Transcript Document

High Productivity Computing
Systems
Robert Graybill
DARPA/IPTO
March 2003
High Productivity
Computing Systems
Goal:
 Provide a new generation of economically viable high productivity computing
systems for the national security and industrial user community (2007 – 2010)
Impact:
 Performance (time-to-solution): speedup critical national
security applications by a factor of 10X to 40X
 Programmability (time-for-idea-to-first-solution): reduce
cost and time of developing application solutions
 Portability (transparency): insulate research and
operational application software from system
 Robustness (reliability): apply all known techniques to
protect against outside attacks, hardware faults, &
programming errors
HPCS Program Focus Areas
Applications:
 Intelligence/surveillance, reconnaissance, cryptanalysis, weapons analysis, airborne contaminant
modeling and biotechnology
Fill the Critical Technology and Capability Gap
Today (late 80’s HPC technology)…..to…..Future (Quantum/Bio Computing)
RBG 10/9/2002
2
Vision: Focus on the Lost Dimension of HPC –
“User & System Efficiency and Productivity”
Parallel Vector
Systems
1980’s
Technology
Vector
Tightly Coupled
Parallel Systems
Commodity HPCs
2010
High-End
Computing Solutions
Moore’s Law
Double Raw
Performance every
18 Months
New Goal:
Double Value Every
18 Months
Fill the high-end computing technology and capability gap
for critical national security missions
RBG 10/9/2002
3
HPCS Technical Considerations
Communication
Programming
Models
Shared-Memory
Multi-Processing
Distributed-Memory
Multi-Computing
“MPI”
Architecture Types
Custom Vector
Parallel
Vector
Scalable
Vector
Vector
Supercomputer
Microprocessor
Symmetric
Multiprocessors
Distributed Shared
Memory
Massively
Parallel
Processors
Commodity
Clusters, Grids
HPCS Focus
Tailorable Balanced Solutions
Performance
Characterization
& Precision
Programming
Models
System
Architecture
Software
Technology
Hardware
Technology
Commodity
HPC
Single Point Design Solutions are no longer
Acceptable
RBG 10/9/2002
4
HPCS Program Phases I - III
Metrics,
Metrics
and
Benchmarks
Benchmarks
Early
Software
Academia
Research
Early
Pilot
Tools
Platforms
Platforms
HPCS
Capability or
Products
Application
Analysis
Performance
Assessment
Products
Requirements
and Metrics
Concept
Reviews
System
Design
Review
Research
Prototypes
& Pilot Systems
Technology
Assessments
PDR
DDR
Industry
Industry Evolutionary
Development Cycle
Phase II
Readiness Reviews
Fiscal Year
02
Phase III Readiness Review
03
04
05
06
07
08
09
Reviews
Industry Procurements
Critical Program
Milestones
RBG 10/9/2002
Phase I
Industry
Concept Study
Phase III
Full Scale Development
Phase II
R&D
5
10
HPCS Phase I Industry Teams
Industry:
Cray, Inc. (Burton Smith)
Hewlett-Packard Company (Kathy Wheeler)
International Business Machines Corporation
(Mootaz Elnozahy)
Silicon Graphics, Inc. (Steve Miller)
Sun Microsystems, Inc. (Jeff Rulifson)
Application Analysis/Performance Assessment Team:
MIT Lincoln Laboratory
RBG 10/9/2002
6
Application Analysis/
Performance Assessment
Activity Flow
Inputs
DDR&E
& IHEC
Mission
Analysis
Mission
Partners:
