首页> 外文学位 >New methodologies for low-power high-performance digital VLSI design.
【24h】

New methodologies for low-power high-performance digital VLSI design.

机译:低功耗高性能数字VLSI设计的新方法。

获取原文
获取原文并翻译 | 示例

摘要

Historically, VLSI designers have focused on increasing the speed and reducing the area of digital systems. However, the evolution of portable systems and advanced Deep Sub-Micron (DSM) fabrication technologies, has brought power dissipation as another critical design factor. Low-power design reduces cooling cost and increases reliability especially for high-density systems. Moreover, it reduces the weight and size of portable devices. Yet, high-performance is still the main criterion for most digital systems, which may not be sacrificed to achieve lower power dissipation. This thesis presents new low-power high-performance digital VLSI design methodologies for process, circuit, and algorithm level design.; On the process level, future challenges in device scaling such as short channel effects, subthreshold leakage currents, and hot carrier effects are discussed. The influence of technology scaling on the performance, power, and area of different CMOS logic styles is then analyzed and simulated. This study covers five logic families; namely, CMOS, CPL, Domino, DCVS and CML. The scalability of each logic style and its potential in future technology generations are explored.; On the circuit level, a new logic family for low-power high-performance applications is presented. This logic family combines the speed, low supply voltage, and noise immunity of CML circuits with the low standby current and design simplicity of dynamic circuits. The new logic style reduces the power by 7% and the delay by 73% compared to conventional CMOS logic. A 16-bit CLA adder is designed, simulated, fabricated, and tested using 0.61μ m CMOS technology. Test results have confirmed the functionality of the new logic family at various supply voltages.; Also, a new Domino logic style, called High-Speed Domino (HS-Domino), has been developed. HS-Domino resolves the trade-off between noise margin and speed associated with the conventional Domino logic. Simulation results show that HS-Domino circuits are superior to conventional Domino ones in terms of power, speed, and tolerance to the leakage currents in DSM technologies.; This study also presents new Multiple Threshold CMOS (MTCMOS) scheme for dynamic circuits. This scheme is applied to Domino and DDCVS logic styles. The new implementations reduce the leakage power by orders of magnitude keeping the noise margin intact, and maintain the high performance and low dynamic power of low VT circuits. Unlike other MTCMOS Domino logic implementations, the new scheme does not require additional hardware.; At the algorithm level, a new algorithm for high radix division is presented. The algorithm uses a look-up table to estimate the quotient digit at each iteration. The look-up table is optimized to reduce power dissipation and delay of the divider. Simulation results show that the new algorithm reduces the power dissipation by 22% and 12% for radix 8 and radix 16 division, respectively, compared to other division algorithms. The algorithm also increases the speed by a factor of 13% and 10%, for radix 8 and radix 16 division, respectively.
机译:从历史上看,VLSI设计人员一直致力于提高速度和减少数字系统的面积。但是,便携式系统和先进的深亚微米(DSM)制造技术的发展带来了功耗,这是另一个至关重要的设计因素。低功耗设计降低了冷却成本并提高了可靠性,特别是对于高密度系统。此外,它减轻了便携式设备的重量和尺寸。然而,高性能仍然是大多数数字系统的主要标准,为达到更低的功耗而不能牺牲高性能。本文提出了用于过程,电路和算法级设计的新型低功耗高性能数字VLSI设计方法。在工艺方面,讨论了器件缩放方面的未来挑战,例如短通道效应,亚阈值泄漏电流和热载流子效应。然后,分析和模拟了技术扩展对不同CMOS逻辑样式的性能,功耗和面积的影响。这项研究涵盖了五个逻辑家族。即CMOS,CPL,Domino,DCVS和CML。探索了每种逻辑样式的可扩展性及其在未来技术中的潜力。在电路级,提出了一种适用于低功耗高性能应用的新逻辑系列。该逻辑系列将CML电路的速度,低电源电压和抗扰性与低待机电流和动态电路的设计简单性结合在一起。与传统的CMOS逻辑相比,新的逻辑样式将功耗降低了7%,将延迟降低了73%。使用0.61μ m CMOS技术设计,仿真,制造和测试了16位CLA加法器。测试结果证实了新逻辑系列在各种电源电压下的功能。此外,还开发了一种新的Domino逻辑样式,称为高速Domino(HS-Domino)。 HS-Domino解决了噪声裕度和与常规Domino逻辑相关联的速度之间的折衷问题。仿真结果表明,在DSM技术中,HS-Domino电路在功率,速度和对泄漏电流的耐受性方面均优于传统的Domino。这项研究还提出了用于动态电路的新的多阈值CMOS(MTCMOS)方案。此方案适用于Domino和DDCVS逻辑样式。新的实现方式将泄漏功率降低了几个数量级,从而保持了噪声容限不变,并保持了低VT电路的高性能和低动态功耗。与其他MTCMOS Domino逻辑实现不同,新方案不需要额外的硬件。在算法级别,提出了一种新的高基数除法算法。该算法使用查找表来估计每次迭代的商位数。优化查找表以减少功耗和分频器的延迟。仿真结果表明,与其他除法算法相比,该算法将基数8和基数16的功耗分别降低22%和12%。对于基数8和基数16的划分,该算法还将速度分别提高了13%和10%。

著录项

  • 作者

    Allam, Mohamed Walred.;

  • 作者单位

    University of Waterloo (Canada).;

  • 授予单位 University of Waterloo (Canada).;
  • 学科 Engineering Electronics and Electrical.
  • 学位 Ph.D.
  • 年度 2000
  • 页码 207 p.
  • 总页数 207
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 无线电电子学、电信技术;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号