"... In PAGE 7: ...Table4 : Industrial HPCC Applications 1 to 5: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 1 Computational PDE, FEM SIMD, MIMD for Fluid Dynamics Turbulence irregular adaptive Aerospace Mesh Generation HPF(+) but Military, Civilian Unclear for Vehicles adaptive irregular Propulsion mesh 2 Structural PDE, FEM MIMD as complex Dynamics Dominated by geometry Vendor Codes such HPF(+) as NASTRAN 3 Electromagnetic PDE solved by Simulation moment method SIMD Antenna Design Matrix solve HPF Stealth Vehicles dominates Noise in high Newer FEM and SIMD, MIMD frequency circuits FD Methods? HPF(+) Mobile Phones Also fast multipole 4 Scheduling Expert Systems MIMD Manufacturing and/or (unclear Speedup) Transportation AsyncSoft (Dairy delivery to Neural Networks SIMD Military deployment) Simulated annealing HPF University Classes Linear Programming MIMD Airline Scheduling (hard sparse matrix) HPF+? of crews, planes in static or dynamic (Syracuse snow storm) cases 5 Environmental PDE, FD, FEM SIMD but Modeling| Sensitivity to MIMD for Earth/Ocean/Atmos- Data irregular adaptive pheric Simulation mesh HPF(+) except this unclear for adaptive irregular mesh... In PAGE 8: ...Table4 : Industrial HPCC Applications 6 to 10: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 6 Environmental Empirical Models Some SIMD but Phenomenology Monte Carlo and MIMD more |Complex Systems, Histograms natural (Lead Concentration HPF in blood) 7 Basic Chemistry Calculate Matrix Probably SIMD Chemical Potentials Elements with perhaps Elemental Reaction Matrix Eigenvalue SIMD possible Dynamics determination, HPF Inversion, Multiplication 8 Molecular Dynamics Particle Dynamics HPF(+) except in Physics amp; Chemistry with irregular need MPF for Biochemistry cuto forces fast multipole Discrete Simulation Fast Multipole Monte Carlo for Methods CFD (DSMC) Mix of PDE and Particle in the Cell Particle methods (PIC) in PIC and DSMC 9 Economic Modelling Single nancial Real Time instrument by SIMD, HPF Optimization Monte Carlo Mortgaged backed Full Simulations MIMD or SIMD Securities of complete with Integration Option Pricing portfolios Software 10 Network Simulations Sparse matrices MIMD Electrical Circuit Zero structure HPF for matrix Microwave and VLSI de ned by elements Biological (neural) connectivity MPF or library Circuit for matrix solve... In PAGE 10: ...Table4 : Industrial HPCC Applications 14 to 18: Information Analysis|\DataMining quot; Application Area Problem Machine Item and Examples Comments and Software 14 Seismic and Parallel Computers SIMD useful but Environmental already important MIMD might be Data Analysis but necessary No oil in New HPF York State 15 Image Processing Many commercial Metacomputer Medical Applications of Low Level Instruments Defense Technology Vision is EOS (mission to Component of many SIMD and HPF Planet Earth) \metaproblems quot; Medium/High Defense (Information Level Vision is Surveillance Integration category) MIMD and HPF(+) Computer Vision e.g.... In PAGE 11: ...Table4 : Industrial HPCC Applications 19 to 22 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand Application Problem Machine amp; Item Area Comments Structure Software 19 Transaction Database-most Embarrassingly MIMD Processing transactions short. Parallel Database ATM As add \value quot; (automatic this becomes teller machine) Informationintegration 20 Collaboration Research Center or Asynchronous High Speed Telemedicine doctor(s)|patient Network Collaboratory interaction without for Research regard to physical Education location Business 21 Text on Multimedia Embarrassingly MIMD Demand database Parallel Database Digital (existing) (see areas 22, 23) libraries Full text search ERIC Education database, United Nations- Worldwide newspapers 22 Video on Multimedia Embarrassingly MIMD Demand Database Parallel for Database Movies, News Interactive VCR, multiple Users Video Editing (CNN Newsource Video Browsing, Software amp; Newsroom), Link of Interesting Current cable, video and text parallel United Nations- database compression SIMD Policy Support compression... In PAGE 12: ...Table4 : Industrial HPCC Applications 23 to 24 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand Application