Scheda bridge NVLink 2-Way passiva per GPU Hopper H200 NVL — NVIDIA

NVLink Bridge Board 2-Way Passive for Hopper H200 NVL GPUs is a passive interconnect board designed to create a direct high-bandwidth link between two compatible NVIDIA Hopper H200 NVL graphics accelerator cards. The board enables low-latency peer-to-peer communication and improved data transfer rates between GPUs in supported server and workstation platforms, helping to optimize multi-GPU compute and AI workloads without requiring active cooling or additional power connections.

Advantages

The passive NVLink bridge provides a straightforward physical connection that reduces latency and increases effective bandwidth between two Hopper H200 NVL GPUs. By enabling direct GPU-to-GPU transfers, it simplifies data movement for parallel processing and scales multi-GPU configurations in systems that support NVIDIA NVLink. The passive design lowers complexity and power requirements compared with active interconnect solutions, and the board’s form factor is intended to match the mechanical and electrical interfaces of compatible accelerator cards.

Suitable for

• High-performance compute servers requiring multi-GPU communication
• AI training and inference systems using Hopper H200 NVL accelerators
• Workstations configured with two compatible NVL GPUs for parallel workloads
• Environments where a passive, low-power NVLink interconnect is preferred

Key specifications

• Type: Passive 2-way NVLink bridge board
• Compatibility: Designed for use with NVIDIA Hopper H200 NVL GPU cards that support NVLink
• Function: Direct GPU-to-GPU high-bandwidth interconnect for peer-to-peer communication
• Cooling: Passive design (no integrated active cooling)
• Form factor: Board designed to align with the mechanical connector positions of compatible dual-card configurations

How to use

Install the NVLink bridge by powering down the host system, opening the server or workstation chassis, and seating each end of the bridge into the NVLink connector on the top edge of each Hopper H200 NVL GPU. Ensure the GPUs are properly secured in their PCIe slots and that the bridge is fully engaged with the connectors on both cards. After installation, close the chassis and power on the system. Verify that the platform firmware and GPU drivers recognize NVLink and that system settings permit NVLink operation between the two cards.

Recommendations

Use this passive NVLink bridge only with systems and GPU cards that explicitly support NVLink on Hopper H200 NVL accelerators. Confirm mechanical clearance and airflow requirements inside the chassis, and ensure firmware and driver versions are up to date to enable NVLink features. Avoid installing or removing the bridge while the system is powered; handle the board by its edges and follow electrostatic discharge (ESD) precautions.