Performance Analysis and Optimization of Automotive GPUs

Journal of Parallel and Distributed Computing

Mazzocchetti

Benedicte

et al. 2021

Self Cite

Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) bring unprecedented performance requirements for automotive systems. Graphic Processing Unit (GPU) based platforms have been deployed with the aim of meeting these requirements, being NVIDIA Jetson TX2 and its high-performance successor, NVIDIA AGX Xavier, relevant representatives. However, to what extent high-performance GPU configurations are appropriate for ADAS and AD workloads remains as an open question.This paper analyzes this concern and provides valuable insights on this question by modeling two recent automotive NVIDIA GPU-based platforms, namely TX2 and AGX Xavier. In particular, our work assesses their microarchitectural parameters against relevant benchmarks, identifying GPU setups delivering increased performance within a similar cost envelope, or decreasing hardware costs while preserving original performance levels. Overall, our analysis identifies opportunities for the optimization of automotive GPUs to further increase system efficiency.

Section: Related Workmentioning

confidence: 99%

Performance Analysis and Optimization Opportunities for NVIDIA Automotive GPUs

Journal of Parallel and Distributed Computing

Mazzocchetti

Benedicte

et al. 2021

Self Cite

“…25 frames per second for camerabased input data). Since heavy parallel computations need to be performed at such high rates, hardware accelerators such as GPUs are needed [16]. This is the case for Apollo in general, and its Perception module in particular [5], whose most heavy computations are offloaded onto a GPU.…”

Section: Roadblocks For Resource Usage Testing On Ad Softwarementioning

confidence: 99%

En-Route

Alcon

Proceedings of the 35th Annual ACM Symposium on Applied Computing

Abella

et al. 2020

Self Cite

Software resource usage testing, including execution time bounds and memory, is a mandatory validation step during the integration of safety-related real-time systems. However, the inherent complexity of Autonomous Driving (AD) systems challenges current practice for resource usage testing. This paper exposes the difficulties to perform resource usage testing for AD frameworks by analyzing a complex and critical module of an AD framework, and provides some guidelines and practical evidence on how resource usage testing can be effectively performed, thus enabling end users to validate their safety-related real-time AD frameworks.

“…The latter are covered, under the stringent time constraints of the automotive domain, by executing AD software on powerful hardware [6,7,11,18], e.g. high-end GPUs despite the existing challenges [8,16,24]. Naturally, these hardware devices consume significant amounts of energy, which recent studies show can reduce the driving range (i.e.…”

Section: Introductionmentioning

confidence: 99%

IntPred

Proceedings of the 35th Annual ACM Symposium on Applied Computing

Fusi

Kosmidis

et al. 2020

Self Cite

Deep Neural-Network (DNN) based Object Detection is one of the most important and time-consuming stages of Autonomous Driving software in cars. In non-critical domains, the performance and energy requirements of object detection can be reduced at the cost of accuracy in the detected objects. This is not the case in a critical domain like automotive, for which a delicate balance between performance/energy overheads and accuracy of object detection must be found. We propose IntPred to achieve such a balance by leveraging on the fact that, with high frame rates, objects do not move significantly across frames. IntPred tailors object interpolation for the case of object detection in autonomous driving frameworks, in line with approaches devised for other domains, thus heavily reducing the performance requirements of full-fledged DNN-based object prediction. IntPred results in comparable accuracy to the original object detection, while saving more than 70% of the computations. The latter allows using lower-performance and cheaper platforms resulting in saving energy and reducing heat dissipation: for instance, in an NVIDIA Jetson TX2 platform, specific for autonomous driving systems, our technique increases the frame processing rate by 4.6x. IntPred also allows consolidating additional applications onto the same platform.