Capo

The idea

Capo is the min-cut school engineered to its logical conclusion, built around the question in its founding paper's title: can recursive bisection alone produce routable placements? Its answers: use state-of-the-art multilevel Fiduccia-Mattheyses bisection at every level; propagate terminals rigorously; manage balance tolerances dynamically (loose early for freedom, tight late for legality); handle the end-case with exact or enumerative placers when regions shrink to a few cells; and treat whitespace as a managed resource, distributing it where congestion needs it rather than packing rows solid.

That last point made Capo famous for routability: placements that routers could actually finish, at a time when denser-looking placements from other tools drowned in overflow. Capo shipped as reliable open source, became the standard academic baseline for a decade, and its benchmarks and evaluation discipline (the same group drove standard wirelength contests) shaped how placers are compared.

The verdict history delivered

On pure wirelength, analytical placers eventually and decisively won: smooth optimization simply sees more than recursive cutting. But Capo's virtues, determinism, speed, congestion sanity, whitespace discipline, migrated into the winners. Modern analytical flows quietly contain min-cut ideas in their clustering and their density management.

Where you meet it today

In the literature as the strongest min-cut reference, in courses as the worked example of engineering a heuristic carefully, and in the design of every benchmark suite it helped standardize.