VDSM技术中容错数据总线的时延和功耗

A. Sathish, M. Chennakesavulu, M. Latha, K. Kishore
{"title":"VDSM技术中容错数据总线的时延和功耗","authors":"A. Sathish, M. Chennakesavulu, M. Latha, K. Kishore","doi":"10.1109/INTERACT.2010.5706173","DOIUrl":null,"url":null,"abstract":"In Very Deep-submicron (VDSM) systems, the scaling of ULSI ICs has increased the sensitivity of CMOS technology to cause various noise mechanismssuch as power supply noise, crosstalk noise, leakage noise, etc. In VDSM technology distance between the data bus lines is reduced, so coupling capacitance is dominating factor. Unfortunately, in VDSM systems, the coupling capacitance is of magnitude several times larger than the loading capacitance. The coupling capacitance causes logical malfunction, delay faults, and power consumption on long on-chip data buses. An important effect of the coupling capacitance is Cross talk. Crosstalk is mainly dependent on several factors: drive strength, wire length/spacing, edge rate and propagation duration. The crosstalk noise produces from the coupling capacitance. Such faults may affect data on data bus. To avoid this condition and to guarantee signal integrity on the on-chip communication, a fault tolerant bus can be adopted. This could be achieved by implementing error-correcting codes (ECCs), providing on-line correction and do not require data retransmission. The 4,8,16, and 32-bit data bus is implemented in 180nm, 120nm, and 65nm technologies using Bsim4 model. For reliable transmission of the data ECC techniques is placed on the data bus. We employed a Hamming code and Dual rail as ECC for 4,8,16 and 32-bit fault tolerant data bus. This is implemented in 180nm, 120nm and 65nm technology. The simulation results show that Average power varies from 0.737mw to 0.176mw, and Maximum delay varies from 0.143nsec to 0.077nsec, for hamming 4 bit ECC, Average power varies from 2.135mw to 0.365mw and Maximum delay varies from 0.385nsec to 0.192nsec for hamming 8 bit ECC, Average power varies from 2.288mw to 0.377mw and Maximum delay varies from 0.721nsec to 0.353nsec for hamming 16 bit ECC, Average power varies from 3.064mw to 0.437mw and Maximum delay varies from 1.562nsec to 0.796nsec for hamming 32 bit ECC. The simulation results show that Average power varies from 0.206mw to 0.0459mw, and Maximum delay varies from 0.241nsec to 0.133nsec, for dual rail 4 bit ECC, Average power varies from 0.417mw to 0.0768mw and Maximum delay varies from 0.479nsec to 0.262nsec for dual rail 8 bit ECC, Average power varies from 0.726mw to 0.156mw and Maximum delay varies from 1.026nsec to 0.554nsec for dual rail 16 bit ECC, Average power varies from 0.926mw to 0.108mw and Maximum delay varies from 2.129nsec to 1.145nsec for dual rail 32 bit ECC respectively.","PeriodicalId":201931,"journal":{"name":"INTERACT-2010","volume":"37 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Delay and power consumption of fault tolerant data busses in VDSM technology\",\"authors\":\"A. Sathish, M. Chennakesavulu, M. Latha, K. Kishore\",\"doi\":\"10.1109/INTERACT.2010.5706173\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In Very Deep-submicron (VDSM) systems, the scaling of ULSI ICs has increased the sensitivity of CMOS technology to cause various noise mechanismssuch as power supply noise, crosstalk noise, leakage noise, etc. In VDSM technology distance between the data bus lines is reduced, so coupling capacitance is dominating factor. Unfortunately, in VDSM systems, the coupling capacitance is of magnitude several times larger than the loading capacitance. The coupling capacitance causes logical malfunction, delay faults, and power consumption on long on-chip data buses. An important effect of the coupling capacitance is Cross talk. Crosstalk is mainly dependent on several factors: drive strength, wire length/spacing, edge rate and propagation duration. The crosstalk noise produces from the coupling capacitance. Such faults may affect data on data bus. To avoid this condition and to guarantee signal integrity on the on-chip communication, a fault tolerant bus can be adopted. This could be achieved by implementing error-correcting codes (ECCs), providing on-line correction and do not require data retransmission. The 4,8,16, and 32-bit data bus is implemented in 180nm, 120nm, and 65nm technologies using Bsim4 model. For reliable transmission of the data ECC techniques is placed on the data bus. We employed a Hamming code and Dual rail as ECC for 4,8,16 and 32-bit fault tolerant data bus. This is implemented in 180nm, 120nm and 65nm technology. The simulation results show that Average power varies from 0.737mw to 0.176mw, and Maximum delay varies from 0.143nsec to 0.077nsec, for hamming 4 bit ECC, Average power varies from 2.135mw to 0.365mw and Maximum delay varies from 0.385nsec to 0.192nsec for hamming 8 bit ECC, Average