CITI (fwd)

From: radev@umich.edu
Date: Fri Sep 08 2006 - 12:29:39 EDT


>From radev Fri Sep 8 12:28:33 2006
Return-Path: <parallel-request@eecs.umich.edu>
Received: from mail.umich.edu [141.211.14.95]
        by localhost with IMAP (fetchmail-6.2.5)
        for radev@localhost (single-drop); Fri, 08 Sep 2006 12:28:33 -0400 (EDT)
Received: from murder (mail.umich.edu [141.211.14.91])
         by copland.mail.umich.edu (Cyrus v2.3.7) with LMTPA;
         Fri, 08 Sep 2006 12:28:25 -0400
X-Sieve: CMU Sieve 2.3
Received: from murder ([unix socket])
         by mail.umich.edu (Cyrus v2.2.12) with LMTPA;
         Fri, 08 Sep 2006 12:28:25 -0400
Received: from madman.mr.itd.umich.edu (madman.mr.itd.umich.edu [141.211.14.75])
        by jacknife.mail.umich.edu (8.12.11.20060308/8.12.11) with ESMTP id k88GSP6c016840
        for <radev@mail.umich.edu>; Fri, 8 Sep 2006 12:28:25 -0400
Received: FROM krusty.si.umich.edu (krusty.si.umich.edu [141.211.202.33])
        BY madman.mr.itd.umich.edu ID 45019A28.AF052.9849 ;
         8 Sep 2006 12:28:24 -0400
Received: by krusty.si.umich.edu (Postfix)
        id CABFBC242; Fri, 8 Sep 2006 12:08:13 -0400 (EDT)
Delivered-To: radev@si.umich.edu
Received: from smtp.eecs.umich.edu (smtp.eecs.umich.edu [141.213.4.43])
        by krusty.si.umich.edu (Postfix) with ESMTP id C4D95C23E
        for <radev@si.umich.edu>; Fri, 8 Sep 2006 12:08:13 -0400 (EDT)
Received: from smtp.eecs.umich.edu (localhost.eecs.umich.edu [127.0.0.1])
        by smtp.eecs.umich.edu (8.13.8/8.13.6) with ESMTP id k88GS60x015324
        (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO);
        Fri, 8 Sep 2006 12:28:06 -0400
Received: (from list@localhost)
        by smtp.eecs.umich.edu (8.13.8/8.13.2/Submit) id k88GS5r3015318;
        Fri, 8 Sep 2006 12:28:05 -0400
Resent-Date: Fri, 8 Sep 2006 12:28:05 -0400
X-Authentication-Warning: smtp.eecs.umich.edu: list set sender to parallel-request@smtp.eecs.umich.edu using -f
Date: Fri, 8 Sep 2006 12:28:05 -0400
From: "Quentin F. Stout" <qstout@eecs.umich.edu>
Message-Id: <200609081628.k88GS55P021963@wellington.eecs.umich.edu>
To: parallel@eecs.umich.edu
Subject: CITI
X-Virus-Scan: : UVSCAN at UoM/EECS
Resent-Message-ID: <WGBHLPKsG5D.A.QvD.VoZAFB@smtp.eecs.umich.edu>
Resent-From: parallel@eecs.umich.edu
X-Mailing-List: <parallel@smtp.eecs.umich.edu> archive/latest/102
X-Loop: parallel@smtp.eecs.umich.edu
List-Post: <mailto:parallel@smtp.eecs.umich.edu>
List-Help: <mailto:parallel-request@smtp.eecs.umich.edu?subject=help>
List-Subscribe: <mailto:parallel-request@smtp.eecs.umich.edu?subject=subscribe>
List-Unsubscribe: <mailto:parallel-request@smtp.eecs.umich.edu?subject=unsubscribe>
Precedence: list
Resent-Sender: parallel-request@eecs.umich.edu
X-Spam-Status: No, score=-0.0 required=5.0 tests=BAYES_50,SPF_HELO_PASS,
        SPF_PASS autolearn=ham version=3.2.0-r431796
X-Spam-Checker-Version: SpamAssassin 3.2.0-r431796 (2006-08-16) on
        smtp.eecs.umich.edu
X-Virus-Scan: : UVSCAN at UoM/EECS

UM's CITI is part of this (though not a large part)

 From: HPCwire <hpcmore@news.hpcwire.com>

DOE SCIDAC NEWS ===============================================================
[ ] M876006 ) Carnegie Mellon to Lead Petascale Data Storage Institute....6.5K

  
  The U.S. Department of Energy (DOE) has awarded a five-year, $11
  million grant to researchers at three universities and five national
  laboratories to find new ways of managing the torrent of data that
  will be produced by the coming generation of supercomputers.
  
