Clockwork.io
Executive perspectives from Clockwork.io across NextGenInfra's event and research showcases.
1 video interview · 2026
Solving GPU Cluster Inefficiency: From 30% to Peak AI Performance
- GPU training clusters typically run at only 30-50% efficiency
- Root causes are inter-GPU communication, availability issues, and straggler tail latency
- Nanosecond-precision clock synchronization mitigates these on InfiniBand or RoCE networks
Full summary
Suresh Vasudevan, CEO of Clockwork.io, explains how GPU clusters for AI training operate at only 30-50% efficiency due to poor inter-GPU communication, availability issues, and tail latency from slow straggler GPUs. He explains how nanosecond-precision clock synchronization can be leveraged to address availability and latency issues across Infiniband or RoCE networks.