CloudBolt Brings GPU Optimization and Bill-Accurate GPU Cost Allocation to StormForge

CloudBolt Software has made GPU optimization and bill-accurate GPU cost allocation available in its Kubernetes optimization product, StormForge. This release allows platform and FinOps teams to allocate GPU costs to specific namespaces, workloads, and labels, using data directly from the cloud provider’s invoice rather than relying on estimate prices.

The new feature provides per-workload visibility into GPU usage and inefficiencies, offering recommendations to reduce waste and insights into node usage patterns for better node type selection. Additionally, StormForge now includes node pool cost and idle reporting, with cost attributes from Reserved Instances and Savings Plans woven into the billing data.

Users can delve into detailed cost, idle capacity, Spot usage, and commitment coverage for each cluster's node pool and instance type. Reliability-focused recommendations now address CPU contention, ensuring latency-sensitive workloads receive necessary CPU resources, and respecting namespace LimitRange policies to maintain compliance.

These updates aim to help teams gain a clearer understanding of their GPU spend, facilitating more informed discussions and implementing automated solutions they can confidently rely on.