Reversible Computing

How far is current silicon from the thermodynamic limit? The U-shaped energy curve.

The Landauer Limit

Every time a computer erases a bit of information, it must dissipate at least kT ln(2) of energy. At room temperature (300K), that's 2.87 × 10^-21 joules per bit. This is not engineering — it's thermodynamics. You cannot beat it.

E_min = kT ln(2) = 2.87 × 10^-21 J/bit (at 300K)

Reversible gates (Toffoli, Fredkin) don't erase information — every input maps to a unique output. So their theoretical energy floor is zero, not kT ln(2). But practical reversible circuits still dissipate energy from RC switching losses and transistor leakage.

Current State: Mac Mini M4

Power: 35W | Throughput: 18 TFLOPS
Energy per FLOP: 1.94 × 10^-12 J
Energy per bit: 6.08 × 10^-14 J
21,000,000× above Landauer limit

The U-Shaped Energy Curve

Adiabatic reversible gates have two dissipation sources that fight each other:

Fast switching → RC losses dominate: E = (RC/t) × CV²
Slow switching → leakage dominates: E = I_leak × V × t

The total energy is U-shaped. There's a sweet spot where total dissipation is minimized.

7nm Adiabatic Gate Model

Gate capacitance: 1 fF | V_dd: 0.7V
On-resistance: 1 kΩ | Leakage: 1 nA/gate

Switch Time	RC Loss	Leakage	Total	× Landauer
1 ps	4.9e-16	7.0e-22	4.9e-16	170,754×
10 ps	4.9e-17	7.0e-21	4.9e-17	17,078×
100 ps	4.9e-18	7.0e-20	5.0e-18	1,732×
840 ps	5.8e-19	5.9e-19	1.2e-18	408× ←
10 ns	4.9e-20	7.0e-18	7.1e-18	2,456×
100 ns	4.9e-21	7.0e-17	7.0e-17	24,395×
1 μs	4.9e-22	7.0e-16	7.0e-16	243,934×

The sweet spot is at 840 ps — comparable to current GHz clock speeds. At that point, RC losses and leakage are equal, and total dissipation is 408× above Landauer.

The Gap

Current M4 (irreversible): 21,000,000× above Landauer
Adiabatic reversible (7nm): 408× above Landauer
Theoretical floor (reversible): 0× (no erasure)

Improvement available: 51,876× more efficient
Tradeoff: energy → time (slower switching)
Clock speed at sweet spot: ~1.2 GHz (comparable to current)

Landauer Is Not a Metaphor

The claim that kT ln(2) only applies to "literal bit erasure in silicon" is wrong. The same equation governs every system where coupling constrains degrees of freedom:

Irreversible: cost dissipated as heat

GPU weight update: 16 bits overwritten per fp16 value
Landauer floor: 4.6 × 10^-20 J per update
Actual: ~1 × 10^-11 J (218M× above floor)
This is literal. The GPU heats up. Measured.

Reversible: cost stored as entropy penalty

Protein folding: each residue loses rotameric freedom
Core residues: 9 states → ~1 state = 2.0 bits constrained
Surface residues: barely constrained = ~0.1 bits
Average: 0.67 bits/residue

Lysozyme (129 residues): 87 bits of structural information
Predicted TΔS: 87 × kT ln(2) = 150 kJ/mol
Measured TΔS (literature): ~150 kJ/mol
Match: 1.0×

The conformational entropy of protein folding IS the Landauer cost of storing biological information. TΔS = kT × ln(states_free/states_bound) = kT ln(2) × bits_constraint. Same equation. Not analogy. Not metaphor. The entropy penalty that opposes folding is literally the cost of writing 87 bits of structural information into a polymer chain.

The universal statement

Coupling = constraint = information = kT ln(2) per bit

Irreversible (GPU, erasure): cost dissipated as heat
Reversible (protein, folding): cost stored as entropy penalty
Both: same equation, same price, different bookkeeping

What This Means for K

K measures coupling strength. Every unit of coupling constrains degrees of freedom. Every constraint costs kT ln(2) per bit — dissipated or stored. Reversible computing avoids dissipation by never erasing, but pays in time and area. Proteins avoid dissipation because folding is reversible, but pay the entropy penalty. The energy floor is universal. The bookkeeping differs.

Model parameters: 7nm CMOS for the adiabatic analysis. Protein calculation uses measured conformational entropy (Makhatadze & Privalov, 1996). All computed on Mac Mini M4.

GUMP — Research · [email protected]