Products
AWS adds enhanced infrastructure observability to Amazon SageMaker HyperPod
AWS announced enhanced observability in Amazon SageMaker HyperPod to help diagnose GPU training issues across EKS- and Slurm-orchestrated clusters. The capability combines CloudWatch DetectGroupOutliers-based GPU outlier detection with static diagnostic rules for hardware faults, EFA throughput drops, thermal throttling, and overheating. Default metrics are enabled automatically, while advanced DCGM/Neuron and EFA metrics are opt-in; CloudWatch is now the default metric store with Prometheus still supported. The feature is available in all AWS Regions where SageMaker HyperPod is available.
Amazon SageMaker HyperPodAWSGPU training infrastructure