Please refer to CAC '05 website
for
details on the upcoming CAC '05 workshop.
Workshop on Communication Architecture for Clusters

April 26 (Monday)
7:50 - 8:00 Welcome and Workshop Introduction
8:00 - 9:00 Keynote Talk
Mark Seager,
Lawrence Livermore National Laboratory
Title: Communication Architecture Challenges for a High Performance Multi-Cluster Simulation Environment
9:00 - 10:00 Session I
Communication and NIC Support
Session Chair: Fabrizio Petrini, Los Alamos National Laboratory
Each presentation is for 25 minutes. The last 10 minutes
of the session is for discussion.
- Opportunities and Challenges of Modern Communication
Architectures: QsNet a Case Study,
Sameer Kumar and Laxmikant Kale
(UIUC)
-
10:00 - 10:30 Break
10:30 - 12:00 Session II
Novel MPI Implementation
Session Chair: Scott Pakin, Los Alamos National Laboratory
Each presentation is for 25 minutes. The last 15 minutes
of the session is for discussion.
- An Analysis of NIC Resource Usage for Offloading MPI, Ron
Brightwell and Keith Underwood, (Sandia National Laboratory)
- Implementing Efficient and Scalable Flow Control Schemes in MPI
over InfiniBand, Jiuxing Liu and Dhabaleswar K. Panda (Ohio State
University)
- An MPICH2 Channel Device Implementation over VAPI on
InfiniBand, Wolfgang Rehm, Rene Grabner, Frank Mietke, Torsten Mehlan
and Christian Siebert, (Chemnitz Univ. of Technology, Germany)
12:00 - 1:30 Lunch (On your own)
1:30 - 2:30 Session III
Routing and Reconfiguration in InfiniBand
Session Chair: Jarek Nieplocha, Pacific Northwest National Laboratory
Each presentation is for 25 minutes. The last 10 minutes
of the session is for discussion.
- A Transition-Based Fault Tolerant Routing Algorithm For
InfiniBand Networks, Jose Miguel Montaņana, Jose Flich, Antonio
Robles, Pedro Lopez, and Jose Duato (UPV, Spain)
- Use of Provisional Routes to Speed-up Change Assimilation in
InfiniBand Networks, Aurelio Bermudez, Rafael Casado, Francisco J.
Quiles (UCLM, Spain) and Jose Duato (UPV, Spain)
2:30 - 2:50 Break
2:50 - 4:20 Session IV
Communication Impact on Systems
Session Chair: Arthur B. (Barney) Maccabe, Univ. of New Mexico
Each presentation is for 25 minutes. The last 15 minutes
of the session is for discussion.
- SHAK: Eliminating Faked Three-way Handshaking in Socket Handoff,
Hai Jin, Dan Tang, and Yongkun Zhang (Huazhong Univ. of Science
and Technology, China)
- Performance Analysis of Remote File System Access over a
High-Speed Local Network, Brice Goglin and Loic Prylli (LIP/ENS-Lyon,
France)
- Experimental Studies of Scalability in Clustered Web Systems,
Ibrahim Haddad and Greg Butler (Concordia University, Canada)
4:20 - 4:30 Short Break
4:30 - 6:00 Panel Session
Title: Cluster Architecture: Standardized Dystopia, or Teething Pains?
Description:
Clusters are a successful system architecture by any standard. Reflecting
that, many key cluster elements are formally standardized or on the way to
it. Yet problems are still reported in areas like efficiency, debugging,
and tuning. Are these misperceptions, "teething pains" of new tools, or
are we freezing fundamental problems into standards - and if so, what, and
what directions should we take to fix this?
Moderator: Greg Pfister, IBM
Panelists:
Ron Brightwell, Sandia National Lab
Arthur B. (Barney) Maccebe, Univ. of New Mexico
Jarek Nieplocha, Pacific Northwest National Lab
Fabrizio Petrini, Los Alamos National Lab
6:00 - Adjourn
Workshop registration is handled by the IPDPS '04 conference.
There is a single registration for the conference and all of its 17
workshops.
Please
visit IPDPS'04 web page for
registration and hotel information.
Deadline for hotel reservation is March 31, 2004 and
deadline for advance registration is April 12, 2004.
THEME:
The availability of commodity PCs/workstations and high-speed networks
(Local Area Networks and System Area Networks) at low prices enabled
the development of low-cost clusters. These clusters are being
targeted for support of traditional high-end computing applications as
well as emerging applications, especially those requiring
high-performance servers. Designing high-performance and scalable
clusters for these emerging applications requires design and
development of high-performance communication and I/O subsystems,
low-overhead programming environment support and support for Quality
of Service (QoS). New standards such as
InfiniBand
Architecture (IBA) and PCI Express AS, and availability of
high-speed networking products (Myrinet, Quadrics, IBA 4X, and
10GigEthernet)
are providing exciting ways to design
high-performance communication and I/O architectures for clusters.
