inference latency

2 articles

Kubernetes 1.37 Memory QoS: GPU Inference Tradeoffs

A 2026 operator guide to Kubernetes 1.37 Memory QoS for GPU inference: inactive defaults, host DRAM protection, page-cache risks, reproducible canaries, SLO evidence, and rollback checks.

October 30, 2026• 22 min read
kubernetes memory qoskubernetes 1.37gpu inference

GPU Power Cap Optimization for Inference: Cost and SLOs

A 2026 protocol for choosing inference GPU power caps using same-run energy, latency-qualified throughput, replica rounding and monthly cost scenarios, with NVIDIA, NERSC and Dell evidence.

October 4, 2026• 27 min read
gpu power cap optimization for inferencegpu power cappinginference latency