=========================== Performance, 4096 to 110592 =========================== Neighbour lists and primitive rings against freud, OVITO, mdapy, pyscal3, vesin, IRA, PTM, cycless, and every other code that has a neighbour or ring row in ``benchmarks/*.csv``. IRA and PTM have no neighbour or ring row in those files. The molecule field is ``molecules``. The interval is ``q1_ms`` and ``q3_ms`` when the file stores them, and ``iqr_ms`` when that is the interval it stores. ``n`` is the sample count in the file. A row with no time, or a verdict of absent or DNF, is written DNF. The classifier, polyhedral template matching, and IRA rows are ``benchmarks/paper-numbers/sitting-large.csv``. ``ratio_vs_dseams`` is the ratio of medians. ``ratio_lo`` and ``ratio_hi`` are the interval stored in that file. ``source_commit`` is the commit column. The tables are rendered from those files in the benchmark records dataset (``docs/records/README.md`` names the revision). .. include:: performance-range-tables.rst