Delay-driven Layer Assignment for Sub-16nm Technology Nodes

碩士 === 國立清華大學 === 資訊工程學系 === 104 === As technology advances, the interconnect delay gradually dominates the circuit performance but the emerging issues makes timing optimization become more complicated. Since 32nm nodes, different tiers of layers have different default wire widths such that the laye...

Full description

Bibliographic Details
Main Authors: Han, Szu-Yuan, 韓儩源
Other Authors: Wang, Ting-Chi
Format: Others
Published: 2016
Online Access:http://ndltd.ncl.edu.tw/handle/33664868921084002217
id ndltd-TW-104NTHU5392030
record_format oai_dc
spelling ndltd-TW-104NTHU53920302017-08-27T04:29:59Z http://ndltd.ncl.edu.tw/handle/33664868921084002217 Delay-driven Layer Assignment for Sub-16nm Technology Nodes 在次16奈米半導體製程下以延遲為取向之佈局層配置 Han, Szu-Yuan 韓儩源 碩士 國立清華大學 資訊工程學系 104 As technology advances, the interconnect delay gradually dominates the circuit performance but the emerging issues makes timing optimization become more complicated. Since 32nm nodes, different tiers of layers have different default wire widths such that the layer assignment of routing wires greatly impacts interconnect delay. Moreover, below 16nm nodes, because the wire coupling impacts delay more significantly and wire width cannot be arbitrarily sized to reduce wire capacitance, the timing optimization problem is harder to resolve. This thesis addresses a delay-driven layer assignment problem with consideration of via delay and coupling effect in the global routing stage. A look-up table method is proposed to estimate the coupling capacitance modeled by a probabilistic approach. In order to consider routability, a negotiation-based framework is presented to strike a good balance between delay, congestion, and via count. Finally, the proposed algorithm is able to use parallel wires and non-default-rule (NDR) wires to pursue better delay reduction instead of using wire sizing. The effectiveness of our layer assignment algorithm is well supported by extensive experimental results. Wang, Ting-Chi 王廷基 2016 學位論文 ; thesis 34
collection NDLTD
format Others
sources NDLTD
description 碩士 === 國立清華大學 === 資訊工程學系 === 104 === As technology advances, the interconnect delay gradually dominates the circuit performance but the emerging issues makes timing optimization become more complicated. Since 32nm nodes, different tiers of layers have different default wire widths such that the layer assignment of routing wires greatly impacts interconnect delay. Moreover, below 16nm nodes, because the wire coupling impacts delay more significantly and wire width cannot be arbitrarily sized to reduce wire capacitance, the timing optimization problem is harder to resolve. This thesis addresses a delay-driven layer assignment problem with consideration of via delay and coupling effect in the global routing stage. A look-up table method is proposed to estimate the coupling capacitance modeled by a probabilistic approach. In order to consider routability, a negotiation-based framework is presented to strike a good balance between delay, congestion, and via count. Finally, the proposed algorithm is able to use parallel wires and non-default-rule (NDR) wires to pursue better delay reduction instead of using wire sizing. The effectiveness of our layer assignment algorithm is well supported by extensive experimental results.
author2 Wang, Ting-Chi
author_facet Wang, Ting-Chi
Han, Szu-Yuan
韓儩源
author Han, Szu-Yuan
韓儩源
spellingShingle Han, Szu-Yuan
韓儩源
Delay-driven Layer Assignment for Sub-16nm Technology Nodes
author_sort Han, Szu-Yuan
title Delay-driven Layer Assignment for Sub-16nm Technology Nodes
title_short Delay-driven Layer Assignment for Sub-16nm Technology Nodes
title_full Delay-driven Layer Assignment for Sub-16nm Technology Nodes
title_fullStr Delay-driven Layer Assignment for Sub-16nm Technology Nodes
title_full_unstemmed Delay-driven Layer Assignment for Sub-16nm Technology Nodes
title_sort delay-driven layer assignment for sub-16nm technology nodes
publishDate 2016
url http://ndltd.ncl.edu.tw/handle/33664868921084002217
work_keys_str_mv AT hanszuyuan delaydrivenlayerassignmentforsub16nmtechnologynodes
AT hánsìyuán delaydrivenlayerassignmentforsub16nmtechnologynodes
AT hanszuyuan zàicì16nàimǐbàndǎotǐzhìchéngxiàyǐyánchíwèiqǔxiàngzhībùjúcéngpèizhì
AT hánsìyuán zàicì16nàimǐbàndǎotǐzhìchéngxiàyǐyánchíwèiqǔxiàngzhībùjúcéngpèizhì
_version_ 1718519354834812928