Problem Machine amp; Item Area Comments Structure Software 23 Imagery on Multimedia Metaproblem MIMD but Demand database Embarrassingly much SIMD Kodak GIODE Image Parallel plus image \clip art quot; on Understanding for Loosely analysis demand Content searching Synchronous Medical images and (terrain) Image Satellite images medical feature Understanding identi cation 24 Simulation on Multimedia map Synchronous SIMD terrain Demand database terrain engine (parallel Education, Generalized rendering with rendering) Tourism, City ight simulator Asynchronous MIMD planning, Geographical Hypermedia database Defense Information Integration mission planning System software... In PAGE 13: ...Table4 : Information Integration Applications 25 to 28 These involve combinations of Information Production, analysis, Access and Dissemination and thus need the Integration of the various Software and Machines Architecture Issues discussed under previous application areas. Sometimes Called System of Systems 25: Military and Civilian Command and Control (C2, C3, C4I : : :) Battle Management, Command, Control, Communication, Intelligence and Surveillance (BMC3IS) Military Decision Support Crisis Management|Police and other Government Operations SIMNET simulates this and with people and computers in the loop has many of same issues 26 to 28: Applications of InfoVision Services Generalize Compuserve, Prodigy, America Online, Dialog and Other In- formation Services 26: Decision Support for Society Community Information Systems Travel and Generalized Yellow Page Services 27: Business Decision Support|One example is: Health Care with Image and Video databases supporting telemedicine 28: Public Administration and Political Decision Support Government Information Systems Maxwell School at Syracuse University teaches use of realtime video to aid world wide decisions (United Nations)... In PAGE 14: ...Table4 : Information Integration Applications 29 to 33 29: Real-Time Control Systems Robotics uses Imagery to make decisions (control vehicles) Energy management controls power use and generation 30: Electronic Banking Requires Security, Privacy, Electronic Cash, etc. 31: Electronic Shopping 32: Agile Manufacturing|Multidisciplinary Design and Concurrent Engineering Combines CAD with Applications 1 to 3 Requires major changes to Manufacturing Infrastructure and Approach 33: Education InfoMall Living Textbook|6 Schools on ATM network linked to HPCC InfoVision Servers at NPAC [Mills:95a] Table 1 describes the general guidelines used in organizing Table 4.... In PAGE 14: ... 29: Real-Time Control Systems Robotics uses Imagery to make decisions (control vehicles) Energy management controls power use and generation 30: Electronic Banking Requires Security, Privacy, Electronic Cash, etc. 31: Electronic Shopping 32: Agile Manufacturing|Multidisciplinary Design and Concurrent Engineering Combines CAD with Applications 1 to 3 Requires major changes to Manufacturing Infrastructure and Approach 33: Education InfoMall Living Textbook|6 Schools on ATM network linked to HPCC InfoVision Servers at NPAC [Mills:95a] Table 1 describes the general guidelines used in organizing Table4 . Note that we did not directly cover academic areas, and a more complete list (which included our industrial table) was produced by the Peta ops meeting [Peta:94a].... In PAGE 14: ... Here, \Info quot; refers to the information based application focus and \Mall quot; to the use of a virtual corporation (groups of \storeholders quot;) to produce the complex integrated applications enabled by HPCC. The rst column of Table4 contains the area label and some sample applications. Algorithmic and other comments are in column two.... In PAGE 15: ... Our Caltech work was mainly on the hypercube, but the total of 300 references used in original classi cation covered work on the Butter y, transputers, the SIMD Connection Machine, and DAP. We originally iden- ti ed three temporal structures and one especially important (as it was so simple) spatial structure, which are the rst four entries in Table4 . Chap- ter 3 of [Fox:94a] describes a \complex systems quot; approach to computation and introduces the spatial and temporal structure of problems and comput- ers.... In