Graphics Processing Units (GPUs), originally developed for computer graphics, are now commonly used to accelerate parallel applications. Given that GPUs are designed to be as efficient as possible, evaluating their performance is crucial. This problem has been tackled in the last years by researchers that started to propose solutions such as analytical models and digital simulators, which are, however, often complex to use and/or to adapt to the needs of the user. Thanks to its high flexibility, model-based analysis is widely used to evaluate systems’ properties, including performance. Researchers started working on developing GPU models that can represent both their architecture and the software in execution, but they often use strong assumptions that undermine their usability. In this work we develop a Stochastic Activity Network model to evaluate the performance of CUDA applications running on NVIDIA GPUs. The model takes as input a representation of the program’s instruction, parsed from the CUDA SASS assembly file, and a list of parameters to offer configurability to the user. We tune our model to match the architecture of two different NVIDIA GPUs and simulate the execution of a CUDA program. We then compare the results with those obtained from the execution of the program over the real GPUs.

General Purpose GPUs (GPGPUs) are highly susceptible to both transient and permanent faults. This is a serious concern for their safe and reliable usage in many domains, from autonomous driving to High Performance Computing. The research and industrial community responded fiercely to this issue, by analyzing failures impact and devising failure mitigation strategies. This led to the definition of several failure modes and mitigation approaches. Unfortunately, these are often based on different foundations, and it is not easy to position them in a consistent view. This work elaborates a GPGPU failures model, identifying relations between the GPGPU failure modes and components, and then it analyzes mitigations proposed in the literature. By proposing a unified view on failures and mitigations, the resulting model i) positions each research on the subject, ii) easily identifies the current gaps, and iii) sets the basis for further research on GPGPU failures.