Vera Rubin access is drop-everything, the goal is proving it works for Google rather than benchmarks, and kernel pre-work starts now

Situation

NVIDIA is working to give CIQ a short window (likely about 5 days, one node) on Vera Rubin hardware, with no advance notice of when it opens. Peter set three things: when the window opens, whoever is involved drops all other work; the goal is getting Rocky/RLC working so CIQ can tell Google it works, not benchmarking; and Sultan starts pre-work now (patch identification, test branch) so the window is productive.

Reasoning

The window is scarce and unschedulable, so readiness is the only controllable lever. Setting the goal as function for Google keeps a short window from being spent on performance numbers. Pre-work cannot be wasted because RLC will support Vera Rubin eventually; leading the charge outranks waiting for Red Hat and 6.18 to carry some of the work. Demand from both Google GCE and GDC is being used as leverage with NVIDIA.

Additional Context

Nathan wanted 3 to 6 weeks; 5 days means a bloated kernel. Peter asked Scott Hara for 10 days and is seeking a confidence estimate on access within two months.

Observed Evidence

Direct quotes in two 1:1s; email to NVIDIA forwarding GCP demand; GDC pricing call noting NVIDIA asks for the smallest number of days.

Matching Patterns

Confidence Breakdown

Reasoning Depth Analysis