Temporal State Machine Architecture Using Time-Encoded Wavefronts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational architectures have stagnated due to limitations in transistor design and performance, and the adoption of race logic encoding schemes has been hindered by the lack of practical, systematized methods for building computer architectures that utilize temporal encoding.
Innovation Solution
The development of temporal memory and computational circuit components that utilize tunable delay components and race logic to encode data in the timing of electrical signals, allowing for the implementation of temporal state machines and computational units that perform operations in the time domain.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional digital clocked schemes are used for storing data and performing operations, then computational components can operate reliably with well-established transistor design, but progress in improving computational performance has stagnated due to transistor design limitations
Solution Approach 1:
The patent replaces the traditional digital clocked computational system with a temporal encoding system that uses arrival time of signals rather than clock cycles. This substitution fundamentally changes the computational paradigm from time-discrete clocked operations to continuous time-domain operations, enabling improved computational performance without being constrained by transistor switching speed limitations
Solution Approach 2:
The patent changes the encoding parameter from voltage level (0 and 1) to arrival time of signals. By encoding data in the temporal domain rather than voltage domain, the system can achieve higher computational density and performance while using existing transistor technology, effectively decoupling performance improvements from transistor design advances
2Productivity
If race logic encoding schemes are adopted to improve computational efficiency, then energy efficiency and computational speed can be enhanced, but practical implementation has been hindered by the lack of systematized methods for building computer architectures
Solution Approach 1:
The patent segments the computational architecture into distinct functional units including temporal memory, computational units, and interconnect structures. Each unit is designed to handle temporal encoding specifically, making the overall system easier to implement by breaking down the complex architecture into manageable, specialized components
Solution Approach 2:
The patent creates universal building blocks such as temporal memory units and computational units that can be configured for different applications. These multi-functional components can implement various operations (AND, OR, NOT, etc.) using the same temporal encoding mechanism, simplifying the manufacturing and deployment process
3Loss of energy
If multiple bits are encoded on a single wire using arrival time encoding, then energy consumption is reduced and computational simplicity is improved, but the complexity of decoding and synchronization increases
Solution Approach 1:
The patent employs self-synchronizing mechanisms where the temporal encoding scheme inherently provides synchronization information through the arrival times themselves. The system uses the relative timing of signal arrivals to automatically synchronize operations without requiring external clock signals or complex synchronization circuits, thereby reducing decoding complexity despite multi-bit encoding
Data Source
AI summary
Systems and methods are provided herein that implement and utilize temporal state machines and other temporal computational/memory circuits. For example, a temporal state machine is provided that comprises a temporal computational unit comprising at least one circuit to perform a time-domain operation, and a temporal memory unit. Both the temporal computational unit and temporal memory unit can operate using solely, or partially, time-encoded wavefronts. The computation unit accesses time-encoded wavefronts from the memory and can output and write time-encoded wavefronts back to the memory.