1 Wide-Sense Nonblocking Under New Compound Routing Strategies Junyi David Guo( 郭君逸 ) 師範大學 數學系 2011/06/29 Joint work with F.H. Chang.

Slides:



Advertisements
Similar presentations
Communication Lower Bound for the Fast Fourier Transform Michael Anderson Communication-Avoiding Algorithms (CS294) Fall 2011.
Advertisements

Routing and Congestion Problems in General Networks Presented by Jun Zou CAS 744.
Wide-sense Nonblocking for Multi-log(d^n,m,k) Networks under the Minimum Index Strategy Speaker: Fei-Huang Chang Coauthers: Ding-An Hsien, Chih-Hung Yen.
Finding the Sites with Best Accessibilities to Amenities Qianlu Lin, Chuan Xiao, Muhammad Aamir Cheema and Wei Wang University of New South Wales, Australia.
Lecture 4. Topics covered in last lecture Multistage Switching (Clos Network) Architecture of Clos Network Routing in Clos Network Blocking Rearranging.
Data Structures by R.S. Chang, Dept. CSIE, NDHU1 Chapter 2 Algorithm Design Human Problems Result Input Data Structures Processing Output Data Structures.
FPGA Intra-cluster Routing Crossbar Design Dr. Philip Brisk Department of Computer Science and Engineering University of California, Riverside CS 223.
Performances of List Scheduling for set partition problems 吳邦一歐秀慧 樹德科技大學 資訊工程系 報告人:歐秀慧.
國立暨南國際大學 National Chi Nan University A Study of (k, n)-threshold Secret Image Sharing Schemes in Visual Cryptography without Expansion Presenter : Ying-Yu.
1 Advancing Supercomputer Performance Through Interconnection Topology Synthesis Yi Zhu, Michael Taylor, Scott B. Baden and Chung-Kuan Cheng Department.
Coalition Formation and Price of Anarchy in Cournot Oligopolies Joint work with: Nicole Immorlica (Northwestern University) Georgios Piliouras (Georgia.
1 Wide-Sense Nonblocking Multicast in a Class of Regular Optical Networks From: C. Zhou and Y. Yang, IEEE Transactions on communications, vol. 50, No.
Routing Permutation in the Baseline Network and in the Omega Network Student : Tzu-hung Chen 陳子鴻 Advisor : Chiuyuan Chen Department of Applied Mathematics.
Multiple Processor Systems
Complexity 7-1 Complexity Andrei Bulatov Complexity of Problems.
1 Data Structures Performance Analysis. 2 Fundamental Concepts Some fundamental concepts that you should know: –Dynamic memory allocation. –Recursion.
Advanced Topics in Algorithms and Data Structures Page 1 Parallel merging through partitioning The partitioning strategy consists of: Breaking up the given.
Dynamic Index Coding User set N Packet set P Broadcast Station N N p p p Michael J. Neely, Arash Saber Tehrani, Zhen Zhang University.
Chapter 3 Growth of Functions
CS541 Advanced Networking 1 Spectrum Sharing in Cognitive Radio Networks Neil Tang 3/23/2009.
048866: Packet Switch Architectures Dr. Isaac Keslassy Electrical Engineering, Technion Scaling.
Dean H. Lorenz, Danny Raz Operations Research Letter, Vol. 28, No
EE 685 presentation Optimization Flow Control, I: Basic Algorithm and Convergence By Steven Low and David Lapsley Asynchronous Distributed Algorithm Proof.
CSE 291-a Interconnection Networks Lecture 7: February 7, 2007 Prof. Chung-Kuan Cheng CSE Dept, UC San Diego Winter 2007 Transcribed by Thomas Weng.
1 A Linear Space Algorithm for Computing Maximal Common Subsequences Author: D.S. Hirschberg Publisher: Communications of the ACM 1975 Presenter: Han-Chen.
Bioinformatics Protein structure prediction Motif finding Clustering techniques in bioinformatics Sequence alignment and comparison Phylogeny Applying.
Limits of Local Algorithms in Random Graphs
Providing Resiliency to Load Variations in Distributed Stream Processing Ying Xing, Jeong-Hyon Hwang, Ugur Cetintemel, Stan Zdonik Brown University.
