Title :
Maximum-information storage system: Concept, implementation and application
Author_Institution :
Electr. & Comput. Eng. Dept., Carnegie Mellon Univ., Pittsburgh, PA, USA
Abstract :
The aggressive technology scaling has made it increasingly difficult to design high-performance, high-density SRAM circuits. In this paper, we propose a new SRAM design methodology that is referred to as maximum-information storage system (MISS). Unlike most traditional SRAM circuits that are designed for maximum cell density, MISS aims to maximize the information density (i.e., the number of information bits per unit area). Towards this goal, an information model is derived to quantitatively measure the information bits stored in a given SRAM system. In addition, a convex optimization framework is developed to optimize SRAM cells to achieve maximum information storage. Our design example in a commercial 65nm CMOS process demonstrates that MISS achieves more than 3.5× area reduction over the traditional SRAM design, while storing the same amount of information. Furthermore, two real-life signal processing examples show that given the same area constraint, MISS can increase signal-to-noise ratio by more than 30 dB compared to the traditional SRAM system.
Keywords :
CMOS integrated circuits; SRAM chips; integrated circuit design; optimisation; CMOS process; SRAM design methodology; area reduction; cell density; convex optimization; information density; maximum-information storage system; signal-to-noise ratio; size 65 nm; Entropy; Noise; Quantization; Random access memory; Random variables; Silicon; Transistors;
Conference_Titel :
Computer-Aided Design (ICCAD), 2010 IEEE/ACM International Conference on
Conference_Location :
San Jose, CA
Print_ISBN :
978-1-4244-8193-4
DOI :
10.1109/ICCAD.2010.5653971