Thermodynamic Reversibility in O(1) Stateless Computing: Micro-watt vQPU Architecture via Adiabatic Charge-Recovery Logic (ACRL)
Min Ho Jung · Zenodo (CERN European Organization for Nuclear Research) · 2026
Modern hyper-scale AI models and spatial edge applications face severe power dissipation bottlenecks and the von Neumann memory wall. According to Landauer's principle, traditional non-reversible computing systems inevitably generate thermodynamic heat during state transitions. Here, we report the hardware-level experimental realization of an ultra-low-power virtual Quantum Processing Unit (vQPU) based on Adiabatic Charge-Recovery Logic (ACRL). By executing logically reversible coordinate mappings (ΔS_logic = 0), this architecture intrinsically bypasses the thermodynamic limits of energy dissipation (ΔE_dissip = 0). Instead of dumping residual charge to ground, the ACRL circuit utilizes a multi-phase power clock to recover the charge back to the power supply. Experimental validation demonstrates that the vQPU operates at a micro-watt scale power consumption (23.9 µW), while materializing a 4.2 GB 4D spatiotemporal dataset utilizing an invariant 11.2 MB heap memory (O(1) space complexity) and zero network bandwidth. This design has been patented under the Korean Patent Application No. 10-2026-0123049, titled "Adiabatic Charge-Recovery Logic (ACRL) Based Logically Reversible virtual Quantum Processing Unit (vQPU) and Ultra-Low-Power Reversible Computing System Using the Same".