CSE Advanced Computer Architecture Week-11 April 1, 2004 engr.smu.edu/~rewini/8383.
寓數學於魔方 DavidGuo 師大數學系 郭君逸 助理教授. 魔方的流行 當幸福來敲門 瓦力 蕭敬騰 2 圖片來源:開眼電影網開眼電影網 圖片來源:開眼電影網開眼電影網.
NETWORK CODING. Routing is concerned with establishing end to end paths between sources and sinks of information. In existing networks each node in a.
Sorting. Quick Sort Example 2, 1, 3, 4, 5 2, 4, 3, 1, 5, 5, 6, 9, 7, 8, 5 S={6, 5, 9, 2, 4, 3, 5, 1, 7, 5, 8} 1, 234, 5 5 5, 6, 7, 8, 9.
1 Dynamic Interconnection Networks Miodrag Bolic.
CS 345: Chapter 10 Parallelism, Concurrency, and Alternative Models Or, Getting Lots of Stuff Done at Once.
MINATO ZDD Project Efficient Enumeration of the Directed Binary Perfect Phylogenies from Incomplete Data Toshiki Saitoh (ERATO) Joint work with Masashi.
EE384y EE384Y: Packet Switch Architectures Part II Scaling Crossbar Switches Nick McKeown Professor of Electrical Engineering and Computer Science,
Real Zeros of Polynomial Functions. Quick Review.
1 The Price of Defense M. Mavronicolas , V. Papadopoulou , L. Michael ¥, A. Philippou , P. Spirakis § University of Cyprus, Cyprus  University of Patras.
Shanghai Jiao Tong University 2012 Indirect Networks or Dynamic Networks Guihai Chen …with major presentation contribution from José Flich, UPV (and Cell.
CSE 20 – Discrete Mathematics Dr. Cynthia Bailey Lee Dr. Shachar Lovett Peer Instruction in Discrete Mathematics by Cynthia Leeis licensed under a Creative.
1 Multicasting in a Class of Multicast-Capable WDM Networks From: Y. Wang and Y. Yang, Journal of Lightwave Technology, vol. 20, No. 3, Mar From:
 Rubik’s cube = 『魔術方塊』  Invented by a Hungarian professor named Ernő Rubik in 1974.
Chapter 6 Relaxation (1) CDS in unit disk graph
Introduction to Probability Theory ‧ 2- 2 ‧ Speaker: Chuang-Chieh Lin Advisor: Professor Maw-Shang Chang National Chung Cheng University Dept. CSIE, Computation.
EE 685 presentation Optimization Flow Control, I: Basic Algorithm and Convergence By Steven Low and David Lapsley.
2.4/2.52.4/2.5 Real Zeros of Polynomial Functions.
Basic Linear Algebra Subroutines (BLAS) – 3 levels of operations Memory hierarchy efficiently exploited by higher level BLAS BLASMemor y Refs. FlopsFlops/
Computer Science and Engineering Copyright by Hesham El-Rewini Advanced Computer Architecture.
Circuit Switching Circuit switching networks,
Ch 8. Switching. Switch  Devices that interconnected with each other  Connecting all nodes (like mesh network) is not cost-effective  Some topology.
Vasilis Syrgkanis Cornell University
Based on An Engineering Approach to Computer Networking/ Keshav
Zaiben Chen et al. Presented by Lian Liu. You’re traveling from s to t. Which gas station would you choose?
Of 17 Limits of Local Algorithms in Random Graphs Madhu Sudan MSR Joint work with David Gamarnik (MIT) 7/11/2013Local Algorithms on Random Graphs1.
1 Computer Architecture & Assembly Language Spring 2001 Dr. Richard Spillman Lecture 26 – Alternative Architectures.
Complexity Classes.
Does Topology Control Reduce Interference?
New Characterizations in Turnstile Streams with Applications
Dynamic connection system
Refer example 2.4on page 64 ACA(Kai Hwang) And refer another ppt attached for static scheduling example.
Monomer-dimer model and a new deterministic approximation algorithm for computing a permanent of a 0,1 matrix David Gamarnik MIT Joint work with Dmitriy.
Static and Dynamic Networks
Multiprocessors Interconnection Networks
Analysis and design of algorithm
High Performance Computing & Bioinformatics Part 2 Dr. Imad Mahgoub
Dynamic Interconnection Networks
Birds Eye View of Interconnection Networks
Circuit Switch Design Principles
Design Principles of Scalable Switching Networks
Introduction and Overview
Presentation transcript:

