The Value Proposition of Campus High Performance Computing Facilities to Institutional Productivity: A Production Function Model

Abstract

Abstract High-performance computing (HPC) is used by higher education institutions, industry, and research labs around the world for modeling, simulation, and data analysis to enable scientific discovery. However, the operation of a high-performance computing center comes with significant operating and capital costs. While some previous attempts have been made to explore the return on investment (ROI) of investment in HPC, the academic cyberinfrastructure community lacks a quantitative model with which an institution of higher education may analyze its own investments in HPC facilities and measure if it is meeting its own stated goals. To create such a model, this study seeks to answer the question if investment in campus HPC facilities leads to measurable impact on institutional outputs. To do so, an analysis of a set of value metrics collected from Purdue University are performed to measure the ROI of the institution’s investment in HPC facilities, and create a model based upon an application of a production function that will measure the HPC facility investment’s impact on the financial, academic, and reputational outputs of the institution. The resulting models demonstrate a strong relationship between institutional outputs and the investments of labor and capital supporting the campus cyberinfrastructure. By itself, investment in either computing resources or staff expertise pays dividends in terms of multiple institutional outputs. While investment in human resources accounts for the largest relative importance in models for all prospective outputs, a coordinated strategy that invests in computing resources and staff to support them together will yield the highest returns.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.