Wave GPU
Wave GPU is available now. See the product page for what it does and how it was measured.
What is Wave GPU?
Wave GPU is the GPU Operations Layer for Kubernetes. It starts with full-stack observability and rule-based waste detection — and evolves toward smart GPU right-sizing and GPU sharing through HAMi integration.
Available today
- See Everything — GPU inventory, utilization, memory, power, and per-workload breakdown via DCGM
- Track Every Dollar — Per-namespace and per-label-group cost aggregation with configurable per-model pricing
- Reclaim Waste — 5 detection rules for idle, underutilized, unassigned, and over-allocated GPU/memory
On the Roadmap
- Smart GPU Sizing — Right-size GPU allocations the way Wave already does for CPU and memory
- GPU Sharing via HAMi — Fractional GPU allocation and multi-tenant sharing
See it on your cluster
Want to see it on your own cluster? Request a demo.