High-Integrity GPU Designs for Critical Real-Time Automotive Systems

Abstract

Autonomous Driving (AD) imposes the use of highperformance hardware, such as GPUs, to perform object recognition and tracking in real-time. However, differently to the consumer electronics market, critical real-time AD functionalities require a high degree of resilience against faults, in line with the automotive ISO26262 functional safety standard requirements. ISO26262 imposes the use of some source of independent redundancy for the most critical functionalities so that a single fault cannot lead to a failure, being dual core lockstep (DCLS) with diversity the preferred choice for computing devices. Unfortunately, GPUs do not support diverse DCLS by construction, thus failing to meet ISO26262 requirements efficiently. In this paper we propose lightweight modifications to GPUs to enable diverse DCLS for critical real-time applications without diminishing their performance for non-critical applications. In particular, we show how enabling specific mechanisms for software-controlled kernel scheduling in the GPU, allows guaranteeing that redundant kernels can be executed in different resources so that a single fault cannot lead to a failure, as imposed by ISO26262. Our results on a GPU simulator and an NVIDIA GPU prove the viability of the approach and its effectiveness on high-performance GPU designs needed for AD systems.

GPU4S: Major Project Outcomes, Lessons Learnt and Way Forward

“…For this reason, other solutions leveraging the inherent parallelism of the GPU hardware and software can achieve similar degree of reliability. For example, hardware [34] or software [35] GPU reliability solutions proposed for the automotive sector can be reused in space as well. Software solutions [36] are particularly interesting, since they can be applied directly in existing systems.…”

Multi-core Devices for Safety-critical Systems

“…In addition to multi-core devices, heterogeneous computing platforms including GPUs and application specific architectures (e.g., tensor cores) can potentially be used for the development of mixed-criticality systems. For example, GPUs provide high performance computing platforms for application domains such autonomous driving solutions [193,194] (e.g., NVIDIA Xavier ASILC)). Mittal et.…”

“…Mittal et. al [194] provide a survey of GPU based heterogeneous computing techniques and Alcaide et al [193] an analysis and techniques for the usage of GPUs in high-integrity autonomous driving solutions with respect to ISO 26262 standard. Besides, the development of artificial intelligence based autonomous systems is leading to open research problems such as the distribution of learning and inference activities over heterogeneous computing platforms.…”

GPU4S: Embedded GPUs in space - Latest project updates

“…For the GPU IP solutions, since their hardware design can be modified, we propose a hardware solution, again from our previous work in the automotive domain [28]. This solution applies the same concept with [27], however in this case it is the hardware which guarantees the spatial and temporal diversity.…”