SHADOWFI: An Open-source Framework For Fault Evaluation of Complex IC Designs Using Hyperscale Computing

Cutting-edge integrated circuits (ICs) are increasingly vulnerable to hardware faults, which can jeopardize the overall reliability of the system. Hence, it is crucial to evaluate the impact of faults to identify hardware vulnerabilities that later designers can use to devise fault mitigation solutions during the circuit design stages. Fault injection (FI) through in-circuit emulation tackles the inherent complexity of fault evaluations by adopting FPGA emulation strategies. Although various works report frameworks using this FI approach, most of them are technology-dependent and hardly scalable when dealing with the increasing complexity of modern IC architectures. In addition, none of these frameworks is disclosed, which limits the adoption of fault emulation strategies due to the inherent complexity and the required time to set up a functional FI environment. This work introduces SHADOWFI, a generic, open-source, and netlist-based fault emulation framework that leverages the computational capabilities of hyperscale infrastructures for fault characterization and reliability estimation of complex IC designs. SHADOWFI offers two different functional workflows: i) simulation, which enables the parallelization of FI tasks on high-performance computing systems, and ii) emulation, which leverages the flexibility of FPGA cluster implementations. The framework automates saboteur insertion, FI campaign execution, and report generation, requiring minimal user configuration. Each SHADOWFI workflow was evaluated on a set of IC design benchmarks, demonstrating practical usability and significant fault injections speedup. SHADOWFI is publicly available at https://github.com/divadnauj-GB/SHADOWFI.git.