  The innovations developed by the new Petascale Data Storage Institute
  will enable U.S. scientists to fully exploit the power of these new
  computing systems, which will be capable of performing millions of
  billions of calculations each second.
  
  The institute combines the talents of computer scientists at Carnegie
  Mellon University, the University of California at Santa Cruz and the
  University of Michigan with those of researchers at the DOE's Los
  Alamos, Sandia, Oak Ridge, Lawrence Berkeley and Pacific Northwest
  national laboratories.
  
  Increased computational power is necessary because scientists depend
  on computer modeling to simulate extremely complicated phenomena, such
  as global warming, earthquake motions, the design of fuel-efficient
  engines, nuclear fusion and the global spread of disease. Computer
  simulations provide scientific insights into these processes that are
  often impossible through conventional observation or experimentation.
  This capability is critical to U.S. economic competitiveness,
  scientific leadership and national security, the President's
  Information Technology Advisory Committee concluded last year.
  
  But simply building computers with faster processing speeds -- the new
  target threshold is a quadrillion (a million billion) calculations per
  second, or a "petaflop" -- will not be sufficient to achieve those
  goals. Garth Gibson, a Carnegie Mellon computer scientist who will
  lead the data storage institute, said new methods will be needed to
  handle the huge amounts of data that computer simulations both use and
  produce.
  
  "Petaflop computers will achieve their high speeds by adding
  processors -- hundreds of thousands to millions of processors," said
  Gibson, an associate professor of computer science. "And they likely
  will require up to hundreds of thousands of magnetic hard disks to
  handle the data required to run simulations, provide
  checkpoint/restart fault tolerance and store the output of these
  modeling experiments.
  
  "With such a large number of components, it is a given that some
  component will be failing at all times," he said.
  
  Today's supercomputers, which perform trillions of calculations each
  second, suffer failures once or twice a day, said Gary Grider, a
  co-principal investigator at the Los Alamos National Laboratory. Once
  supercomputers are built out to the scale of multiple petaflops, he
  said, the failure rate could jump to once every few minutes. Petascale
  data storage systems will thus require robust designs that can
  tolerate many failures, mask the effects of those failures and
  continue to operate reliably.
  
  "It's beyond daunting," Grider said of the challenge facing the new
  institute. "Imagine failures every minute or two in your PC and you'll
  have an idea of how a high performance computer might be crippled. For
  simulations of phenomena such as global weather or nuclear stockpile
  safety, we're talking about running for months and months and months
  to get meaningful results," he explained.
  
  Collaborating members in the Petascale Data Storage Institute
  represent a breadth of experience and expertise in data storage. "We
  felt we needed to bring the best and brightest together to address
  these problems that we don't yet know how to solve," said Grider,
  leader of Los Alamos' High Performance Computing Systems Integration
  Group.
  
  Carnegie Mellon and the University of California at Santa Cruz are the
  two leading academic centers for storage systems research, while the
  University of Michigan is a leader in network file systems. All three
  universities have sizable government and industrial collaborations.
  
  Los Alamos, a national security lab, and Oak Ridge National Laboratory
  are both in the process of building petaflop supercomputers, while a
  third member, Sandia National Laboratories, is another national
  security lab that recently built a leadership-class supercomputer.
  Both remaining members, the National Energy Research Scientific
  Computing Center at Lawrence Berkeley National Laboratory and Pacific
  Northwest National Laboratory, provide supercomputing resources for a
  diverse array of scientists.
  
  The data storage institute will focus its efforts in three areas:
  collecting field data about computer failure rates and application
  behaviors, disseminating knowledge through best practices and
  standards, and developing innovative system solutions for managing
  petascale data storage. The latter category could include so-called
  "self-star" systems that use computers to manage computers.
  
  The Petascale Data Storage Institute is part of the DOE's Scientific
  Discovery Through Advanced Computing program, which develops new tools
  and techniques for computational modeling and simulations. It is
  funded by a grant from the DOE Office of Science Programs.
  
  For more information, visit the Petascale Data Storage Institute Web
  site at <http://www.pdl.cmu.edu/PDSI/> and the Scientific Discovery
  Through Advanced Computing site at <http://www.scidac.gov/>.
  
  -----
  
  Source: DOE/Carnegie Mellon University

-- 
Dragomir R. Radev                    Associate Professor
SI, CSE, Ling                     U. Michigan, Ann Arbor 
http://www.eecs.umich.edu/~radev         radev@umich.edu              



This archive was generated by hypermail 2b30 : Tue Jun 09 2009 - 05:00:13 EDT