Roofline: An Insightful Visual Performance Model for Floating-Point Programs and Multicore Architectures

Communications of the ACM, vol 52 no 4. The canonical arithmetic-intensity roofline model: attainable performance = min(peak FLOPS, memory bandwidth x operational intensity). Establishes that a kernel with AI below the machine's balance point is bandwidth-bound no matter how fast the multiply units are. Verified via Crossref API (DOI 10.1145/1498765.1498785, issued 2009-04). | SOURCE AUDIT 2026-10-03: the 'publisher' field held a paragraph of prose (authors + affiliations, 100 chars) instead of a publisher name; trimmed to 'ACM Communications of the ACM' with the authorship detail moved here.