IGS: The Novel Fast IC Power Ground Network Optimization Flow Based on Improved Gauss-Seidel Method
IGS: The Novel Fast IC Power Ground Network Optimization Flow Based on Improved Gauss-Seidel Method
复制标题
IGS：基于改进高斯-赛德尔法的新型快速IC电源地网优化流程
DOI：
10.25046/aj020391
复制
发表时间：
2017-06
期刊：
影响因子：
--
通讯作者：
Xiaoxiao Wang
中科院分区：
文献类型：
--
作者：
Qinghao Ye;Jiansong Gong;Xiaoxiao Wang
随着硅技术的进一步扩展，开关活动与GHz工作频率一起通过产生大的IR降噪声极大地影响功率完整性。过大的IR压降会导致功能故障，如定时故障、异常复位和SRAM翻转。需要优化PGN以减少IR压降。传统的EDA优化程序重复了版图生成、参数提取和仿真等步骤，大大增加了集成电路设计的成本。提出了一种基于改进的高斯-赛德尔法（IGS）的PGN快速优化流程，该流程根据PGN的初始电压分布和三维寄生电阻分布计算精确的IR压降分布。该优化流程将计算得到的IR压降分布与IR压降分布约束进行比较，在不进行全芯片仿真的情况下，通过改变电源带参数，使PGN的性能满足约束要求，从而减少了PGN优化的时间开销。文中还对基于IGS的优化流程与传统的EDA优化流程进行了比较。用该优化流程计算的IR液滴分布与传统EDA优化程序计算的IR液滴分布的差异小于4.7%。对于ITC 99基准s9234、s13207、s35932和b19，PGN优化时间平均减少了96.2%。通过分析和仿真结果表明，基于IGS的优化流程对提高PGN的性能是可靠的。
With silicon technology further scaling, the switching activities together with GHz operation frequency greatly a ff ects the power integrity by generating large IR-drop noises. Excessive IR-drop causes functional failures such as timing failure, abnormal reset and SRAM flipping. The PGN needs to be optimized to reduce IR-drop. The traditional EDA optimization routine repeats the steps such as generating layout, extracting parameter and simulation, which greatly increases the cost of integrated circuit design. This paper proposes a novel fast optimization flow optimizing the PGN based on the improved Gauss-Seidel method (IGS), which calculates the accurate IR-drop distribution according to the initial voltage distribution and the 3-D parasitic resistance distribution of PGN. The proposed optimization flow compares the calculated IR-drop distribution with the IR-drop distribution constraints, and changes the parameters of power straps until the performance of PGN meets the requirement of constraints without full chip simulation, which reduces the time cost of PGN optimization. This paper also provides the comparison between the IGS-based optimization flow and traditional EDA optimization routine. The di ff erence of IR-drop distributions calculated by the proposed optimization flow and traditional EDA optimization routine is less than 4.7%. And the PGN optimization time has been reduced by 96.2% on average for ITC99 benchmark s9234, s13207, s35932 and b19. According to the results and analysis, the IGS-based optimization flow is reliable to improve the performance of PGNs.
登录
查看更多内容
DOI：
10.1109/3dic.2009.5306539
发表时间：
2009-10
期刊：
2009 IEEE International Conference on 3D System Integration
影响因子：
--
作者：
Nauman H. Khan;S. Alam;S. Hassoun
通讯作者：
Nauman H. Khan;S. Alam;S. Hassoun
DOI：
10.1109/icept.2014.6922723
发表时间：
2014-10
期刊：
2014 15th International Conference on Electronic Packaging Technology
影响因子：
--
作者：
Yue-Hui Huang;Mu-Shui Zhang;Hongzhou Tan
通讯作者：
Yue-Hui Huang;Mu-Shui Zhang;Hongzhou Tan
DOI：
10.1109/ectc.2014.6897340
发表时间：
2014-05
期刊：
2014 IEEE 64th Electronic Components and Technology Conference (ECTC)
影响因子：
--
作者：
Minghui Han;A. Amirkhany;Wei Xiong
通讯作者：
Minghui Han;A. Amirkhany;Wei Xiong
DOI：
10.1109/aspdac.2012.6164995
发表时间：
2012-03
期刊：
17th Asia and South Pacific Design Automation Conference
影响因子：
--
作者：
Zuowei Li;Yuchun Ma;Qiang Zhou;Yici Cai;Yu Wang;Tingting Huang;Yuan Xie
通讯作者：
Zuowei Li;Yuchun Ma;Qiang Zhou;Yici Cai;Yu Wang;Tingting Huang;Yuan Xie
DOI：
10.1109/iccad.2005.1560093
发表时间：
2005-05
期刊：
ICCAD-2005. IEEE/ACM International Conference on Computer-Aided Design, 2005.
影响因子：
--
作者：
Yu Zhong;Martin D. F. Wong
通讯作者：
Yu Zhong;Martin D. F. Wong