Chandra Gandu

This paper introduces Stochastic Activity Prediction (SAP) to reduce dynamic power consumption in tensor accelerator multipliers by predicting low switching activity based on Hamming weight of arriving operands.

This paper presents a methodology for transitioning from Known Good Die (KGD) to Known Good Reliable Die (KGRD) screening in chiplet-based artificial intelligence systems-on-chips (SoCs) as a constrained inference problem, and proposes a Bayesian probabilistic risk model, safety-gated decision architecture, uncertainty-aware disposition boundaries, and a closed-loop feedback mechanism.