Work

A Physically Anchored LoE Driver for GPU Architectures

Year: 2025

Type: article

Abstract: We present a self-contained Logic of Existence driver for modern GPUs that requiresno external assets (figures or data files) to compile or run its documentation. The driveranchors each CUDA thread (Sub-Neuron, SN) and each Streaming Multiprocessor (SM) to afixed logical block of the Logic of Existence field, thereby turning the GPU topology into apersistent neuromorphic substrate. We formalize the field equations, describe a persistentblock-per-SM kernel, outline multi-tone driving and annealing, detail boundary conditionsand multi-GPU synchronization, and provide profiling & telemetry practices. This paper isa drop-in documentation module for Overleaf: it uses only standard LaTeX packages and noexternal graphics.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.17336311

Author: Sawa, Krzysztof

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7090140958

FWCI:

Cites:

Cited by:

Related to: