AI networking and full-stack optimization for AMD clusters

AMD and DriveNets are helping cluster builders deploy high-performance AMD-based AI environments faster and with lower risk. We are working closely together to provide a complete AI infrastructure solution that includes AMD Instinct GPU systems and DriveNets networking domains, spanning scale-up, scale-out, scale-across and frontend & storage networking. The joint solution provides a fully integrated software stack that is optimized for performance and functionality, delivering a competitive alternative to a full NVIDIA AI solution.

Download Reference Architecture

System-Level Optimization of AI Infrastructure with AMD Instinct GPUs

Validated end-to-end AI infrastructure delivery: Delivered a fully integrated and automated AMD & DriveNets based cluster stack with a repeatable deployment model and built-in validation to reduce bring-up risk and effort.
RCCL and system-level optimization: Joint AMD & DriveNets collaboration on collective communication and network optimization, spanning RCCL development and end-to-end performance validation and tuning.
Large-scale validation: Demonstrated the solution at scale under real workloads, achieving measurable performance gains vs. Spectrum-X, InfiniBand, and reference AMD clusters.
Future-safe collaboration: Joint R&D collaboration, including Helios rack-scale, liquid-cooled switch, future model optimization using automated, adaptive, and agentic techniques to continuously improve AI infrastructure performance.

Instinct Fabric Reference Architecture

AMD and DriveNets released a validated reference architecture document for clusters built with AMD Instinct MI355X GPUs, AMD Pollara NICs, and DriveNets scale-out and frontend solution.

The reference architecture document provides a comprehensive end-to-end blueprint for building a high-performance, scalable AI GPU cluster, and a repeatable deployment model that reduces integration and configuration risk.