1 Wide-Sense Nonblocking Under New Compound Routing Strategies Junyi David Guo( 郭君逸 ) 師範大學 數學系 2011/06/29 Joint work with F.H. Chang

2 Applications Come from the need to interconnect telephones Interconnect processors with memories Data transmission Conference calls Satellite communication

3 One frequently discussed topic in switching networks is its nonblocking property.

4 Preliminaries Crossbars, X nm inlets, outlets 1 2 n 1 2 m 1 2 n 1 2 m

5 Multi-stage interconnection network inputs outputs stage

6 Symmetry C(n, m, r) C(2, 4, 3) n 1 2 r 1 2 m n 1 2 r

7 Definitions Request Strictly nonblocking(SNB) Wide-sense nonblocking(WSNB)

8 Matrix O1O1 O2O2 O3O3 I1I1 I2I2 I3I3 1 3

9 3-stage Clos Network [Clos 1953] C(n 1, r 1, m, n 2, r 2 ) is SNB iff m ≥ n 1 +n C(n, m, r) is SNB iff m ≥ 2n-1.

10 [n+5,2n-2] n+4 n+3 [12,n-1]11[6,10]2n-14,51,2,3 n+2 n,n+1

11 Algorithms Cyclic dynamic (CD) Cyclic static (CS) Save the unused (STU) Packing (P) Minimum index (MI) Compound

12

13 Literature review

14 Cyclic Dynamic CD can reach any state. Theorem C(n, m, r) is WSNB under CD iff m ≥ 2n-1. [1,n-1] [n,2n-2]2n-1

15 Cyclic Static C(n, m, r) is WSNB under CS iff m ≥ 2n-1 [1,n] [n,2n-2] [1,n-1] [n,2n-2]

16 Packing For P, hence STU, C(n, m, r), r ≥ 3, is WSNB iff m ≥ 2n-1. [1,n] n [1,n-1] n+1 n [1,n-1] n+1 n [1,n-1] n+1 n [1,n-1] n+1 n+2 n [1,n-1] n+1 n+2 n [1,n-1] n+1,n+2 n+2 n [1,n-1] [n+1,2n-2]

17 MI C(n, m, r) is WSNB under MI iff m ≥ 2n-1 X Y 2n-3

18 Packing+MI, STU+MI C(n, m, r), r ≥ 3, is WSNB iff m ≥ 2n-1.

19 3-stage Clos Network C(n 1, r 1, m, n 2, r 2 ) C(2, 4, 3, 3, 2) n1n1 n2n2 1 2 r1r1 r2r m

20 Asymmetry C(n 1, r 1, m, n 2, r 2 ) is WSNB iff m ≥ n 1 +n 2 -1 under every known routing strategies.

21 Multi-log d N Networks First proposed by Lea(1990). Baseline

22 Baseline d-nary baseline network BL d (n) d n inputs, d n outputs n stages d n-1 d  d crossbars BL d (n-1) 1 2 d n X dd

23 Baseline Example BL 2 (4)

24 Multi-log d N Network Theorem(Shyy & Lea 1991, Hwang 1998) Multi- log d N network is SNB if p ≥ p(n), where

25 Theorem Multi-log d N network is SNB only if p ≥ p(n). x yxy

26 Multi-log d N Network

27 Theorem Multi-log d N network is WSNB under CD and CS iff p ≥ p(n).

28 Graph Model The graph model of BL 2 (4)

29 P or STU Multi-log d N network is WSNB under P or STU iff p ≥ p(n).

30 MI Multi-log d N network is WSNB under MI iff p ≥ p(n). I1I1 I2I2 O1O1 O2O2 I1I1 I2I2 O1O1 O2O2 stage

31 For n even

32 P+MI and STU+MI Multi-log d N network is WSNB under P+MI and STU+MI iff p ≥ p(n).

33 Generalizations General vertical-copy network.

34 Generalizations A vertical-copy network V is WSNB under the CS and CD routing iff V is SNB. Theorem A vertical-copy network V is WSNB under the P routing iff V is SNB.

