Dr. Muhammad Shafique from Chair of Embedded Systems (KIT Germany) and VISpro Lab (SEECS) are collaborating on the area of Automatic Co-Processor Generation for Computer Vision (Co-Pro-Vision). Dr. Shafique shall be giving two talks on 17th & 18th of January 2013(1700 hrs @ SEECS Seminar Hall). An abstract of the two talks and an introduction to the presenter is provided below.
Title: Cross-Layer Techniques for Reliable Code Generation and Execution on Unreliable Hardware
Venue & Time: SEECS Seminar Hall @ 1700 hrs on 17th of January 2013
Abstract: In the deep nano-scale regime, reliability has emerged as one of the major design constraints for high-density integrated systems. Among others, key reliability-related issues are soft errors, temperature related issues, and aging effects like Negative Bias Temperature Instability, which may corrupt the correct execution of application software. Tremendous amount of research effort has been invested at device, circuit, and architecture levels to counter these problems. However, these techniques incur significant area and power cost. We believe that designing a highly reliable and robust system requires combating the reliability issues at all the system layers, i.e. ranging from device to all the way up to the application software. In particular, we propose various reliability-optimizing techniques at compilation, system software, and application levels that account for the hardware’s susceptibility to various types of errors. Our proposed techniques enable generation and execution of reliable application software code, which minimizes the error probability on potentially unreliable hardware. This talk provides an overview of important reliability issues and state-of-the-art techniques along with several compilations and system software techniques for reliable systems and resiliency modeling approaches currently under investigation in our lab.
Title: Developing Power-Efficient Embedded Multimedia Systems for 3D-Video Coding
Venue & Time: SEECS Seminar Hall @ 1700 hrs on 18th of January 2013
Abstract: Due to the recent advancements in display and camera technologies, new 3D-video services are emerging that will change the landscape of future entertainment, security, and communication industries. Prominent application scenarios are Free Viewpoint TV, Realistic-TV, 3D-personal video recording and playback, 3D-surveillance, in-car 3D-infotainment, etc. However, gigantic amount of data processing poses a grand challenge towards next-generation embedded multimedia computing. We believe that exploiting the extensive application specific knowledge and leveraging algorithm- architecture collaborative techniques is the key to achieve high energy efficiency. This talk will provide an overview of performance and power related challenges for designing emerging embedded multimedia systems for 3D-video coding. Various energy-efficient algorithm and architecture level techniques will be presented that enable low-power processing of 3D-video coding.
About the Presenter
Dr. Muhammad Shafique is a research group leader at the Chair for Embedded Systems (CES), Department of Computer Science, Karlsruhe Institute of Technology (KIT), Germany. He has 9+ years' research and development experience in low-power and high-performance embedded systems (covering both hardware & software perspectives) in leading industrial and research organizations. Dr. Shafique holds one US patent on the low cost operational control for video encoders, 6 Gold Medals, CODES+ISSS'11 Best Paper Award, AHS'11 Best Paper Award, DATE'08 Best Paper Award, ICCAD'10 Best Paper Nomination, seven HiPEAC (European Network of Excellence on High Performance and Embedded Architecture and Compilation) Paper Awards, and Best Master's Thesis Award. He has (co) -authored 1 book and 45+ publications in premier conferences and journals on various aspects of low- power, reliable, and adaptive embedded systems.