power varies from 2.288mw to 0.377mw and Maximum delay varies from 0.721nsec to 0.353nsec for hamming 16 bit ECC, Average power varies from 3.064mw to 0.437mw and Maximum delay varies from 1.562nsec to 0.796nsec for hamming 32 bit ECC. The simulation results show that Average power varies from 0.206mw to 0.0459mw, and Maximum delay varies from 0.241nsec to 0.133nsec, for dual rail 4 bit ECC, Average power varies from 0.417mw to 0.0768mw and Maximum delay varies from 0.479nsec to 0.262nsec for dual rail 8 bit ECC, Average power varies from 0.726mw to 0.156mw and Maximum delay varies from 1.026nsec to 0.554nsec for dual rail 16 bit ECC, Average power varies from 0.926mw to 0.108mw and Maximum delay varies from 2.129nsec to 1.145nsec for dual rail 32 bit ECC respectively.\",\"PeriodicalId\":201931,\"journal\":{\"name\":\"INTERACT-2010\",\"volume\":\"37 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2010-12-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"INTERACT-2010\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/INTERACT.2010.5706173\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"INTERACT-2010","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/INTERACT.2010.5706173","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

在甚深亚微米(VDSM)系统中,ULSI集成电路的缩放增加了CMOS技术对各种噪声机制(如电源噪声、串扰噪声、泄漏噪声等)的灵敏度。在VDSM技术中,数据总线之间的距离减小,因此耦合电容是主要因素。不幸的是,在VDSM系统中,耦合电容比负载电容大几倍。耦合电容会导致长片上数据总线的逻辑故障、延迟故障和功耗。耦合电容的一个重要影响是串扰。串扰主要取决于几个因素:驱动强度、导线长度/间距、边缘速率和传播持续时间。串扰噪声是由耦合电容产生的。这些故障可能会影响数据总线上的数据。为了避免这种情况并保证片上通信的信号完整性,可以采用容错总线。这可以通过实施纠错码(ECCs)来实现,提供在线纠错并且不需要数据重传。4、8、16和32位数据总线采用Bsim4模型,采用180nm、120nm和65nm技术实现。为了保证数据的可靠传输,ECC技术被置于数据总线上。我们采用汉明码和双轨作为ECC的4、8、16和32位容错数据总线。这是实现在180nm, 120nm和65nm的技术。仿真结果表明,对于4位ECC,平均功率为0.737mw ~ 0.176mw,最大时延为0.143 ~ 0.077nsec;对于8位ECC,平均功率为2.135mw ~ 0.365mw,最大时延为0.385 ~ 0.192nsec;对于16位ECC,平均功率为2.288mw ~ 0.377mw,最大时延为0.721 ~ 0.353nsec。平均功率从3.064mw到0.437mw,最大延迟从1.562nsec到0.796nsec不等。仿真结果表明:双轨4位ECC的平均功率为0.206 ~ 0.0459mw,最大时延为0.241 ~ 0.133nsec;双轨8位ECC的平均功率为0.417 ~ 0.0768mw,最大时延为0.479 ~ 0.262nsec;双轨16位ECC的平均功率为0.726 ~ 0.156mw,最大时延为1.026 ~ 0.554nsec。双轨32位ECC的平均功率范围为0.926mw ~ 0.108mw,最大延迟范围为2.129nsec ~ 1.145nsec。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Delay and power consumption of fault tolerant data busses in VDSM technology
In Very Deep-submicron (VDSM) systems, the scaling of ULSI ICs has increased the sensitivity of CMOS technology to cause various noise mechanismssuch as power supply noise, crosstalk noise, leakage noise, etc. In VDSM technology distance between the data bus lines is reduced, so coupling capacitance is dominating factor. Unfortunately, in VDSM systems, the coupling capacitance is of magnitude several times larger than the loading capacitance. The coupling capacitance causes logical malfunction, delay faults, and power consumption on long on-chip data buses. An important effect of the coupling capacitance is Cross talk. Crosstalk is mainly dependent on several factors: drive strength, wire length/spacing, edge rate and propagation duration. The crosstalk noise produces from the coupling capacitance. Such faults may affect data on data bus. To avoid this condition and to guarantee signal integrity on the on-chip communication, a fault tolerant bus can be adopted. This could be achieved by implementing error-correcting codes (ECCs), providing on-line correction and do not require data retransmission. The 4,8,16, and 32-bit data bus is implemented in 180nm, 120nm, and 65nm technologies using Bsim4 model. For reliable transmission of the data ECC techniques is placed on the data bus. We employed a Hamming code and Dual rail as ECC for 4,8,16 and 32-bit fault tolerant data bus. This is implemented in 180nm, 120nm and 65nm technology. The simulation results show that Average power varies from 0.737mw to 0.176mw, and Maximum delay varies from 0.143nsec to 0.077nsec, for hamming 4 bit ECC, Average power varies from 2.135mw to 0.365mw and Maximum delay varies from 0.385nsec to 0.192nsec for hamming 8 bit ECC, Average power varies from 2.288mw to 0.377mw and Maximum delay varies from 0.721nsec to 0.353nsec for hamming 16 bit ECC, Average power varies from 3.064mw to 0.437mw and Maximum delay varies from 1.562nsec to 0.796nsec for hamming 32 bit ECC. The simulation results show that Average power varies from 0.206mw to 0.0459mw, and Maximum delay varies from 0.241nsec to 0.133nsec, for dual rail 4 bit ECC, Average power varies from 0.417mw to 0.0768mw and Maximum delay varies from 0.479nsec to 0.262nsec for dual rail 8 bit ECC, Average power varies from 0.726mw to 0.156mw and Maximum delay varies from 1.026nsec to 0.554nsec for dual rail 16 bit ECC, Average power varies from 0.926mw to 0.108mw and Maximum delay varies from 2.129nsec to 1.145nsec for dual rail 32 bit ECC respectively.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信