🚗 Smart Manufacturing
Leading NEV Enterprise
Core challenge: Insufficient computing for autonomous driving
Scale: 256 GPU Cluster
Core challenge
Insufficient computing for autonomous driving
Solution
Built H100 GPU training cluster with high-bandwidth network and distributed storage
Result
Training cycle reduced 60%
Project details
Huizhi Jiachuan built one of the leading autonomous driving AI training platforms in China for this top NEV enterprise. The project uses 256 NVIDIA H100 SXM5 GPUs with InfiniBand 400G high-speed interconnects and distributed parallel file systems.
Project gallery

