Emerging deep learning workloads urgently need fast general matrix multiplication (GEMM). To meet such demand, one of the critical features of machine-learning-specific accelerators such as NVIDIA Tensor Cores, AMD Matrix Cores, and Google TPUs is the support of mixed-precision enabled GEMM. For DNN models, lower-precision FP data formats and computation offer acceptable correctness but significant performance, area, and memory footprint improvement. While promising, the mixed-precision computation on error resilience remains unexplored. To this end, we develop a fault injection framework that systematically injects fault into the mixed-precision computation results. We investigate how the faults affect the accuracy of machine learning applications. Based on the error resilience characteristics, we offer lightweight error detection and correction solutions that significantly improve the overall model accuracy if the models experience hardware faults. The solutions can be efficiently integrated into the accelerator's pipelines.
@article{arxiv.2311.05782,
title = {MPGemmFI: A Fault Injection Technique for Mixed Precision GEMM in ML Applications},
author = {Bo Fang and Xinyi Li and Harvey Dam and Cheng Tan and Siva Kumar Sastry Hari and Timothy Tsai and Ignacio Laguna and Dingwen Tao and Ganesh Gopalakrishnan and Prashant Nair and Kevin Barker and Ang Li},
journal= {arXiv preprint arXiv:2311.05782},
year = {2023}
}