PAGE 16: ... This is important in determing the performance, as shown in Chapter 3 of [Fox:94a] of an implementation, but it does not a ect the broad software issues discussed here. In Table4 , we only single out one special spatial structure, \embarrassingly parallel, quot; where there is little or no connection between the individual parallel program components. For embarrassingly parallel problems, illustrated in Figure 4, the synchronization (both soft- ware and hardware) issues are greatly simpli ed.... In PAGE 23: ... Each asynchronous task is now a synchronous or loosely synchronous mod- estly parallel evaluation of a given chess position. There were a few examples mentioned above of metaproblems in our orig- inal survey, but a major part of Table4 , from our New York State activity,... In PAGE 34: ...4 Summary Each of three case studies illustrates how di erent applications and di erent numerical approaches to a given problem, lead to very di erent problem architectures and correspondingly the needed software support. Although our discussion is not complete, we do think that it is quite typical, and that a similar situation is seen in the other applications of Table4 , and summarized in the last two columns.... ..."

"... In PAGE 8: ...Table4 : Industrial HPCC Applications 1 to 5: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 1 Computational PDE, FEM SIMD, MIMD for Fluid Dynamics Turbulence irregular adaptive Aerospace Mesh Generation HPF(+) but Military, Civilian Unclear for Vehicles adaptive irregular Propulsion mesh 2 Structural PDE, FEM MIMD as complex Dynamics Dominated by geometry Vendor Codes such HPF(+) as NASTRAN 3 Electromagnetic PDE solved by Simulation moment method SIMD Antenna Design Matrix solve HPF Stealth Vehicles dominates Noise in high Newer FEM and SIMD, MIMD frequency circuits FD Methods? HPF(+) Mobile Phones Also fast multipole 4 Scheduling Expert Systems MIMD Manufacturing and/or (unclear Speedup) Transportation AsyncSoft (Dairy delivery to Neural Networks SIMD Military deployment) Simulated annealing HPF University Classes Linear Programming MIMD Airline Scheduling (hard sparse matrix) HPF+? of crews, planes in static or dynamic (Syracuse snow storm) cases 5 Environmental PDE, FD, FEM SIMD but Modeling| Sensitivity to MIMD for Earth/Ocean/Atmos- Data irregular adaptive pheric Simulation mesh HPF(+) except this unclear for adaptive irregular mesh... In PAGE 9: ...Table4 : Industrial HPCC Applications 6 to 10: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 6 Environmental Empirical Models Some SIMD but Phenomenology Monte Carlo and MIMD more |Complex Systems, Histograms natural (Lead Concentration HPF in blood) 7 Basic Chemistry Calculate Matrix Probably SIMD Chemical Potentials Elements with perhaps Elemental Reaction Matrix Eigenvalue SIMD possible Dynamics determination, HPF Inversion, Multiplication 8 Molecular Dynamics Particle Dynamics HPF(+) except in Physics amp; Chemistry with irregular need MPF for Biochemistry cuto forces fast multipole Discrete Simulation Fast Multipole Monte Carlo for Methods CFD (DSMC) Mix of PDE and Particle in the Cell Particle methods (PIC) in PIC and DSMC 9 Economic Modelling Single nancial Real Time instrument by SIMD, HPF Optimization Monte Carlo Mortgaged backed Full Simulations MIMD or SIMD Securities of complete with Integration Option Pricing portfolios Software 10 Network Simulations Sparse matrices MIMD Electrical Circuit Zero structure HPF for matrix Microwave and VLSI de ned by elements Biological (neural) connectivity MPF or library Circuit for matrix solve... In PAGE 11: ...Table4 : Industrial HPCC Applications 14 to 18: Information Analysis|\DataMining quot; Application Area Problem Machine Item and Examples Comments and Software 14 Seismic and Parallel Computers SIMD useful but Environmental already important MIMD might be Data Analysis but necessary No oil in New HPF York State 15 Image Processing Many commercial Metacomputer Medical Applications of Low Level Instruments Defense Technology Vision is EOS (mission to Component of many SIMD and HPF Planet Earth) \metaproblems quot; Medium/High Defense (Information Level Vision is Surveillance Integration category) MIMD and HPF(+) Computer