Ancient Greek had a three way contrast between voiced, unvoiced, and aspirated stops. It seems to be assumed that the unvoiced stops were pulmonic, but how do we know this?
A fact that may or may not have any bearing is that when there would otherwise be a sequence of two aspirated stops in successive syllables, the first one lost its aspiration (Grassmann's Law). Another is that certain initial stop clusters are uncommon or non-existent. The relatively common ones are pt, kt, and their aspirated correlates. Bd is uncommon and mainly used in words for ugly and disgusting things, while gd is more or less non-existent.