35 Larger Perspective Theorem(Tsai, Wang, and Hwang 2001) For, C(n, m, r) is WSNB if and only if m ≥ 2n-1.

36 Conclusion In 3-stage Clos Networks, the conclusions for all r and all known algorithms are known. The proofs are not messy. The necessity condition of Multi-log d N network is SNB. Extended to multi-log n N netowrks and vertical-copy netowrks. Find a relation between multi-log n N netowrks and 3-stage Clos network. The concept of MBC

37 Open Problems Find a single-selected case for P, even STU, under 3-stage Clos network or multi-log d network. Multi-log network with extra stage. MI under vertical-copy network Find a general argument independent of any routing algorithm. Conjecture: There is no good WSNB algorithm under one-to-one traffic.

38 Thank you!

39 Beneš(1965) Theorem C(n, m, 2) is WSNB under STU if m ≥ 3n/2. a b c d |a  b  c  d| < 3n/2 |a  d| < n |b  c| < n

40 Minimum Index 1,3,4,5,6,7 2,8 4,5,6,7 1,2,3,8

41 Packing

42 Reduce the result of Tsai et al. to polynomial bound of r and extend to multi-log d N networks.

Fewest move Alexander H. Frey, Jr. and David Singmaster. Handbook of Cubik Math. Enslow Publishers, ◦ 最少 17 步,最多 52 步 ◦ 猜測 : 「上帝的數字 ( God’s number) 」為 20 。 43

Fewest move Michael Reid. ◦ Lower bounds: Superflip requires 20 face turns. (1995) U R2 F B R B2 R U2 L B2 R U' D' R2 F R' L B2 U2 F2Superflip requires 20 face turns ◦ Upper bounds: New upper bounds. (1995)New upper bounds  Face turn metric(FTM): 29 moves.  Quarter turn metric(QTM): 42 moves. 44

Fewest move Silviu Radu. ◦ Rubik can be solved in 27f. (2006/04) Rubik can be solved in 27f ◦ Paper. (June 30, 2007) Paper  Face turn metric(FTM): 27 moves.  Quarter turn metric(QTM): 34 moves. ◦ Using GAP(Groups, Algorithms, Programming): a System for Computational Discrete AlgebraGAP 45

Fewest move Daniel Kunkle and Gene Cooperman, 26 Moves Suffice for Rubik's Cube, Proc. of International Symposium on Symbolic and Algebraic Computation (ISSAC '07), ACM Press, ◦ FTM: 26 moves. ◦ Gene Cooperman, Larry Finkelstein, and Namita Sarawagi. Applications of Cayleygraphs. In AAECC: Applied Algebra, Algebraic Algorithmsand Error-Correcting Codes, InternationalConference, pages LNCS, Springer- Verlag,  FTM: 11 moves.  QTM: 14 moves. 46

Fewest move Tomas Rokicki. ◦ 25 Moves Suffice for Rubik’s Cube. (2008.3) 25 Moves Suffice for Rubik’s Cube ◦ 23 Moves Suffice for Rubik’s Cube. (2008.4) 23 Moves Suffice for Rubik’s Cube ◦ 22 Moves Suffice for Rubik’s Cube. (2008.8) 22 Moves Suffice for Rubik’s Cube ◦ God’s Number is 20. (2010.7) God’s Number is 20 47

3x3 方塊的變化數 There are Cube subgroups. 48

Partition into 2,217,093,120 sets of 19,508,428,800 positions. Reduce to 55,882,296. About 35 CPU years. 49 God’s Number is 20

50 Number of Positions