新闻动态

本周将有多位来自著名大学的学者来中心做报告。

3月19日

Title: Power Constrained Performance Optimization in Computer Systems

Speaker: Kai Ma (The Ohio State University)

Abstract: Power consumption has imposed a first-order design constraint to the entire spectrum of computer systems, from the smallest hand-held devices to the largest data centers. Limitations on the power consumption due to the small battery capacity in mobile devices, or the limited cooling capacity in desktops, or the tight electricity expense budget in data centers, require innovations that allow computer systems with larger numbers of cores to dynamically adapt to time-varying needs of modern workloads within a limited power budget. In the first part of my talk, I will provide the background of power management for computer systems. In the second part of my talk, I will introduce FreqPar, a power management solution that controls the power consumption of a many-core processor under a fixed power budget, as well as to optimize the performance of the processor by dynamically adjusting the frequency of each core on a multi-core processor. In the third part of my talk, I will summarize my previous projects on power management for multi-core processor, GPU/CPU heterogeneous systems, cooling power and computational power co-optimization, and data center power optimization.

3月21日

Title: Operating System Support for Overlapping-ISA Heterogeneous Multi-core Architectures

Speaker: Tong Li (Duke University)

Abstract: A heterogeneous processor consists of cores that are asymmetric in performance, power and, sometimes, functionality. Such a design enables the processor to achieve a wider operating range of different performance and power characteristics. This talk discusses the design space of heterogeneous architectures and one particular architecture model, where cores have asymmetric performance and overlapping, but non-identical instruction sets. We will look at some of the hardware and software challenges and discuss our early experience in building an operating system to support this architecture.

3月22日

Title: High-Performance Communication Protocols for Asynchronous Duty-Cycling Wireless Networks

Speaker: Lei Tang (Rice University)

Abstract: Duty cycling is a technique for saving energy in resource-limited wireless networks such as sensor networks. With duty-cycling, each node periodically switches between active and sleeping state, for example being active for only 1 or 10 percent of the time. Duty cycling networks face many challenges such as maintaining high energy efficiency, efficient packet delivery under dynamic traffic conditions, and resilience to wireless interference and possibly even to wireless jamming attacks. In this talk, I will present a progression of two wireless medium access control (MAC) protocols that address these challenges.

First, I will present a single-channel energy-efficient MAC protocol, called the Predictive-Wakeup MAC (PW-MAC). The key idea behind PW-MAC is to allow each node to wake up at randomized, asynchronous times chosen by the node’s own random wakeup scheduler, while enabling senders to predict receiver wakeup times to save energy. Second, I will present a multichannel extension of PW-MAC, called the Efficient-Multichannel MAC (EM-MAC). EM-MAC enables each node to dynamically optimize the selection of wireless channels it utilizes based on the channel conditions it senses. By adapting to changing channel conditions and spreading wireless packet transmissions to randomized wakeup channels and wakeup times, EM-MAC substantially enhances wireless channel utilization and packet transmission efficiency while resisting wireless interference and even jamming on limited numbers of channels.

Implemented in TinyOS on a testbed of MICAz sensor nodes, EM-MAC significantly outperformed all other MAC protocols tested, while PW-MAC achieved the lowest energy consumption and delivery latency among all single-channel protocols. EM-MAC maintained the lowest sender and receiver duty cycles, the lowest packet delivery latency, and 100% packet delivery ratio across all experiments, including those with concurrent multihop traffic flows, and those with heavy ZigBee interference, constant ZigBee jamming, or Wi-Fi interference.

3月22日

Title: Towards Efficient Real-Time Multicore Computing Systems

Speaker: Cong Liu (The University of North Carolina at Chapel Hill)

Abstract: Current trends in multicore computing are towards building more powerful, intelligent, yet space- and power- efficient systems. A key requirement in correctly building such intelligent systems is to ensure real-time performance, i.e., “make the right move at the right time in a predictable manner”. Current research on real-time multicore computing has been limited to simple systems for which complex application runtime behaviors are ignored; this limits the practical applicability of such research. In practice, complex but realistic application runtime behaviors often exist, such as I/O operations, data communications, parallel execution segments, critical sections etc.. Such runtime behaviors are currently dealt with by over-provisioning systems, which is an economically wasteful practice. In this talk, I will present predictable real-time multicore computing system design, analysis, and implementation methods that can efficiently support common types of application runtime behaviors. I will show that the proposed methods are able to avoid over-provisioning systems and to reduce the number of needed hardware components to the extent possible while providing timing correctness guarantees.