University AI Compute: academic infrastructure and access reference
Repository-documented university-compute inventory, data pipeline and dashboard relating documented GPU infrastructure to academic research resources and allocation/access regimes. A reference for infrastructure planning and research on access to AI compute, not an automated allocator or a validated performance ranking.
Workflow level n/a Source: Repository-documented Outcome: Not empirically evaluated here
Research applicability
Explicitly documents research-university compute inventories, allocation/control regimes and national academic AI-pilot context. Supports infrastructure-access research and planning; any local resource-selection use is an editorial, untested adaptation.
Getting started
Editorial suggestion, not executed: inspect the README count concepts and access-regime methodology, follow the dated institutional sources for a relevant system, and verify actual eligibility, availability and units before using records for a research plan. Do not equate a fleet total with an obtainable allocation.
Maintenance
Live documentation was retrieved; maintenance quality and future availability are not guaranteed.
License
Consult original licence files and third-party terms; not independently audited.
Workflow & human oversight
Infrastructure and access research reference; no AI-executed or autonomous workflow is claimed. Researchers verify source records and institutional access conditions.
Workflow level is not an evidence score; higher is not better evidence.
Approach
Repository-documented university-compute inventory, data pipeline and dashboard relating documented GPU infrastructure to academic research resources and allocation/access regimes. A reference for infrastructure planning and research on access to AI compute, not an automated allocator or a validated performance ranking.
Limitations
README-only evidence; underlying CSV records, institutional primary sources, dashboard calculations and validation code were not independently inspected or executed. Inventory is documented lower bounds, not complete capacity: missing/unresolved is not zero, overlapping aggregates must not be double-counted, and generation evidence is sparse. Price-equivalent is a price proxy, not performance equivalence. Correlations are exploratory associations, not causation or evidence of under-investment. GPU counts, node-hours and project counts are different units; cloud/planned systems and merely managed national labs are not owned-GPU totals. No live allocation availability, access entitlement or compute-performance guarantee. Source documents GPU information through October 2026, HERD FY2024 and mixed endowment source years; these are source-described dates, not independently verified currency.
Reproducibility
No software, dashboard calculation or research experiment was run. Original saved source body, retrieval date and SHA-256 retained; mutable main branch, no commit pin.
Adoption requirements
For reference use, read the documented dashboard and methodology; check dated institutional source records and current allocation conditions. Local rebuilding requires inspecting the documented Python pipeline and its data/dependencies first. Data rights, source licences and calculation accuracy need separate review.
Comments
No comments yet.
Add a comment
Comments are held for moderation before appearing. Critical or corrective comments are welcome — especially evidence that an entry overstates what has been verified.