Cisco HCI-GPU-A16-M6 | NVIDIA A16 GPU, 64GB (4x16GB) GDDR6, PCIe 4.0, VDI

Q: What is the primary use case for the HCI-GPU-A16-M6?
A: The HCI-GPU-A16-M6 is purpose-built for high-density virtual desktop infrastructure (VDI) deployments. It excels at serving knowledge workers running Office 365, web browsers, video conferencing, and multimedia applications, supporting up to 64 concurrent users per board. It can also provision entry-level virtual workstations for CAD and design applications when paired with NVIDIA RTX vWS software.

Q: How does the quad-GPU design benefit VDI deployments?
A: The unique quad-GPU architecture provides four independent 16GB GPUs on a single dual-slot card, enabling flexible user profile provisioning. IT administrators can mix user types on one board—allocating one GPU for power users running virtual workstations while splitting the other three across standard virtual desktop sessions. This delivers superior user density and TCO compared to single-GPU alternatives.

Q: Which Cisco servers support the HCI-GPU-A16-M6?
A: The HCI-GPU-A16-M6 is validated for Cisco UCS C240 M6, C240 M5, C245 M6, and Cisco HyperFlex systems. It requires PCIe 4.0 x16 slots and passive cooling support from the host server. The GPU must be procured from Cisco as it requires a unique SBIOS ID for CIMC and UCSM management.

Q: What software is required to enable VDI functionality?
A: The A16 GPU requires NVIDIA virtual GPU (vGPU) software to enable virtualization features. For knowledge worker VDI, pair it with NVIDIA Virtual PC (vPC) or Virtual Applications (vApps) software. For virtual workstations with professional graphics acceleration, use NVIDIA RTX Virtual Workstation (vWS) software. NVIDIA AI Enterprise software is also supported for compute workloads.

Q: How does the A16 compare to other NVIDIA data center GPUs?
A: The A16 is optimized specifically for VDI user density rather than raw compute performance. While GPUs like the A40 or L40S deliver more horsepower for workstation-class graphics and 3D rendering, the A16 maximizes seats per server for general productivity workloads. It delivers double the user density versus the previous-generation M10 GPU while providing up to 70% better performance than CPU-only VDI.