SDSCNPACI Overview NPACI Parallel Computing Institute August 19, 2002

1
SDSC/NPACI OverviewNPACI Parallel Computing
InstituteAugust 19, 2002

• Richard Moore
• NPACI Executive Director
• San Diego Supercomputer Center
• rlm_at_sdsc.edu

2
Welcome to SDSC and NPACI

• SDSC is the Leading Edge Site for the National
  Partnership for Advanced Computational
  Infrastructure (NPACI)

• UCSDs San Diego Supercomputer Centers mission
  is to develop and use technology to advance
  science.

3
Computing Today NSFs Cyberinfrastructure
4
Focus for activities SDSC Programs
Networking
5
NPACI A National Partnership

• NPACI is developing the human, software, data,
  networking and computational infrastructure to
  address critical science and technology needs
• 48 Partner Institutions
• 21 states CA, TX, MI, MD, KN, NM, MS, MT, NY,
  OH, OR, NJ, TN, VA, MO, WI, CO, AZ, MA, WA, PA
• Italy, Australia, Spain, Sweden
• Hundreds of participating users, developers,
  scientists, students

6
Science that Makes a Difference

• NPACI human, software and hardware infrastructure
  is key to enabling the next generation of
  advances in technology, science, and engineering.

Deep Impact
Broad Impact
Individual Impact
7
2002 NPACI Leadership Team

• Susan Graham, UC Berkeley Chief Computer
  Scientist
• Peter Taylor, SDSCChief Applications Scientist
• Wayne Pfeiffer, SDSCResources Lead
• Greg Moses, U WisconsinEducation, Outreach, and
  Training Thrust Lead

Francine Berman, SDSCPI Director Richard
Moore, SDSC Executive Director Carl Kesselman,
USC/ISI, Chief Software Architect
8
2002 NPACI Executive Committee
Rich Wolski, UC Santa Barbara Metasystems/Grid
Computing Joel Saltz, Ohio State Programming
Tools and Environments Reagan Moore,
SDSCData-Intensive Computing Arthur Olson,
TSRIInteraction Environments William Martin, U
MichiganResource Representative Jim Pool,
Caltech Resource Representative
Helen Berman, Rutgers Molecular Science Mark
Ellisman, UCSDNeuroscience Cherri Pancake,
Oregon State Earth Systems Science Tinsley Oden,
U TexasEngineering Gwen Jacobs, Montana
StateMember at Large Aron Kuppermann,
Caltech User Representative
9
NPACI Is Organized by Thrusts Goal of all
thrusts is to develop and deploy infrastructure
EOT
APPLICATIONS
TECHNOLOGIES
Molecular Science Neuroscience Earth Systems
Science Engineering
Grid Computing Programming Tools
Environments Data-Intensive Computing Interaction
Environments
RESOURCES
10
NPACI allocable resources are at 4 sites
U MichiganIBM SPsAMD Athlon ClusterADSM Storage
CaltechHP V2500sHPSS Storage
U TexasIBM p690 HPCs Cray T3ECray SV1 DMF
Storage
SDSCIBM SP Blue HorizonSun HPC 10000HPSS
Storage
11
Complementary rolesof NPACI resource sites

• Leading-edge site (UCSD/SDSC)
• Very high-performance resources Blue Horizon, an
  IBM SP, the first teraflops system (in 2000) for
  U.S. academic community
• Very large amounts of data 370 terabytes as of
  May 2002
• Major network hub one of four initial TeraGrid
  partners
• Caltech
• Alternate architecture systems from HP (being
  phased out)
• One of four initial TeraGrid partners
  data-intensive apps
• U Texas U Michigan mid-range sites
• Allocable IBM Cray systems smaller than at LES
• Training, documentation, consulting help
• New IBM p690 HPCs at UT AMD cluster at UM
• Testbeds for grid-based computing such as in
  TeraGrid

12
NPACI allocable compute resourcesas of May 2002
13
NPACI Vision Software Roadmap

• Goal Deliver SW which is usable and useful, and
  promotes maximal use of the underlying resources
• SW development efforts focus on
• Minimizing the impact of heterogeneity (e.g. SRB,
  NWS, Globus)
• Coordination across remote resources (e.g. APST,
  NetSolve, Globus)
• Usability (e.g. NPACI Rocks, mySRB, GridPort,
  HotPage)
• Greater access to high-end and geographically
  distributed resources (e.g. TeraGrid, portals,
  NPACI HotPage, GridPort)
• Maximizing the impact of the resources (e.g.,
  NMI, Catalina scheduler, Scalable Vis Toolkits)

14
NPACI Grid modeled after evolving Community Grid
Model

• Roll your own software but agree on interfaces,
  service architecture, standards

15
The Beginning of Cyberinfrastructure PACI and
TeraGrid
August 9, 2001 NSF Awarded 53,000,000 to
SDSC/NPACI and NCSA/Alliance for TeraGrid

• TeraGrid will provide in aggregate
• 13.6 trillion calculations per second
• Over 600 trillion bytes of immediately accessible
  data
• 40 gigabit per second network speed
• TeraGrid will provide a new paradigm for
  data-oriented computing
• Critical for disaster response, genomics,
  environmental modeling,

16
TeraGrid 13.6 TF, 6.8 TB memory, 79 TB internal
disk, 576 network disk
ANL 1 TF .25 TB Memory 25 TB disk
Caltech 0.5 TF .4 TB Memory 86 TB disk
Extreme Blk Diamond
574p IA-32 Chiba City
256p HP X-Class
32
32
32
32
24
128p Origin
128p HP V2500
32
24
32
24
HR Display VR Facilities
92p IA-32
5
4
5
8
8
HPSS
HPSS
OC-48
NTON
OC-12
Calren
ESnet HSCC MREN/Abilene Starlight
Chicago LA DTF Core Switch/Routers Cisco 65xx
Catalyst Switch (256 Gb/s Crossbar)
Juniper M160
OC-48
OC-12 ATM
OC-12
GbE
NCSA 8 TF 4 TB Memory 240 TB disk
SDSC 4.1 TF 2 TB Memory 225 TB SAN
Juniper M40
Juniper M40
OC-12
vBNS Abilene Calren ESnet
OC-12
OC-12
OC-3
Myrinet
8
4
UniTree
8
HPSS
2
Sun Server
Myrinet
4
1024p IA-32 320p IA-64
1176p IBM SP Blue Horizon
16
14
4
15xxp Origin
Sun E10K