infiniband-rdma

name	infiniband-rdma
description	Configure SR-IOV InfiniBand and verify RDMA bandwidth

Infiniband Rdma

When to use this skill

• HPC or distributed ML training

Key concepts

• SR-IOV exposes IB adapter to the VM; OFED + IB drivers required (built into HPC images).
• Verify with ibv_devinfo, ib_send_bw, nccl-tests.
• MPI bootstraps over IB; ensure correct UCX/HCOLL build.
• All VMs must be in the same VMSS or PPG to share IB fabric.

Common pitfalls

• Mixing IB and non-IB SKUs in one cluster — fabric reachability breaks.

References

• Microsoft Learn: search for "Infiniband Rdma Azure VM" on https://learn.microsoft.com

---

Validation-first: verify every Azure fact against the Microsoft Learn MCP server (cite the Learn URL); Learn overrides built-in knowledge. If no Learn MCP server is configured, flag the answer ⚠️ unverified and mark specs as indicative. Analysis only — verify against Microsoft documentation before applying.