17th Asia and South Pacific Design Automation Conference 2012
DOI: 10.1109/aspdac.2012.6164969
|View full text |Cite
|
Sign up to set email alerts
|

Block-level 3D IC design with through-silicon-via planning

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

1
11
0

Year Published

2012
2012
2017
2017

Publication Types

Select...
5
2
1

Relationship

2
6

Authors

Journals

citations
Cited by 28 publications
(12 citation statements)
references
References 10 publications
1
11
0
Order By: Relevance
“…This increase is mainly due to longer interconnects passing multiple dies (notably caused by nets connecting to external pins), which undermines wirelength reduction by shorter inter-die routes. Depending on die thickness, inserting multiple TSVs guided by tree construction may reduce wirelength [19]. However, this would increase TSV count notably and thus cost as well.…”
Section: Experimental Results Inmentioning
confidence: 99%
See 1 more Smart Citation
“…This increase is mainly due to longer interconnects passing multiple dies (notably caused by nets connecting to external pins), which undermines wirelength reduction by shorter inter-die routes. Depending on die thickness, inserting multiple TSVs guided by tree construction may reduce wirelength [19]. However, this would increase TSV count notably and thus cost as well.…”
Section: Experimental Results Inmentioning
confidence: 99%
“…Our optimization methodology MoDo is presented in Section III; an experimental investigation is provided in Section IV. Our conclusions on optimizing deadspace for 3D ICs and its benefits are given in Section V. [25]) [19], [20] rarely aligned nonuniform; Thermal ≈ 2 − 40µm may be encouraged low -medium irregular small -medium contigous regions; [7] [8], [9], [27] possibly aligned nonuniform; Power/Ground ≈ 10 − 40µm strongly preferred low irregular small contigous regions; [7], [13], [14], [16] [11], [13], [14], [16] necessarily aligned nonuniform; Clock ≈ 2 − 20µm may be encouraged low irregular small contigous regions; [33] [33], [34] possibly aligned…”
Section: Introductionmentioning
confidence: 99%
“…As the table shows, the maximum difference is 12.1%, but the average difference for the arithmetic circuits (AL1-AL7) is −3.6% and that for the microprocessor circuits (MP1-MP5) is −3.0%. 2 Table II compares the wirelength of the layouts of the four benchmark circuits designed by [21] and the wirelength predicted by our TSV-aware block-level 3-D wirelength distribution model. The maximum difference is 9.0% and the average difference is 3.1%, which are acceptable as a prediction result in early design stages.…”
Section: B Validation Of the Tsv-aware 3-d Wirelength Distribution Mmentioning
confidence: 99%
“…A typical example is core + memory stack [7], which provides very high memory access bandwidth. The second one is block-level integration [8], where functional blocks are partitioned into different tiers based on their logical connections. In block-level integration, the number of vertical connections is usually more than core + memory stacking.…”
Section: Benefits Of Monolithic 3d For 3d Integrationmentioning
confidence: 99%