| Document Date: 2013-12-16 15:05:08
Open Document File Size: 662,10 KBShare Result on Facebook
Company LG / HP Laboratories / TSMC / Synopsys / NVIDIA / / Country United States / / Currency USD / / / Event Reorganization / / Facility GP CMOS library / Krste Asanovi´c University of California / Is building / / IndustryTerm graph-based knowledge network search / data-parallel solutions / signal processors / energy bottleneck / energy breakdown / pattern-based data-parallel processors / instruction delivery energy / search accelerator / data-parallel processors / energy consumption gap / data-parallel processor / energy benefits / programmable solutions / search algorithm / 3.6× more energy / scalar processor / energy / search kernel / beam search approximation / energy consumption / mobile system-on-chip / graphics processors / across-word search kernels / energy efficiency / computing / fixed-function hardware / scalar control processor / search space / vector processor / conventional data-parallel processors / static energy / / Organization US government / University of California / Berkeley / ASIC / US Federal Reserve / / Person Michael Anderson / Morgan Kaufmann / Andrew Waterman / David Sheffield / Krste Asanovi / Kurt Keutzer / / Product 45nm GP CMOS / / ProgrammingLanguage Python / Verilog / C++ / / Technology pattern-based data-parallel processors / data-parallel processors / Speech Recognition / graphics processors / Hwacha vector processor / scalar control processor / vector processor / 2.2 Rocket-Hwacha Vector Processor / SRAM / scalar processor / speech recognition system / ASIC Chip / CMP / Viterbi search algorithm / mobile system-on-chip / ASIC / conventional data-parallel processors / 5-stage Rocket processor / Verilog / data-parallel processor / simulation / /
SocialTag |
Measuring the Gap between Programmable and Fixed-Function Accelerators: A Case Study on Speech Recognition Yunsup Lee, David Sheffield, Andrew Waterman, Michael Anderson, Kurt Keutzer, Krste Asanovi´c