Distributed Computing Seminar, Semester Alef, Fall 2004

 

Professor Yehuda Afek

 Sundays, ??? @ Schreiber ???

This seminar is intended for graduate students interested to pursue research in Asynchronous Distributed Computing.  (Following on the distributed computing class I taught last year (Spring 2004)).  Every week we will review a specific topic, usually covered by one recent paper from a conference or a journal (see below).  The student leading the discussion in a given week should read the paper(s) carefully and prepare a presentation discussion on the paper to be held in the meeting.

 Participants are required to:

  • present a research paper(s), and lead a discussion on the topic;
  • Write a 1-2 page report which critically evaluates the papers (flaws in the model, not exciting when compared to related work, mistakes, etc.), give a vague or specific idea how to improve the paper, and propose what else can be done with this topic;
  • attend the talks of the seminar and actively participate in the discussions.

Your presentation should cover the motivation for the problem as well as some technical parts of the paper in detail. Assume that the other participants know nothing about the subject.   You are encouraged to deviate from the logical structure of the paper and present it in the most lucid (clear) presentation of the topic.

Schedule of Presentations

 

 

date

title

presenter

material

 

2004/10/17

Introduction

Yehuda Afek

 

 

2004/10/24

 

 

 

 

2004/10/31

 

 

 

 

2004/11/7

 

 

 

 

 

2004/11/14

 

 

 

 

 

2004/11/21

 

 

 

 

 

2004/11/28

 

 

 

 

 

2004/12/5

 

 

 

 

 

2004/12/12

 

 

 

 

 

2004/12/19

 

 

 

 

 

2004/12/26

 

 

 

 

 

2005/1/2

 

 

 

 

 

2005/1/9

 

 

 

 

 

2005/1/16

 

 

 

 

 

 

 

 

 

 

Suggested Papers

 

 Title

Comments

Assigned

  1. High performance dynamic lock-free hash tables and list-based sets Maged Michael, SPAA 2002, pg 73-82

Using CAS to implement efficiently hash function, for OS operations.

 

  1. The Concurrency Hierarchy, and Algorithms for Unbounded Concurrency, Gafni, Merritt, Taubenfeld PODC 01

Wait-Free even if there is no bound on the number of processes that may arrive

 

  1. Implementing Shared Memory Overwriting Objects Hanan Weisman, Master Thesis, Tel-Aviv 1996 Relating Sequential Consistency, Linearizability and First-Come-First-Served Gafni and Lamport Manuscript 2002

Imlementing swap from other consensus-2 objects.

 

  1. An Almost Non-blocking Stack Hans-J. Boehm, PODC-2004

 Non-blocking until n approach

 

  1. Bringing Practical Lock-Free Synchronization to 64-bit Applications  Simon Doherty, Maurice Herlihy, Victor Luchangco and Mark Moir Practical Lock-free and Wait-Free LL/SC/VL Implementations using 64bit CAS Maged Michael, IBM DISC 2004

Implementing in single 64 bit word what was implemented in a split 64 (into two 32).

 

  1. A round-by-round failure detector - unifying synchrony and asynchrony,  Eli Gafni PODC-98  pg 143-152

Model a system by its round-by-round failure detector, and then relating the diff models

 

  1. A Simple Algorithmically Reasoned Characterization of Wait-free Computations Elizabeth Borowsky and Eli Gafni, PODC-97

Iterated simulation with immediate snapshots.  New model and reduction technique.

 

  1. Split-Ordered Lists - Lock-Free Extensible Hash Tables <http://www.cs.tau.ac.il/%7Eorish/shalev-shavit.ps.gz>/ Shalev and Shavit PODC 2003.

 

 

  1. Concurrent Data Structure,  book chapter, Moir and Shavit,

 

 

  1. Obstruction-free Synchronization: Double-ended Queues as an Example Herlihy, V. Luchangco, and M. Moir ICDCS03

 

 

  1. The Repeat Offender Problem: A Mechanism for Supporting Dynamic-Sized, Lock-Free Data Structures Herlihy, V. Luchangco, and M. Moir, DISC02

 

 

  1. On the Inherent Weakness of Conditional Synchronization Primitives Faith Fich, Danny Handler, Nir Shavit Operation Valency and the Cost of Coordination Danny Handler and Nir Shavit

Sqrt(n) lowerbounds when large valence.

 

 

  1. Efficient Synchronous Snapshots Alex Brodsky and Faith Fich

 Synchronous SnapShot in O(1)-update and O(m)-scan and variants

 

  1. Lower bounds for adaptive collect and related objects Hagit Attiya, Faith Fich, Yaniv Kaplan

F(k)-adaptive total contention implies       f^{-1}(n) multi writer registers.  + variants.

 

  1.  

 

 

 

  1.  

 

 

  1.