Fundamentals of CPU/GPU Power and Performance Modeling

This book presents a comprehensive, academically rigorous, and industry-grounded framework for modeling power and performance in modern CPUs and GPUs. It bridges architectural theory, workload-driven performance analysis, and realistic chip-level power modeling, explicitly incorporating variations and constraints that are often treated as secondary or external—such as thermals, yield, reliability, and power delivery. By integrating these dimensions into a single framework, the book connects early architectural decisions to post-silicon behavior and manufacturing out-comes, and demonstrates how accurate power–performance (PnP) modeling can directly inform high-impact decisions including binning strategy, product stacking, and wafer utilization.