Events that invalidate the baseline

Rerun after a PHP minor or major upgrade, a server migration, a processor class change, a material operating-system update or a change to runtime extensions involved in a study.

A rerun is also useful after hosting-plan changes that alter CPU limits or storage characteristics, even when the visible software versions stay the same.

Preserve old results

Do not silently replace history when a comparison matters. Archive the old JSON and generation timestamp, note the environment change and publish a correction if earlier prose no longer matches the new data.

Continue reading

Full benchmark methodology Browse measured studies