A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU
碩士 === 國立交通大學 === 電子研究所 === 100 === This thesis proposes an innovative force-directed parallel algorithm, FDPrior, to solve the multilayer partitioning problem of 3DICs. The purpose of our research is providing a new field of vision in the partition problem of 3DICs. The growing scale and multi-laye...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2011
|
Online Access: | http://ndltd.ncl.edu.tw/handle/10451639615044959591 |
id |
ndltd-TW-100NCTU5428017 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-100NCTU54280172015-10-13T20:37:27Z http://ndltd.ncl.edu.tw/handle/10451639615044959591 A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU 三維積體電路在通用圖形處理器裡基於力使量法的平行分割演算法 Chen,Wan-Jing 陳琬菁 碩士 國立交通大學 電子研究所 100 This thesis proposes an innovative force-directed parallel algorithm, FDPrior, to solve the multilayer partitioning problem of 3DICs. The purpose of our research is providing a new field of vision in the partition problem of 3DICs. The growing scale and multi-layered structure of the 3DIC technology make it computationally expensive for EDA tools to achieve optimization goals. Exploiting the algorithmic parallelism on multi-core architectures becomes the key to attain scalable runtime. The objective is to minimize the total number of Through Silicon Vias (TSVs) while meeting the area constraint for each layer. By adopting the N-body simulation scheme and novel techniques to reduce synchronization overhead, FDPrior successfully exposes the massive parallelism on the multi-core GPGPU architecture. The experimental results on ISPD98 benchmark show that FDPrior outperforms the conventional FM algorithm by achieving in average 5.95X better TSVs and up to 303.66X runtime speedup. Compared with PP3D, a parallel 3DIC partitioning algorithm, FDPrior achieves 7.71X better TSVs with 3.35 X runtime enhancements. In recent years, the multilevel hypergraph partitioning algorithms could earn better performances than non-multilevel methods. This is why our thesis also proposes an algorithm, MFDPrior, which fulfills the multilevel framework. MFDPrior exercises the FDPrior as the essential partitioning part. When comparing with the single level FDPrior, MFDPrior demonstrates an average of 1.46X better solution quality and earns 1.44X speedup. Lai, Bo-Cheng 賴伯承 2011 學位論文 ; thesis 54 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立交通大學 === 電子研究所 === 100 === This thesis proposes an innovative force-directed parallel algorithm, FDPrior, to solve the multilayer partitioning problem of 3DICs. The purpose of our research is providing a new field of vision in the partition problem of 3DICs. The growing scale and multi-layered structure of the 3DIC technology make it computationally expensive for EDA tools to achieve optimization goals. Exploiting the algorithmic parallelism on multi-core architectures becomes the key to attain scalable runtime. The objective is to minimize the total number of Through Silicon Vias (TSVs) while meeting the area constraint for each layer. By adopting the N-body simulation scheme and novel techniques to reduce synchronization overhead, FDPrior successfully exposes the massive parallelism on the multi-core GPGPU architecture. The experimental results on ISPD98 benchmark show that FDPrior outperforms the conventional FM algorithm by achieving in average 5.95X better TSVs and up to 303.66X runtime speedup. Compared with PP3D, a parallel 3DIC partitioning algorithm, FDPrior achieves 7.71X better TSVs with 3.35 X runtime enhancements.
In recent years, the multilevel hypergraph partitioning algorithms could earn better performances than non-multilevel methods. This is why our thesis also proposes an algorithm, MFDPrior, which fulfills the multilevel framework. MFDPrior exercises the FDPrior as the essential partitioning part. When comparing with the single level FDPrior, MFDPrior demonstrates an average of 1.46X better solution quality and earns 1.44X speedup.
|
author2 |
Lai, Bo-Cheng |
author_facet |
Lai, Bo-Cheng Chen,Wan-Jing 陳琬菁 |
author |
Chen,Wan-Jing 陳琬菁 |
spellingShingle |
Chen,Wan-Jing 陳琬菁 A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
author_sort |
Chen,Wan-Jing |
title |
A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
title_short |
A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
title_full |
A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
title_fullStr |
A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
title_full_unstemmed |
A Force-Directed Based Parallel Partitioning Algorithm for Three Dimensional Integrated Circuits on GPGPU |
title_sort |
force-directed based parallel partitioning algorithm for three dimensional integrated circuits on gpgpu |
publishDate |
2011 |
url |
http://ndltd.ncl.edu.tw/handle/10451639615044959591 |
work_keys_str_mv |
AT chenwanjing aforcedirectedbasedparallelpartitioningalgorithmforthreedimensionalintegratedcircuitsongpgpu AT chénwǎnjīng aforcedirectedbasedparallelpartitioningalgorithmforthreedimensionalintegratedcircuitsongpgpu AT chenwanjing sānwéijītǐdiànlùzàitōngyòngtúxíngchùlǐqìlǐjīyúlìshǐliàngfǎdepíngxíngfēngēyǎnsuànfǎ AT chénwǎnjīng sānwéijītǐdiànlùzàitōngyòngtúxíngchùlǐqìlǐjīyúlìshǐliàngfǎdepíngxíngfēngēyǎnsuànfǎ AT chenwanjing forcedirectedbasedparallelpartitioningalgorithmforthreedimensionalintegratedcircuitsongpgpu AT chénwǎnjīng forcedirectedbasedparallelpartitioningalgorithmforthreedimensionalintegratedcircuitsongpgpu |
_version_ |
1718049976406245376 |