Skip to content

M3 GPU pools and fractional sharing #56

Description

@Rikarin

docs/plan/01 § AI and Machine Learning · ⊂ Compute · M3.

GPU-attached VMs and node pools, via the NVIDIA GPU operator and HAMI for fractional sharing — which is what Cozystack 1.4 added.

⚠ The catalogue frames the whole AI section deliberately and it is worth keeping: selling "AI" without models is selling GPUs, and we should say GPUs. Thirty Azure AI products are out of scope on the grounds that they are a research organisation with a GPU fleet, not a cloud-platform capability. This row is the honest subset.

Depends on #28 (Compute) for the VM and node-pool shapes to attach to, and adds two charts/bundle/ components with the pinning and licence questions in #17 and #18.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    product-surfacePortal, CLI, SDK, provider catalogue

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions