Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper addresses the gap in semiconductor test methodology for chiplet-based AI systems-on-chip by formalizing the transition from Known Good Die (KGD) to Known Good Reliable Die (KGRD) screening as a constrained inference problem. The authors introduce a Bayesian probabilistic risk model that links pre-assembly telemetry to post-assembly failure likelihood, alongside a safety-gated decision architecture that ensures a provable failure probability guarantee. Through Monte Carlo simulations on 4,000 synthetic dies, the study validates the theoretical framework and demonstrates that the safety guarantee is maintained across varying gate thresholds.
Transitioning from KGD to KGRD screening could redefine reliability standards in chiplet-based AI systems, ensuring functional correctness and longevity.
The rapid growth of chiplet-based artificial intelligence systems-on-chip (SoCs) has exposed a fundamental gap in semiconductor test methodology. Existing Known Good Die (KGD) screening guarantees pre-assembly functional correctness, yet it offers no probabilistic assurance of post-assembly reliability lifetime. To address this limitation, the present work formalizes the transition from KGD to Known Good Reliable Die (KGRD) screening as a constrained inference problem over incomplete pre-assembly observability. Building upon this formulation, four interlocking contributions are presented: (i) a Bayesian probabilistic risk model that maps pre-assembly telemetry to post-assembly failure likelihood with a quantified observability bias bound; (ii) a safety-gated decision architecture that provides a provable post-assembly failure probability guarantee; (iii) uncertainty-aware disposition boundaries derived from Bayes-optimal decision theory; and (iv) a constrained closed-loop feedback mechanism that delivers consistent model improvement without violating reliability constraints. A Monte Carlo simulation study on N = 4,000 synthetic dies verifies all four theoretical properties and confirms that the safety guarantee holds uniformly across the full range of tested gate threshold.