1. Our report on the F U T U R E Dennis Gannon, Andy Cleary, Jack Dongarra, Steve Tueke, Sameer Shende, Paul Boggs, Subash Saini, Sonia Sachs

2. Timeline for funding levels DOE 2000 Basic CS research Deployable Technology Pilots DOE2000 becomes base funding Basic CS Research Advanced development Deployment/Hardening/Support Combustion Climate other basic science apps SSI

3. Collaboration Services zTwo classes of important activities. yOne class is deployable and is beginning to merge with new commercial technology, e.g. xCollaboration management xSome of the security work yOther class MUST continue as core research that pushes the envelope in terms of performance, capability and scalability. xMulticast protocol work, multi-rate streams and shared VR.

4. Notebooks zSeeing emerging interoperability standard is very exciting. zFuture research directions ysupport next-gen input devices (PDAs, tablets, voice) ysync/search of handwritten notes & voice annotations ybetter support for collaborative authoring ynon-wimp user interfaces yintegration with mathematical environments yinterface to distributed resources yinterface to metadata services/data mining.

5. Scientific Frameworks zESI &PETSc show how to build linear algebra tools from interchangeable parts. zPooma does this for field based simulations zHow do we do this with other scientific frameworks yhow do we build high level packages/frameworks by building on the results of recent pairwise integration efforts. xProjects like Overture provide a good example. xHow do we do more integration of symbolic and numerical systems?

6. Runtime Frameworks zHow do we get more than 1% of our Tera-flop resources? yHow do we make it available to users? xAnalytical and dynamic performance modeling and Analysis. xPerformance analysis tools that provide info at multiple levels of program abstraction. xDynamic and static restructuring tools for memory hierarchy management/ thread scheduling. xOptimization for multi-disciplinary/component based apps. xLoad balancing for dynamic applications

7. Component Technologies zCCA needs to demonstrate that a substantial multi-lab framework can be built to yaddress applications based on commercial technology xe.g. show how to build a multi-agent framework for data analysis that can be used with/by the notebooks and collaborative tools. yProve that this can be done while addressing the needs of the highest performance DOE testbeds. yuse a scientific computation component library in a significant application

8. Grids zTwo classes of activities yStandardize and deploy initial grid services. xBasic info service, authentication, scheduling yConduct basic grid research xExample: Information Services Component/Resource discovery for CCA Authentication and allocation services Grid scheduling MetaData Services and protocols. Schemas/DTD for other services and resources. xOther areas: Authorization, Accounting, Scheduling, Communication, etc.

9. Interactions with other agencies zNSF - NextGen software program is yfocused on basic CS research on performance and optimization in high performance heterogeneous environments. zPACI Centers. yA completely compatible research agenda. zNASA Information Power Grid yPlatform for demonstrating the power of DOE technology on apps from astrobiology, aerospace industry, spacestation, etc. yAlready committed to working with DOE.