DOD
DOE
NNSA
NSA
NRO
Application Analysis
Benchmarks & Metrics
Impacts
HPCS Applications
Common
Critical
Kernels
Participants
HPCS Technology
Drivers
Compact
Applications
Define System
Requirements and
Characteristics
1. Cryptanalysis
2. Signal and Image
Processing
3. Operational Weather
4. Nuclear Stockpile
Stewardship
5. Etc.
Applications
Mission-Specific
Roadmap
Mission Work Flows
Productivity
Mission Partners
Ratio of
Utility/Cost
Improved Mission
Capability
Metrics
- Development time
(cost)
Participants:
Cray HP
IBM SGI
Sun
RBG 10/9/2002
- Execution time
(cost)
Implicit Factors
7
DARPA
HPCS Program
Motivation
Application Focus Selection
DDR&E Study
•
• Operational weather and ocean
forecasting
•
• Planning activities for dispersion
of airborne/waterborne
•
contaminants
•
• Cryptanalysis
• Intelligence, surveillance,
•
reconnaissance
•
• Improved armor design
• Engineering design of large
•
aircraft, ship and structures
•
• National missile defense
•
• Test and evaluation
• Weapon (warheads and
•
penetrators)
•
• Survivability/stealth design • Bioscience
RBG 10/9/2002
8
IHEC Study
Comprehensive Aerospace
Vehicle Design
Signals Intelligence (Crypt)
Signals Intelligence (Graph)
Operational Weather/Ocean
Forecasting
Stealthy Ship Design
Nuclear Weapons Stockpile
Stewardship
Signal and Image Processing
Army Future Combat Systems
Electromagnetic Weapons
Development
Geospatial Intelligence
Threat Weapon Systems
Characterization
Biomedical Computing Requirements
Computational Biology: from Sequence to Systems
Sequence Genome
TeraOps
Assemble Genome
Trivially Parallel
Find the Genes
Annotate the Genes
1
Map Genes to Proteins
10
Protein-Protein Interactions
Pathways: Normal & Aberrant
100
Protein Functions in Pathways
Protein Structure
Slide provided by IDC
RBG 10/9/2002
9
1000
Identify Drug Targets
Peta-Scale
Cellular Response
Computing
Tissue, Organ &
Whole Body Response
HPCS Mission Work Flows
Overall Cycle
Development Cycle
Theory
Researcher
Code
Days to
hours
Hours to
minutes
Prototyping
Test
Design
Development
Execution
Port Legacy
Software
Port Legacy
Software
Enterprise
Months
to days
Months
to days
Simulation
Orient
Observe
Production
Optimize
Development
Prototyping
Design
Test
Scale
Design
Initial Product
Development
Years to
months
Hours to
Minutes
(Response Time)
Act
Code
Initial
Development
Experiment
Visualize
Design
Evaluation
Code
Test
Maintenance
Operation
Port, Scale,
Optimize
Decide
HPCS Productivity Factors: Performance, Programmability,
Portability, and Robustness are very closely coupled with each work flow
RBG 10/9/2002
10
Workflow Priorities & Goals
•
•
Workflows define scope of
customer priorities
Activity and Purpose benchmarks
will be used to measure
Productivity
HPCS Goal is to add value to
each workflow
– Increase productivity while
increasing problem size
RBG 10/9/2002
System
Requirements
Workstation
Productivity
•
Mission
Needs
Cluster
Researcher
Workflow
Researcher
Enterprise
Production
Implicit Productivity Factors
Perf.
Prog. Port.
Robust.
High
High
High
High
High
High
High
Production
HPCS
Problem Size
11
HPCS Productivity Framework
Activity &
Purpose
Benchmarks
Execution
Time (cost)
Productivity
(Ratio of
Utility/Cost)
Work
Flow
Actual
Productivity
Metrics
System
or
Model
Development
Time (cost)
System Parameters
(Examples)
BW bytes/flop
Memory latency
Memory size
……..
Processor flop/cycle
Bisection BW
Total Connections
………
Size (cuft)
Power/rack
Facility operation
……….
Code size
Restart time (reliability)
Code Optimization time
………
Implicit HPCS Productivity Factors:
Performance, Programmability, Portability, and Robustness
RBG 10/9/2002
12
HPC Community Reactions
• DoD and DOE User Communities
–
–
–
–
Active participation in reviews
Providing challenge problems
Linking with internal efforts
Providing funding synergism
• Industry
– Finally an opportunity to develop a non evolutionary vision
– Active program support (technical, personnel, vision)
– Direct impact to future product roadmaps
• University
– Active support for Phase 1 (2X growth from proposals)
• Extended Community
– HPCS strategy embedded in Congressional IHEC Report
Productivity a new HPC Sub-discipline
RBG 10/9/2002
13