Vision e.g.... In PAGE 12: ...Table4 : Industrial HPCC Applications 19 to 22 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand Application Problem Machine amp; Item Area Comments Structure Software 19 Transaction Database-most Embarrassingly MIMD Processing transactions short. Parallel Database ATM As add \value quot; (automatic this becomes teller machine) Informationintegration 20 Collaboration Research Center or Asynchronous High Speed Telemedicine doctor(s)|patient Network Collaboratory interaction without for Research regard to physical Education location Business 21 Text on Multimedia Embarrassingly MIMD Demand database Parallel Database Digital (existing) (see areas 22, 23) libraries Full text search ERIC Education database, United Nations- Worldwide newspapers 22 Video on Multimedia Embarrassingly MIMD Demand Database Parallel for Database Movies, News Interactive VCR, multiple Users Video Editing (CNN Newsource Video Browsing, Software amp; Newsroom), Link of Interesting Current cable, video and text parallel United Nations- database compression SIMD Policy Support compression... In PAGE 13: ...Table4 : Industrial HPCC Applications 23 to 24 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand Application Problem Machine amp; Item Area Comments Structure Software 23 Imagery on Multimedia Metaproblem MIMD but Demand database Embarrassingly much SIMD Kodak GIODE Image Parallel plus image \clip art quot; on Understanding for Loosely analysis demand Content searching Synchronous Medical images and (terrain) Image Satellite images medical feature Understanding identi cation 24 Simulation on Multimedia map Synchronous SIMD terrain Demand database terrain engine (parallel Education, Generalized rendering with rendering) Tourism, City ight simulator Asynchronous MIMD planning, Geographical Hypermedia database Defense Information Integration mission planning System software... In PAGE 14: ...Table4 : Information Integration Applications 25 to 28 These involve combinations of Information Production, analysis, Access and Dissemination and thus need the Integration of the various Software and Machines Architecture Issues discussed under previous application areas. Sometimes Called System of Systems 25: Military and Civilian Command and Control (C2, C3, C4I : : :) Battle Management, Command, Control, Communication, Intelligence and Surveillance (BMC3IS) Military Decision Support Crisis Management|Police and other Government Operations SIMNET simulates this and with people and computers in the loop has many of same issues 26 to 28: Applications of InfoVision Services Generalize Compuserve, Prodigy, America Online, Dialog and Other In- formation Services 26: Decision Support for Society Community Information Systems Travel and Generalized Yellow Page Services 27: Business Decision Support|One example is: Health Care with Image and Video databases supporting telemedicine 28: Public Administration and Political Decision Support Government Information Systems Maxwell School at Syracuse University teaches use of realtime video to aid world wide decisions (United Nations)... In PAGE 15: ...Table4 : Information Integration Applications 29 to 33 29: Real-Time Control Systems Robotics uses Imagery to make decisions (control vehicles) Energy management controls power use and generation 30: Electronic Banking Requires Security, Privacy, Electronic Cash, etc. 31: Electronic Shopping 32: Agile Manufacturing|Multidisciplinary Design and Concurrent Engineering Combines CAD with Applications 1 to 3 Requires major changes to Manufacturing Infrastructure and Approach 33: Education InfoMall Living Textbook|6 Schools on ATM network linked to HPCC InfoVision Servers at NPAC [Mills:95a] Table 1 describes the general guidelines used in organizing Table 4.... In PAGE 15: ... 