A large number of research groups from academia, industry, and
research labs are currently engaged in the above research directions.
The goal of this workshop is to bring together researchers and
practitioners working in the areas of communication, I/O, and
architecture to discuss state-of-the-art solutions as well as future
trends for designing scalable, high-performance, and cost-effective
communication and I/O architectures for clusters.
The first three workshops in this series
(CAC '01,
CAC '02, and
CAC '03)
were held
in conjunction with IPDPS conferences, and they were very successful.
The CAC '04 workshop plans to continue this tradition.
TOPICS OF INTEREST:
Topics of interest for the workshop include but are not limited to:
-
Router/switch, network, and network-interface architecture for
supporting efficient point-to-point communication, collective
communication, and I/O at intra-cluster and inter-cluster levels.
-
Design, development, and implementation of low-level
communication and I/O protocols
(GM, TCP/IP, VAPI, SDP, DAPL, SRP, iSCSI, RDMA over IP, etc) on
different networking and interconnect
technologies (such as Myrinet, 10Gigabit Ethernet, InfiniBand, Quadrics,
TCP Offload Engine, etc.).
-
High-performance implementation of different programming layers
(Message Passing Interface (MPI), Distributed Shared Memory such
as TreadMarks, Get/Put, Global Arrays, sockets, etc.) and File
Systems (such as PVFS and DAFS).
-
Communication and architectural issues related to switch
organization, flow control, congestion control, routing and
deadlock-handling, load balancing, reliability, and QoS support.
-
Strategies, algorithms, and protocols for management of communication
resources, including topology discovery, hot update/replacement of
components, dynamic reconfigurations, etc.
-
Performance evaluation and tools for different application areas,
including interprocessor communication and I/O, etc.
Results of both theoretical and practical significance will be
considered.
PROCEEDINGS:
The proceedings of this workshop will be published together with the
proceedings of other IPDPS '04 workshops by the IEEE Computer Society
Press.
PAPER SUBMISSIONS:
We are planning a purely web submission and review process.
Authors are requested to submit papers (in PDF format) not exceeding
10 single-spaced pages, including abstract, five key words, contact
address, figures, and references. Detailed instructions on
web submissions will be available soon.
Note: the PDF file must be viewable using the ``acroread'' tool. It
is also important, when creating your PDF file, to use a page size of
8.5x11 inches (LETTER sized output not A4), since an A4 sized page may
be truncated on a LETTER sized printer.
SCHEDULE:
Paper submission: December 3, 2003
Notification of acceptance: January 4, 2004
Camera-ready due: January 23, 2004
ABSTRACT/PAPER SUBMISSION PAGE IS HERE
WORKSHOP CO-CHAIRS:
Dhabaleswar K. Panda (Ohio State),
Jose Duato (Tech. Univ. of Valencia, Spain),
and Craig Stunkel (IBM TJ Watson Research Center)
PROGRAM COMMITTEE:
- Bulent Abali (IBM TJ Watson)
- Mohammad Banikazemi (IBM TJ Watson)
- Angelos Bilas (Univ. of Toronto, Canada)
- Alan Benner (IBM)
- Ron Brightwell (Sandia National Lab)
- Darius Buntinas (Argonne National Lab)
- Toni Cortes (UPC, Spain)
- Wu-Chun Feng (Los Alamos National Lab)
- Jose Flich (Tech. Univ. of Valencia, Spain)
- Mitchell Gusat (IBM, Zurich)
- Mark Heinrich (Univ. of Central Florida)
- Manolis G.H. Katevenis (FORTH and Univ. of Crete, Greece)
- Nectarios G. Koziris (National Technical Univ. of Athens, Greece)
- Mario Lauria (Ohio State)
- Olav Lysne (Univ. of Oslo, Norway)
- Arthur (Barney) Mccabe (Univ. of New Mexico)
- Pankaj Mehra (HP)
- Shubu Mukherjee (Intel)
- Jarek Nieplocha (Pacific Northwest National Lab)
- Scott Pakin (Los Alamos National Lab)
- Fabrizio Petrini (Los Alamos National Lab)
- Greg Pfister (IBM)
- Timothy Pinkston (Univ. of Southern California)
- Wolfgang Rehm (Tech. Univ. of Chemnitz, Germany)
- Antonio Robles (UPV, Spain)
- Tom Rokicki (Instantis)
- Reza Rooholamini (Dell)
- Evan Speight (Cornell Univ.)
- Thomas M. Stricker (ETH, Zurich, Switzerland)
- Peter Varman (NSF and Rice Univ.)
- Pete Wyckoff (Ohio Supercomputer Center)
- Mazin Yousif (Intel)
PUBLICITY COORDINATORS:
Darius Buntinas (Argonne National Lab) and
Nectarios G. Koziris (National Technical Univ. of Athens, Greece)
ADDITIONAL INFORMATION:
For further questions, send e-mail to
cac@cse.ohio-state.edu.
Last updated April 22, 2004