Physicists propose chips that relax the constraints reversible logic removes
In “Solving the compute crisis with physics-based ASICs,” authors including Patrick Coles, Kerem Camsari, Peter McMahon and Logan Wright propose application-specific chips that act as exact realisations of physical processes instead of forcing physics into the conventional digital straitjacket.
They argue that the standard requirements of statelessness, unidirectionality, determinism and synchronisation are precisely what waste energy, and that deliberately relaxing them can boost efficiency and throughput for diffusion models, optimisation, neural networks and scientific simulation. The framing connects directly to reversible computing, since unidirectionality is exactly the constraint reversible logic removes to recover energy.
The paper reads as a manifesto from prominent unconventional-computing researchers, situating reversibility within a broader movement toward letting hardware compute the way nature already does.