29: Real-Time Control Systems Robotics uses Imagery to make decisions (control vehicles) Energy management controls power use and generation 30: Electronic Banking Requires Security, Privacy, Electronic Cash, etc. 31: Electronic Shopping 32: Agile Manufacturing|Multidisciplinary Design and Concurrent Engineering Combines CAD with Applications 1 to 3 Requires major changes to Manufacturing Infrastructure and Approach 33: Education InfoMall Living Textbook|6 Schools on ATM network linked to HPCC InfoVision Servers at NPAC [Mills:95a] Table 1 describes the general guidelines used in organizing Table4 . Note that we did not directly cover academic areas, and a more complete list (which included our industrial table) was produced by the Peta ops meeting [Peta:94a].... In PAGE 15: ... Here, \Info quot; refers to the information based application focus and \Mall quot; to the use of a virtual corporation (groups of \storeholders quot;) to produce the complex integrated applications enabled by HPCC. The rst column of Table4 contains the area label and some sample applications. Algorithmic and other comments are in column two.... In PAGE 16: ... Our Caltech work was mainly on the hypercube, but the total of 300 references used in original classi cation covered work on the Butter y, transputers, the SIMD Connection Machine, and DAP. We originally iden- ti ed three temporal structures and one especially important (as it was so simple) spatial structure, which are the rst four entries in Table4 . Chap- ter 3 of [Fox:94a] describes a \complex systems quot; approach to computation and introduces the spatial and temporal structure of problems and comput- ers.... In PAGE 17: ... This is important in determing the performance, as shown in Chapter 3 of [Fox:94a] of an implementation, but it does not a ect the broad software issues discussed here. In Table4 , we only single out one special spatial structure, \embarrassingly parallel, quot; where there is little or no connection between the individual parallel program components. For embarrassingly parallel problems, illustrated in Figure 4, the synchronization (both soft- ware and hardware) issues are greatly simpli ed.... In PAGE 24: ... Each asynchronous task is now a synchronous or loosely synchronous mod- estly parallel evaluation of a given chess position. There were a few examples mentioned above of metaproblems in our orig- inal survey, but a major part of Table4 , from our New York State activity,... In PAGE 35: ...4 Summary Each of three case studies illustrates how di erent applications and di erent numerical approaches to a given problem, lead to very di erent problem architectures and correspondingly the needed software support. Although our discussion is not complete, we do think that it is quite typical, and that a similar situation is seen in the other applications of Table4 , and summarized in the last two columns.... ..."

"... In PAGE 3: ... However, these issues will not be discussed here. Table 1 describes the general guidelines used in organizing Table4 . Note that we did not directly cover academic areas, and a more complete list (which included our industrial table) was produced by the Peta ops meeting [Peta:94a].... In PAGE 3: ... Here, \Info quot;, in InfoMall, refers to the information based application focus and \Mall quot; to the use of a virtual corporation (groups of \storeholders quot;) to produce the complex integrated applications enabled by HPCC. The rst column of Table4 contains the area label and some sample applications. There is also a pointer to Section 4, if appropriate.... In PAGE 7: ...Table4 : Industrial HPCC Applications 1 to 5: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 1 Computational PDE, FEM SIMD, MIMD for Sec. Fluid Dynamics Turbulence irregular adaptive 4.... In PAGE 8: ...Table4 : Industrial HPCC Applications 6 to 10: SIMULATION Application Area Problem Machine Item and Examples Comments and Software 6 Environmental Empirical Models Some SIMD but Sec. Phenomenology Monte Carlo and MIMD more 4.... In PAGE 10: ...Table4 : Industrial HPCC Applications 14 to 18: Information Analysis|\DataMining quot; Application Area Problem Machine Item and Examples Comments and Software 14 Seismic and Parallel Computers SIMD useful but Environmental already important MIMD might be Data Analysis but necessary No oil in New HPF York State 15 Image Processing Many commercial Metacomputer Sec. Medical Applications of Low Level 4.... In PAGE 11: ...Table4 : Industrial HPCC Applications 19 to 22 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand (Sec. 4.... In PAGE 12: ...Table4 : Industrial HPCC Applications 23 to 24 