+ All Categories
Home > Documents > Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Date post: 04-Jan-2016
Category:
Upload: conrad-mills
View: 214 times
Download: 0 times
Share this document with a friend
Popular Tags:
27
Configurable Multi- product Floorplanni ng Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010
Transcript
Page 1: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Configurable Multi-product Floorplanning

Qiang Ma, Martin D.F. Wong,

Kai-Yuan Chao

ASP-DAC 2010

Page 2: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Outline

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 3: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 4: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction

Conventional ASIC or SoC design floorplan usually targets for one single product; and, high efforts in re-floorplan and re-convergence for different products are still required if there is no pre-design stage multi-product planning

Therefore, the problem of designing floorplans at product or market planning stage that simultaneously optimizes multiple products

Page 5: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Traditional floorplanning problem

Given a set of rectangles, each with a range of aspect ratio

Generate a packing of all the rectangles s.t. The rectangles do not overlap The total area and wire length are minimized

Page 6: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction

Multi-product Floorplanning Generate a floorplan of enough components (basi

c blocks) that is configurable for all the products The rectangular region for each product is as smal

l as possible

Page 7: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 8: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Notations

Problem input {b1, b2, …, bm} be the set of m basic blocks

{P1, P2, …, Pq} be the set of q products

[Di1, Di

2, …, Dim]T be the Demand Vector Di of Pi

Page 9: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Notations

Cutline a line obtained by extending the i

nterval of a block boundary Valid Region

A rectangular region R on a candidate floorplan formed by four Cutlines l, r, t, b (denoted by R(l, r, t, b))

Capacity Vector CR = [CR

1, CR2, …, CR

m]T of a Valid Region R, where CRj den

otes # of basic block bj lying within R Feasible Region

R is Feasible for product Pi if and only if CR D≧ i

Page 10: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Problem Formulation

Multi-product Floorplanning (MPF) Inputs:

{b1, b2, …, bm}

{P1, P2, …, Pq}

Each product Pi is associated with a Demand-Vector Di

Objective: Generate a floorplan of sufficient basic blocks to accom

modate all the q products The total area of the MPF for each product on the result

ant floorplan is minimized

Page 11: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 12: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Preprocessing

Page 13: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Exact Algorithm

Page 14: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Exact Algorithm

Horizontal Scan (H-Scan) Determine a Horizontal Feasible Regions (HFR)

for a product Horizontal interval array H is scanned Fix left boundary, shift right boundary until the

current region contains sufficient basic blocks

Page 15: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Exact Algorithm

Horizontal Scan (H-Scan)

Page 16: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Exact Algorithm

Vertical Scan (V-Scan) Given a HFR(lb, rb), find HFR*(lb, rb) with minimu

m height Find positions for tb and bb s.t. (bb-tb) is minimum At the beginning, tb = Top-Cutlines[0], bb = Botto

m-Cutlines[0] Repeat

Shift bb down until R(lb, rb, tb, bb) is feasible Shift tb down until R(lb, rb, tb, bb) is infeasible Record the current height and update the best

Until bb is no longer able to be shifted downward

Page 17: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Exact Algorithm Vertical Scan (V-Scan)

Page 18: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Greedy Heuristic

Given a candidate floorplan and a product Pi’s demand vector Di, the greedy heuristic behaves as follows The feasible region R(lb, rb, tb, bb) starts with the whole flo

orplan Greedily shrink the four boundaries

Pick one boundary b and shift it one step towards the center Its shifting reduce R by the most

Break the tie by picking the least recently used boundary The feasibility of R is maintained Scan Horizontal interval array H and vertical interval array V Stop when R can not be shrunk any more The resultant region R is a Minimal Feasible Region

Page 19: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Greedy Heuristic

Page 20: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 21: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Experimental Results

Exact Use the exact algorithm

Greedy Use the greedy heuristic

Hybrid Applying the greedy heuristic at the first stage and then

switching to the exact algorithm at the second stage, when the acceptance ratio drops below a certain value

Provide good tradeoff between solution quality and rum time

Page 22: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Experimental Results

The three method are tested and compared on a set of test cases derived from industrial data

Page 23: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Experimental Results

Page 24: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Experimental Results

Page 25: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Experimental Results

Page 26: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Introduction Problem Formulation Methodology

Exact Algorithm Greedy Heuristic

Experimental Results Conclusions

Page 27: Configurable Multi-product Floorplanning Qiang Ma, Martin D.F. Wong, Kai-Yuan Chao ASP-DAC 2010.

Conclusions

This paper introduced the Multi-product Floorplanning problem Simultaneously design for a family of related prod

ucts Significantly reduce overall design time and cost

First work in literature addressing this problem

The effectiveness of this approach is validated by promising results on a set of test cases


Recommended