Overmind NSA: A Unified Neuro-Symbolic Computing Architecture with Approximate Nonlinear Activations and Preemptive Memory Bypass

TL;DR

Overmind is a unified neuro-symbolic computing architecture that uses approximate nonlinear functions and memory bypass techniques to improve efficiency and performance in AI applications.

Contribution

It introduces a novel hardware architecture with Padé approximations and preemptive memory bypass, optimizing neuro-symbolic AI deployment with reduced resource usage.

Findings

Achieves 8.1 TOPS/W energy efficiency.

Reaches 410 GOPS throughput for mixed workloads.

Maintains minimal accuracy loss with adaptive nonlinear approximation.

Abstract

Neuro-symbolic AI is gaining traction in domains such as large language models, scientific discovery, and autonomous systems due to its ability to combine perception with structured reasoning. However, its deployment is often constrained by high memory demands, diverse computation patterns, and complex hardware requirements. Existing hardware platforms struggle with large on-chip memory overheads, frequent pipeline stalls, limited I/O bandwidth, and inefficient handling of nonlinear operations. To address these key computational bottlenecks, we propose Overmind, a unified neuro-symbolic architecture with cross-layer optimizations. Overmind tackles these core bottlenecks through Pad\'e approximations for universal nonlinear functions, preemptive memory bypass that eliminates costly on-chip caches, and a complete software stack that optimizes model deployment. By reconfiguring the Pad\'e orders for approximating nonlinear functions, we also demonstrate adaptive accuracy-performance scaling. Overmind achieves an energy efficiency of 8.1 TOPS/W and a throughput of 410 GOPS for mixed neuro-symbolic workloads with minimal model accuracy loss. Compared to existing solutions, Overmind improves performance and efficiency with significantly fewer hardware resources.