Hardware
How NVIDIA NVLink 6 Delivers Multi-Layer Resiliency for AI Factories
NVIDIA NVLink 6 offers a natively lossless fabric for large‑scale AI factories by integrating lightweight Forward Error Correction, physical‑layer retries, UPHY recovery and credit‑based flow control
NVIDIA NVLink 6 offers a natively lossless fabric for large‑scale AI factories by integrating lightweight Forward Error Correction, physical‑layer retries, UPHY recovery and credit‑based flow control to eliminate packet loss with near‑zero latency penalty. Its NMX controller isolates faults locally via a “contain and drain” state, while Shadow Engine Recovery keeps pre‑warmed replicas that restore inference in 7 s instead of hours for cold restarts; NCCL supports multi‑node checkpoints through cuda‑checkpoint, and NVLink Fusion extends this resiliency stack to third‑party XPUs. These features together enable continuous, high‑throughput training across thousands of GPUs with minimal downtime.
Related
- NVIDIA NVLink: The Scale-Up Network for AI Factories
- NVIDIA NVLink Fusion Brings NVHBM to Next-Generation AI Infrastructure
Source: NVIDIA Developer | 2026-09-15