Skip to content

Latest commit

 

History

History
15 lines (10 loc) · 587 Bytes

README.md

File metadata and controls

15 lines (10 loc) · 587 Bytes

AI Metrics

Performance metrics for AI/ML RoCEv2 network traffic, for example, large scale CUDA compute tasks using NVIDIA Collective Communication Library (NCCL) operations for inter-GPU communications: AllReduce, Broadcast, Reduce, AllGather, and ReduceScatter.

AI Metrics

To install

  1. Download sFlow-RT
  2. Run command: sflow-rt/get-app.sh sflow-rt topology
  3. Run command: sflow-rt/get-app.sh sflow-rt ai-metrics
  4. Restart sFlow-RT

For more information, visit: https://sFlow-RT.com