Description

The NVIDIA NVLink 2-Way Bridge for H200 NVL GPUs provides high-bandwidth, low-latency interconnect between two GPUs, enabling them to function as a unified memory and compute resource. Engineered for next-generation workloads in AI training, large language models (LLMs), high-performance computing (HPC), and data-intensive analytics, the NVLink bridge supports direct GPU-to-GPU communication, bypassing traditional PCIe bottlenecks. With support for multiple NVLink lanes and bandwidth up to hundreds of GB/s, this 2-way bridge ensures optimized data throughput and efficient memory sharing between the paired H200 GPUs. It’s essential for scaling complex models and accelerating parallel processing tasks in demanding enterprise or research environments. This bridge is purpose-built for NVIDIA H200 NVL series GPUs and is a critical component for unleashing full multi-GPU performance in compatible systems.