August 23, 2024
Members of the UNC Real-Time Systems Group were awarded two Outstanding Paper awards and a Best Presentation award at the 36th Euromicro Conference on Real-Time Systems (ECRTS) in Lille, France.
ECRTS is the largest conference in Europe dedicated to the broad research area of real-time systems and one of the top three international conferences on the topic.
Outstanding Paper Awards
“Open Problem Resolved: The ‘Two’ in Existing Multiprocessor PI-Blocking Bounds is Fundamental”
Shareef Ahmed and James H. Anderson
The goal of a real-time locking protocol is to reduce any priority-inversion blocking (pi-blocking) a task may incur while waiting to access a shared resource. This paper closes a pi-blocking-related problem that has been open for 14 years. This problem pertains to upper and lower bounds on pi-blocking under global fixed-priority (G-FP) and global earliest-deadline-first (G-EDF) scheduling.
“Autonomy Today: Many Delay-Prone Black Boxes”
Sizhe Liu, Rohan Wagle, James H. Anderson, Ming Yang (WeRide Corp., UNC alumnus), Chi Zhang (WeRide Corp.), and Yunhua Li (WeRide Corp.)
This paper, a collaboration between UNC Computer Science and autonomous vehicle company WeRide, highlights the danger of using software and hardware components developed for non-safety-critical applications in autonomous vehicles. One such component used with machine-learning algorithms, CUDA, was shown to cause synchronization delays and system failure during testing performed by WeRide.
Best Presentation Award
“Predictable GPU Sharing in Component-Based Real-Time Systems”
Syed W. Ali, Zelin Tong, Joseph Goh, and James H. Anderson
This paper presents a real-time locking protocol whose design was motivated by the goal of enabling safe GPU sharing in time-sliced component-based systems. The protocol enables a GPU to be shared concurrently across, and utilized within, isolated components with predictable execution times. It relies on a novel resizing technique where GPU work is dimensioned on-the-fly to run on partitions of an NVIDIA GPU. The paper was presented at the conference by Ali.