Abstract

Speaker: Yousun Ko

Title: Hardware/Software Co-refinement for Next-Generation Computer Accelerators

Abstract: The speed of processors has been grown exponentially, doubling every two years for last decades. However, the maximum speed of high-end processors has remained at around 4.0 GHz for more than ten years now. To resurrect conventional computing power scaling and meet the ever-growing demand for faster and energy efficient processors, we propose to explore new techniques for developing combined HW/SW accelerator solutions for domain-specific application. Traditional HW/SW co-design methodologies typically limit the design spaces a priori to make the design spaces amenable to handle, which may foreclose promising solutions inadvertently. To offer open-ended design space exploration without loss of optimality and feasibility, we suggest incremental and iterative refinement of both hardware and software from a many-core processor template, Loki.

——— Speaker: Bihao Wang

Title: Mono-vision based moving object detection in complex traffic scenes

Abstract: Traffic scene understanding has been a popular topic for the past few years, especially in the field of advanced driving system and autonomous vehicles. Among these methods, vision based moving object detection from a moving vehicle is still one of the most challenging subjects, because of the complexity of motion models, changing illumination conditions and limited embedded processing capabilities. In this talk, I will present some existing approaches to encourage a broad discussion on this topic.

——— Speaker: Diana Andreea Popescu

Title: Title: A Look at Network Latency in Data Centres and How It Affects Application Performance

Abstract: Businesses and individuals run increasing numbers of applications in the cloud. The performance of an application running in the cloud depends on the data center conditions and upon the resources committed to an application. Small network delays may lead to a significant performance degradation, which affects both the user’s cost and the service provider’s resource usage, power consumption and data center efficiency. We show how to use the Precision Time Protocol (PTP) to infer network latency and packet loss in data centers, and we conduct a small scale study of network latency and packet loss measurements in data centers from different cloud providers, using PTPd, an open-source software implementation of PTP . We quantify the effect of network latency on several typical cloud applications, varying in complexity and use cases. Our results show that different applications are affected by fixed and variable latency to differing amounts.