen4 GPU (HPE Part P40228-B21) is a quad-chip, high-density graphics and compute accelerator purpose-built for virtualized enterprise workloads running on HPE ProLiant DL380 Gen10 Plus servers. Based on NVIDIA's Ampere architecture, the A16 integrates four GA107 GPU dies on a single PCIe Gen4 card, delivering a combined 64GB of GDDR6 memory (16GB per GPU) and 1,024 CUDA cores per chip. The card is optimized for NVIDIA Virtual PC (vPC), Virtual Applications (vApps), and Virtual Workstation (vWS) profiles under the NVIDIA Virtual GPU (vGPU) software stack, enabling multiple concurrent virtual desktop and application sessions with hardware-level GPU isolation.

Designed for data center density and efficiency, the A16 operates within a 250W TDP envelope and uses a passive cooling design that relies on the server chassis airflow of HPE ProLiant systems. PCIe Gen4 x16 host connectivity provides the bandwidth headroom needed to support concurrent multi-user GPU virtualization without bottlenecking I/O. Each of the four A16 GPU dies supports independent vGPU partitioning, allowing administrators to allocate discrete GPU slices to individual virtual machines with guaranteed frame buffer and compute resources.

As an HPE factory-integrated option (HPE option kit P40228-B21), this GPU is validated and supported within the HPE ProLiant DL380 ecosystem, ensuring compatibility with HPE's server management tools, Intelligent Provisioning, and Integrated Lights-Out (iLO). It is particularly well-suited for enterprise IT environments in industries such as healthcare, financial services, oil and gas, and engineering that require scalable, secure, multi-user GPU-accelerated virtual desktops and remote workstation experiences across distributed or centralized data centers.

Ideal for

• Enterprise virtual desktop infrastructure (VDI) deployments requiring GPU-accelerated multi-user sessions via NVIDIA vGPU software on HPE ProLiant DL380 servers
• Remote workstation virtualization for engineering, media, and design teams needing dedicated GPU frame buffer per user without physical workstation hardware
• Healthcare imaging and clinical desktop virtualization where GPU-accelerated rendering of DICOM and diagnostic applications must be delivered securely to thin clients
• Financial services firms virtualizing GPU-intensive analytics dashboards and trading applications across centrally managed data center infrastructure
• Oil, gas, and geoscience environments virtualizing 3D seismic interpretation and reservoir modeling tools for geographically distributed teams
• Enterprise IT consolidation projects migrating physical GPU workstations to a centralized vGPU-enabled server infrastructure for simplified lifecycle management and security policy enforcement

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.