Breaking through power and ILP walls across multiple scales of computing
撰稿：
摄影：
发布时间：2013年08月30日
时间：2013年9月2日（周一）上午10:00-11:30
地点：4层报告厅
摘要
This talk ties together several projects at Harvard that strive to push compute performance and efficiency across a wide range of scales. For Tiny embedded systems, like the "brain" of the Harvard RoboBee, stress the limits of size and weight in addition to energy efficiency. We propose to use a fully-integrated on-chip voltage regulators, voltage-tracking clock generators, and a sea of hardware accelerators. In mobile applications, demands on small form factor and longer battery life motivated us to rethink the power delivery subsystem for multicore processors and to develop 'voltage stacking‘. For large-scale computing systems, In order to break through the ILP wall, we propose an automatic parallelization framework called HELIX. As a general theme, these projects explore cross-layer solutions that leverage synergistic opportunities between circuit, architecture, and software layers.
主讲人简介 David Brooks is Gordon McKay Professor of Computer Science in the School of Engineering and Applied Sciences at Harvard University. Prof. Brooks received his BS in Electrical Engineering at the University of Southern California and MA and PhD degrees in Electrical Engineering at Princeton University. His research interests include technology-aware computer design, with an emphasis on power-efficient computer architectures and system software for high-performance and embedded systems.
附件下载：
Breaking through power and ILP walls across multiple scales of computing
地点：4层报告厅
摘要
This talk ties together several projects at Harvard that strive to push compute performance and efficiency across a wide range of scales. For Tiny embedded systems, like the "brain" of the Harvard RoboBee, stress the limits of size and weight in addition to energy efficiency. We propose to use a fully-integrated on-chip voltage regulators, voltage-tracking clock generators, and a sea of hardware accelerators. In mobile applications, demands on small form factor and longer battery life motivated us to rethink the power delivery subsystem for multicore processors and to develop 'voltage stacking‘. For large-scale computing systems, In order to break through the ILP wall, we propose an automatic parallelization framework called HELIX. As a general theme, these projects explore cross-layer solutions that leverage synergistic opportunities between circuit, architecture, and software layers.
主讲人简介
David Brooks is Gordon McKay Professor of Computer Science in the School of Engineering and Applied Sciences at Harvard University. Prof. Brooks received his BS in Electrical Engineering at the University of Southern California and MA and PhD degrees in Electrical Engineering at Princeton University. His research interests include technology-aware computer design, with an emphasis on power-efficient computer architectures and system software for high-performance and embedded systems.