Transcript Incremental Communication

Advanced Computational Research
Laboratory (ACRL)
Virendra C. Bhavsar
Faculty of Computer Science
University of New Brunswick
Fredericton, NB, E3B 5A3
Canada
OUTLINE
 ACRL Research Groups
 Introduction to Parallel Processing
 ACRL Research Groups
 Conclusion
ARCL
Advanced Computational Research
Laboratory
 High Performance Computational
Problem-Solving Environment and
Visualization Environment
 Computational Experiments in
multiple disciplines: Computer Science,
Science and Engineering
 Located in the Information
Technology Center (ITC)
ACRL: Researchers and Groups
Faculty of Computer Science
 Artificial Intelligence Group
- Dr. Spencer, Dr. Nickerson
 Parallel/Distributed Processing
Group
- Dr. Bhavsar, Dr. Du, Dr. Ghorbani
Dr. Kaser, Dr. Shaw
 Computational Geometry Group
- Dr. Bremner, Dr. Itturiaga
 Automated Reasoning Group
- Dr. Spencer, Dr. Horton
 Bioinformatics Group
ACRL: Researchers and Groups
Faculty of Science
 Physics
- Dr. Hamza (plasma physics,
ionospehere, solar corona)
Dr. Balcolm (magnetic resonance
Imaging)
Dr. Xu (methanol to gasoline
process)
 Chemistry
- Dr. Thakkar (optical computing
materials)
Dr. Grein (ozone related reactions)
Dr. Mattar (cancer drugs, fisheries)
 Bioinformatics Group
ACRL: Researchers and Groups
Faculty of Engineering
 Mechanical Engineering
Dr. Hussein (threat-material
detection)
Dr. Sousa ( fire propagation, CFD)
Dr. Biden (artificial limbs)
 Chemical Engineering
Dr. Bendrich (plastics
manufacturing)
 Electrical Engineering
Dr. Chang (electrical machines
Forestry and Environment Management
 New CFI Application
Scientific Computation
Parallel Computing
• Parallel computing
- simultaneous use of multiple
compute resources to solve a
computational problem
• Why Parallel Computing?
- to save time (wall clock time)
- to solve larger problems
- to alleviate memory
constraints
- larger databases
Parallel Computing
• Grand Challenge Problems”
- weather and climate
- mechanical devices - from
prosthetics to spacecraft
- electronic circuits
- manufacturing processes
- geological, seismic activity
- biological, human genome
- chemical and nuclear reactions
Parallel Computing
• Commercial applications
- parallel databases, data mining
- oil exploration
- computer-aided diagnosis in
medicine
- management of national and multinational corporations
- advanced graphics and virtual
reality, particularly in the
entertainment industry
- networked video and multi-media
technologies
- collaborative work environments
Parallel Computing
Ultimately, parallel computing is an
attempt to maximize the infinite but
seemingly scarce commodity called
time
IBM SP
Shared Memory Model
• Quad-Processor System
Distributed Memory Model
Hybrid Model
• Similar to IBM SP
ARCL
Advanced Computational Research
Laboratory
 High Performance Multiprocessor
(16-processor) System with
24 GFLOPS (peak) performance with
72 GB internal disk storage and
109.2 GB external disk storage
 Software for Computational Studies
and Visualization
 Parallel Programming tools
 E-Commerce Software, including
datamining software
ARCL
Nodes
• 4 Compute Nodes: total of 16 processors.
Switch
• 300 MB/sec bi-directional
• 1.2 µsec latency
ARCL
Node
• 2 x 2-way 375 Mhz POWER3 64-bit
Winterhawk II Processor Cards
• 258 MB Memory (1 GB total)
• 2 x 9.1 GB Ultra-SCSI Disk Drives
• 10/100 Mbit Ethernet Adapter
•Gigabit Ethernet Card
MIMD Processing
• Multiple Instruction Stream Multiple
Data Stream Model
Array Processing
Threads
Message Passing Model
• Example - MPI
Data Parallel Model
Domain Decomposition
Domain Decomposition
Functional Decomposition
Inter-Process Communication
Load Balancing
Monte Carlo Method
Heat Equation
Heat Equation
Conclusion
Future Workshops
Feb. 13, 2001: Parallel Prog.
Workshop
Feb 24, 2001: AC3 Workshop
Feb. 26-27, 2001: IBM
Workshop
- Visualization using Open DX
- Atlantic Canada High
Performance Computing
Workshop
-HPCS’2001 at Windsor, ON
June 18-20, 2001