NVIDIA Expands NVLink Fusion with NVHBM for Enhanced AI Infrastructure

NVIDIA announced the expansion of its NVLink Fusion technology with NVHBM, a new high-bandwidth memory technology designed to improve memory performance and efficiency for AI workloads. According to the announcement, NVHBM integrates NVIDIA’s custom memory controller into the HBM base die, delivering up to 30% greater memory bandwidth and 15% lower HBM power consumption compared to standard HBM4E. This integration also frees up to 25% more area on XPU compute die, allowing for more efficient use of silicon area.

Amazon’s Annapurna Labs will be the first to collaborate on NVHBM technology alongside NVLink Fusion. The collaboration aims to enhance performance and efficiency for AI workloads, with Annapurna Labs supporting NVLink Fusion with its next-generation Trainium chips starting with Trainium4. This builds on AWS’s previously announced support for NVLink Fusion.

NVLink Fusion enables partners to connect custom XPUs and CPUs to NVIDIA’s rack-scale platform, offering a proven technology stack for scale-up and scale-out networking, rack-scale systems, and software. This allows hyperscalers and AI-native companies to focus engineering resources on XPU innovation while using a reliable technology stack for deploying semi-custom AI infrastructure.

Source: nvidia-newsroom