#gpu
Tagged “gpu”
4 articlesOpenCost 1.121.0 pins a dollar figure to every model and token on Kubernetes
OpenCost 1.121.0, announced on the CNCF blog on August 5, adds per-model and per-token cost attribution for AI inference workloads. The release aims at platform teams that have been reading GPU bills and token counters in separate windows with no way to connect them.
Aug 6, 2026 · Maya OkonkwoRunners & infrastructureThe 60% idle GPU that turned out to be a network policy
A CNCF write-up walks through a distributed Kubeflow training job whose GPUs sat around 60% idle while every pod reported healthy. The failure lived one layer down, in a Cilium policy enforcing zone boundaries that gradient sync had to cross.
Jul 23, 2026 · Maya OkonkwoRunners & infrastructureGPUs keep falling off the PCIe bus, and standard node health does not notice
AWS engineers write up what running Kubernetes at Amazon EKS scale taught them about GPU node failure, and why the default node health model has nothing to say when the GPU just vanishes from the bus.
Jul 20, 2026 · Maya OkonkwoRunners & infrastructureDynamic Resource Allocation hits GA in Kubernetes 1.35, and GPU CI jobs finally get a real API
DRA reached general availability in Kubernetes v1.35, and the NVIDIA DRA driver has dropped its Beta label. For teams running GPU workloads through Kubernetes-backed pipelines, it replaces a decade of nodeSelector gymnastics with a first-class request primitive.
Jul 2, 2026 · Priya Nair