NSF Programs in High Performance Computing

15 slides
0.08 MB

Similar Presentations

Presentation Transcript


NSF and Science of Design Avogadro Scale Engineering Center for Bits & Atoms November 18-19, 2003 Kamal Abdali Computing & Communication Foundations Division Computer & Information Sciences Directorate National Science Foundation


Social, Behavioral, and Economic Sciences National Science Foundation Inspector GeneralNational Science BoardDirector Deputy Director Staff OfficesComputer & Info. Science & Engineering EngineeringGeosciencesMathematical & Physical SciencesEducation & Human Resources Budget, Finance & Award Management Information Resource Management Biological Sciences


Reorganized CISE StructureDivisions Administrative units based on intellectual partitions Clusters Comprehensive activity within a Division in a coherent area of research and education Teams of Program Officers and staff working closely with the community Themes Focused areas of research and education that cut across clusters and divisions Address scientific and national priorities Have program announcements and funds


The New CISE and Cross Cutting Themes e.g. Cyber Trust or Science of Design


Computing and Communication Foundations (CCF)Formal and Mathematical Foundations Core computing & communication theory Algorithmic & computational science Application-specific theory Foundations of Computing Processes & Artifacts Software design & productivity High-end software, architecture & design Computer graphics & visualization Emerging models for technology and computation Biologically motivated computing models Quantum computing & communication Computing & communication systems based on nano


Formal & Mathematical Foundations Determine inherent limits of computation and communication, and obtain optimal solutions within those limits. Investigate information representation methods, algorithms, and computational techniques for advancing information technology as well as all scientific and engineering disciplines. Representative Topics of Interest theory of computation and algorithms, algorithmic & computational approaches to mathematics & science, information coding and communication


Foundations of Computing Processes & Artifacts Advance the science, formalisms, and methodologies for building computing and communication systems. Representative Topics of Interest software engineering, programming language design and implementation, computer architecture and design, design test and automation techniques, graphics and visualization


Emerging Models and Technologies for Computation Explore computational models, techniques, and systems based on emerging and future technologies. Representative Topics of Interest Computing systems based on nanotechnology, quantum computing and communication, computational devices and architectures inspired by processing of information in living matter, computational approaches to problems in biology


Support of Design at NSFNearly all engineering programs Design and Manufacturing Innovation Division in ENG Computing Processes & Artifacts cluster in CCF (for software and hardware design, test & design automation tools, …) Science of Design Theme in CISE (being formulated--a workshop held during Nov 2-4)


MotivationChallenges in computing & communication Emerging technologies Nanoscale systems Alternative computing paradigms (e.g., DNA-based, quantum, chemical) Mobile computing & communication Distributed sensors Computational issues Increasing complexity of scientific and engineering problems Massively distributed data Trust and security Creating need and opportunity to design, build and maintain systems which are larger and more complex than existing ones by several orders of magnitude


Design as Mathematical Problem SolvingScientific knowledge base for design domain Ability to express design desiderata as a mathematical problem Available solution techniques, algorithmic and heuristic Optimization techniques needed since usually many solutions feasible


Engineering practice to analyze and synthesize complex systems impose the idea of architecture (functional organization as intercommunicating blocks) use hierarchical decomposition and multiple layers of abstraction different mathematical models, languages and formalisms, to represent component interactions at different layers E.g. Shannon's use of Boolean algebra to analyze relay circuits Bell's (and others') idea of high-level hardware design languages Conway-Mead VLSI design methodology.


Benefits of That ApproachEnables design of the most complex human-made systems: computer hardware, networks, databases, software E.g., VLSI designers can create without being bogged down by complexities of low-level component interactions, yet produce designs that exploit the electronics, physics, and material science involved in components


De-layered designOpposite of traditional practice More efficient design since cross-layer relationships can be exploited Sometimes the two practices can be mixed, combining humans’ layered and local design work with automated cross-layer and global optimization


Random List of IssuesUtilitarian and esthetic aspects. Domain-independent design principles? Coping with complexity of systems approaching Avogadro scale Aggregation and statistical characterization Exploiting sparsity, regularity and structure Self-diagnosis, self-repair, self regeneration, evolving Learning from nature Special problems for software and software-intensive systems Scientific principles, engineering practices, standard components, design automation, verification and test tools Requirement and specification, derivation of design from these Robust, reliable, fault-tolerant design Maintenance, built-in verification. Quality assurance Education and work force developments

Browse More Presentations

Last Updated: 8th March 2018

Recommended PPTs