MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202611079828 A) filed by Chandigarh University on June 29, 2026, for Adaptive Multi-Datapath Alu With Confidence-Guided Verification And Learning.
Inventor includes Garima Thakur.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present invention relates to a self-optimizing Arithmetic Logic Unit (ALU) configured to dynamically select an appropriate execution datapath for arithmetic and logical operations based on real-time operating conditions. The ALU includes a feature extraction module for deriving operand characteristics and telemetry information, including voltage, temperature, and aging parameters. A hierarchical machine-learning prediction engine determines a datapath selection signal and confidence value for choosing among fast, approximate, and safe execution paths. A residue verification module selectively validates computation results when the confidence value falls below a predefined threshold, thereby reducing verification overhead. A rollback buffer enables rapid recovery and re-execution through a safe datapath upon detection of an error. Further, a reservoir sampling engine facilitates incremental on-chip adaptation using operational data. The disclosed architecture enhances computational reliability, performance, energy efficiency, and resilience to workload variations and device aging.
Disclaimer: Curated by HT Syndication.
