Thermodynamic computer (Normal Computing SPU)

Realizes: probabilistic sampling / linear algebra via thermal equilibration

Analog physics-based computers using thermodynamic principles for computation. Normal Computing's Stochastic Processing Unit (SPU) uses RLC circuits as unit cells with all-to-all coupling via switched capacitances, natively simulating Langevin/Ornstein-Uhlenbeck dynamics for probabilistic reasoning, generative design, and scientific computing. The SPU line is realized as analog probabilistic ASICs with memristive elements and thermodynamic noise shaping, targeting AI inference rather than general-purpose compute.

Examples

🔗

Normal Computing Stochastic Processing Unit

Thermodynamic computer using RLC circuits on PCB with all-to-all coupling, natively implementing Langevin dynamics for sampling and linear algebra

SAMPLE MULTIPLY ADD seconds medium mJ

🔗

Normal Computing Carnot Architecture

Silicon-based thermodynamic computing architecture enabling unprecedented scaling for thermodynamic computation, targeting generative design and probabilistic reasoning

SAMPLE MULTIPLY ADD seconds large μJ

🔗

Normal Computing CN101 prototype board

Normal Computing's CN101 prototype board demonstrates analog stochastic inference circuits targeting diffusion models, Langevin samplers, and probabilistic linear solvers.

Stochastic inference workloads for diffusion models, stochastic samplers, and probabilistic matrix operations delivered via analog probabilistic dataflow circuits. high chip pJ-scale