OpenAI Introduces MRC (Multipath Reliable Connection): A New Open Networking Protocol for Large-Scale AI Supercomputer Training Clusters

MarkTechPost / 5/7/2026

📰 NewsDeveloper Stack & InfrastructureIndustry & Market Moves

共有:

Key Points

OpenAI has introduced MRC (Multipath Reliable Connection), an open networking protocol designed to boost GPU networking performance and resilience in large-scale AI training clusters.
MRC improves reliability by sending packets across hundreds of paths simultaneously and rapidly recovering from network failures on the order of microseconds.
The protocol targets very large training systems—enabling supercomputers with 100,000+ GPUs to be constructed using only two tiers of Ethernet switches.
MRC was developed in partnership with major hardware and platform companies including AMD, Broadcom, Intel, Microsoft, and NVIDIA.

MRC (Multipath Reliable Connection) is a new open networking protocol developed by OpenAI in partnership with AMD, Broadcom, Intel, Microsoft, and NVIDIA that improves GPU networking performance and resilience in large-scale AI training clusters by spreading packets across hundreds of paths simultaneously, recovering from network failures in microseconds, and enabling supercomputers with over 100,000 GPUs to be built using only two tiers of Ethernet switches.

The post OpenAI Introduces MRC (Multipath Reliable Connection): A New Open Networking Protocol for Large-Scale AI Supercomputer Training Clusters appeared first on MarkTechPost.