Reliability of Google’s Tensor Processing Units for Convolutional Neural Networks

This abstract presents the result of extensive reliability evaluation of Google’s Coral Tensor Processing Unit (TPU), which is one of the latest low power accelerators for CNNs. We report experimental data equivalent to more than 30 million years of natural irradiation and analyze the behavior of TPUs executing atomic operations (standard or depthwise convolutions) with increasing input sizes as well as eight CNN designs typical of embedded applications, including transfer learning and reduced data-set configurations. We found that, despite the high error rate, most neutrons-induced errors only slightly modify the convolution output and do not change the CNNs detection or classification. By reporting details about the fault model and error rate, we provide valuable information on how to evaluate and improve the reliability of CNNs executed on a TPU.