Safety-critical real-time embedded systems in automotive and aviation industries increasingly demand high-performance computing platforms to execute compute and data intensive workloads (e.g., deep neural networks) in real-time. However, engineers and researchers developing and studying these safety-critical systems have had troubles to deal with modern high-performance computer architectures because of their unpredictable and extremely poor worst-case timing behaviors that are too complex to understand, analyze and control. 

In this talk, I will present my group's two recent proposals to address the challenges. In the first work, we make a case that a fundamental problem that prevents efficient and predictable real-time computing on multicore is the lack of a proper memory abstraction to express memory criticality, which cuts across various layers of the system. We thus present a new memory abstraction, which we call Deterministic Memory, and supporting OS and hardware architecture to enable predictable and efficient real-time computing. The second work focuses on the interference problem on shared memory based Integrated CPU-GPU architectures (e.g., NVIDIA Jetson TX2). Because both CPU and GPU shares the same main memory subsystem in integrated architectures, concurrent use of CPU and GPU by multiple tasks can severely affect each other’s execution timing due to memory contention. We propose a software framework, which we call BWLOCK++, to protect the performance of real-time GPU kernels from co-scheduled memory intensive CPU applications.

Heechul Yun is an assistant professor in the Department of Electrical Engineering and Computer Science at the University of Kansas. He received a Ph.D. degree in Computer Science from the University of Illinois at Urbana-Champaign in 2013. Prior to his Ph.D., he worked at Samsung Electronics and ETRI. His research focuses on OS and architecture support for safety-critical real-time embedded systems. He received the Best Paper Award at IEEE Intl. Conference on Real-Time and Embedded Technology and Applications Symposium (RTAS) in 2016. Also, his work was selected as the Editor’s Pick of the Year 2016 from IEEE Transactions on Computers.