for Information Access InfoVision|Information, Video, Imagery and Simulation on Demand (Sec. 4.... In PAGE 13: ...Table4 : Information Integration Applications 25 to 28 These involve combinations of Information Production, Analysis, Access and Dis- semination and thus need the Integration of the various Software and Machines Architecture Issues discussed under previous application areas. Many need collaboration and \computational steering quot; technology correspond- ing to integration of computers, people, and instruments \in the loop.... In PAGE 14: ...Table4 : Information Integration Applications 29 to 33 29: Real-Time Control Systems Robotics uses Imagery to make decisions (control vehicles) Energy management controls power use and generation 30: Electronic Banking Requires Security, Privacy, Authentication, Electronic Cash, etc. 31: Electronic Shopping 32: Agile Manufacturing|Multidisciplinary Design and Concurrent Engineering (Sec.... In PAGE 14: ...3: Education (Sec. 4.5) Many commonalities with Application 25 InfoMall Living Schoolbook|6 Schools on ATM network linked to HPCC InfoVi- sion Servers at NPAC [Mills:95a] This paper is not intended to advocate a particular parallel software environment or lan- guage. Rather, we want to describe the broad capabilities of, and give examples of the parallel programming paradigm needed for the applications of Table4 . We believe that the program- ming functionality needed by a particular application is broadly determined by the problem architecture described in the following section.... In PAGE 21: ... Data parallelism occurs in large data mining sub-applications on servers with links to Java or VRML clients, that just handle visualization and interpretation modules. There were a few examples of metaproblems in our original survey, but a major part of Table4 , from our New York State activity, is the Information Integration classi cation. This class includes manufacturing and the applications 25{33, all examples of metaproblems.... In PAGE 32: ... These include processing sensor data (signal and image processing|Application 15) and simulations of such things as expected weather patterns and chemical plumes. In this way, many of the components (Applications 1 to 12 in Table4 ) are linked as part of this large metaproblem. One also needs large-scale multimedia databases with HPCC issues related to those described for InfoVISiON in Section 4.... ..."

"... In PAGE 7: ... A crucial issue on the SB-PRAM is interrupt han- dling. Table1 lists all interrupts currently in use on the SB-PRAM. The so-called synchronous inter- rupts are caused by the program execution and occur at predictable times, while asynchronous interrupts, e.... In PAGE 9: ... As described in Section 3, the SB-PRAM hardware guarantees that (in general) all memory accesses take the same number of instruction cycles, regardless of the accessing processor or the accessed memory loca- tion. The interface to the operating system (see Section 5) guarantees that no exceptions (see Table1 ) occur that may delay a single processor and thus destroy exact synchronous execution within the group. The compiler must guarantee that: (a) at (re-)entry into a synchronous region exact synchronous execu- tion is (re-)installed, (b) at branches of control ow, the compiler is responsible for proper group splitting, and (c) at joins of control ow it must restore exact synchronous execution.... ..."

"... In PAGE 16: ... With hot spots the variation is much greater, indicating the nice e ect load balancing has for smoothing the variation and reducing the gradient. Finally Table1 shows the calculated average number of moves made by a node in the entire system, with and without hot spots and with and without load balancing, and the normalized variation of the capacity at each processor from the mean. The table shows that the load balancing reduces the coe cient of variation at the cost of a very small increase in the average moves in the system, indicating that load balancing is e ective with low overhead.... ..."

"... In PAGE 4: ... The power measurements were performed at the highest frequencies permitted by the supply voltage. Table1 summarizes the characteristics of the decoders. 6.... ..."