Answers are grounded in this paper and cite their sources.
Digital integrated circuit design faces growing challenges in evaluating multi-component system characteristics such as computational efficiency, energy performance, thermal dynamics, and reliability. While CMOS remains a foundational technology, the post-Moore era introduces heterogeneous integration strategies—including chiplets, 2.5D interposers, 3D stacking, RF modules, and photonic components—that demand new co-design and validation approaches. This survey examines state-of-the-art co-simulation and benchmarking techniques across multiple abstraction levels, addressing limitations in scalability, synchronization, and multi-physics modeling. We explore frameworks that integrate software-based models (SystemC, TLM), full-system simulators (gem5), and FPGA-based emulation with advanced packaging validation workflows. Additionally, we discuss AI-driven validation techniques, open-source versus proprietary toolchains, and workload-driven benchmarking for HPC and AI systems. By embracing cross-domain modeling and unconventional computing paradigms, this paper provides strategies to ensure performance, energy efficiency, and reliability in next-generation semiconductor designs.
No takes yet. Share an insight, caveat, or question.
Bel et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: