gvw@castle.ed.ac.uk (Greg Wilson) (03/01/91)
This message is a reminder that "Past, Present, Parallel: A Survey of Parallel Computing at the Beginning of the 1990s", edited by Arthur Trew and Greg Wilson (that's me), published by Springer-Verlag, will be on bookshelves from March 11 (or so they tell me). As the table of contents included below (and posted to this group last November) indicates, the book is a survey of who's doing what commercially with parallel computing. Hope you enjoy it, Greg Wilson p.s. "enjoy" is a Scots word meaning "buy". ====================================================================== 1 Introduction 1 1.1 An Introduction to Parallel Processing: 2 1.1.1 B. P. (Before Parallelism) 2 1.1.2 Computer Architectures 6 1.1.3 Computer Software 9 1.1.4 Types of Parallelism 12 2 SIMD: Specialisation Equals Success 13 2.1 Active Memory Technology: 14 2.1.1 The Company 14 2.1.2 The Machine 16 2.1.3 The Development System 20 2.1.4 Programming Languages 21 2.1.5 The Market Niche 22 2.1.6 The Competition 23 2.1.7 The Future 24 2.2 MasPar Computer Corporation: 26 2.2.1 The Company 26 2.2.2 The Machine 27 2.2.3 The Program Development System 32 2.2.4 Programming Languages 32 2.2.5 The Market 33 2.2.6 The Competitors 34 2.2.7 The Future 35 2.3 Thinking Machines Corporation: 36 2.3.1 The Company 36 2.3.2 The Machine 37 2.3.3 The Development System 43 2.3.4 Programming Languages 43 2.3.5 The Market Niche 45 2.3.6 The Competition 46 2.3.7 The Future 47 3 Shared Memory Multiprocessors: The Evolutionary Approach 55 3.1 Alliant Computer Systems: 56 3.1.1 The Company 56 3.1.2 The Hardware 57 3.1.3 The Visualisation Machines 59 3.1.4 Software 60 3.1.5 The Market Niche 61 3.1.6 The Competition 62 3.1.7 The Future 62 3.2 BBN Advanced Computers Inc.: 64 3.2.1 The Company 64 3.2.2 The Machine 66 The Butterfly Switch 66 The Processors 68 The Communications System 69 3.2.3 The Development System 70 3.2.4 Programming Languages 72 3.2.5 The Market Niche 73 3.2.6 The Competition 74 3.2.7 The Future 75 3.3 Concurrent Computer Corporation: 76 3.3.1 Background 76 3.3.2 The 3200 Series 76 3.3.3 The 8000 Series 78 3.3.4 The Market 78 3.3.5 The Future 79 3.4 Convex Computer Corporation: 80 3.4.1 The Company 80 3.4.2 The Hardware 81 3.4.3 Software 83 3.4.4 The Market Niche 84 3.4.5 The Competition 85 3.4.6 The Future 85 3.5 Encore Computer Corporation: 87 3.5.1 The Company 87 3.5.2 The Machines 89 Multimax 89 The Encore 91 91 3.5.3 The Development System 93 3.5.4 Programming Languages 93 3.5.5 The Market Niche 94 3.5.6 The Competition 95 3.5.7 The Future 96 3.6 FPS Computing: 97 3.6.1 The Company 97 3.6.2 The Machine 99 3.6.3 The Development System 102 3.6.4 Programming Languages 103 3.6.5 The Market Niche 103 3.6.6 The Competition 104 3.6.7 The Future 105 3.7 Sequent Computer Systems: 106 3.7.1 The Company 106 3.7.2 The Machine 107 3.7.3 The Operating System 110 3.7.4 Programming languages 111 3.7.5 The Market Niche 112 3.7.6 The Competition 113 3.7.7 The Future 113 4 Hypercubes: A Geometry that Works 125 4.1 Intel Scientific Computers: 126 4.1.1 The Company 126 4.1.2 The Hardware 127 iPSC/1 128 iPSC/2 129 4.1.3 The Concurrent I/O facility 130 iPSC/860 131 4.1.4 The Software 132 Languages 132 Tools 133 4.1.5 Sales 133 4.1.6 The Market Niche 134 4.1.7 The Future 136 4.2 NCUBE Corporation: 137 4.2.1 The Company 137 4.2.2 Hardware 138 The NCUBE/ten 138 The NCUBE/four and NCUBE/seven 140 NCUBE-2 140 4.2.3 The Software 142 4.2.4 The Market Niche 143 4.2.5 The Opposition 143 4.2.6 The Future 144 5 The Transputer and Its Offspring 149 5.1 Inmos Limited: 150 5.1.1 The Company 150 5.1.2 The Transputer 153 T414 153 T800 155 T400 155 Transputer Modules (TRAMs) 156 5.1.3 The Software 156 5.1.4 The Market Niche 156 5.1.5 The Future 157 5.2 Caplin Cybernetics: 159 5.2.1 The Company 159 5.2.2 Caplin's Product Range 160 The QT Series 160 HEX 161 HEXsys 162 5.2.3 The Market Niche 163 5.2.4 Present Applications 163 Computational Chemistry 163 Finite Element Analysis and Computational Fluid Dynamics 164 Relational Database Applications 164 5.2.5 The Future 164 5.3 Meiko Scientific Limited: 165 5.3.1 The Company 165 5.3.2 The Hardware 166 The In-Sun Computing Surface 169 SPARC-based Computing Surface 170 i860-based Computing Surface 171 5.3.3 Programming Environments 172 5.3.4 CSTools 172 5.3.5 Sales and Marketing 173 5.3.6 The Future 175 5.4 Parsys Limited: 176 5.4.1 The Company 176 The Esprit Supernode Projects 176 Parsys Goes It Alone 178 Telmat Informatique 178 5.4.2 The Hardware 179 5.4.3 The Software 182 The IDRIS Operating System 182 X Support 183 Transputer Development System (TDS) 183 Compilers and Libraries 184 5.4.4 The Market Niche 184 5.4.5 The Competition 184 5.4.6 Current Applications 185 Commercial 185 Engineering 185 Finance 186 5.4.7 The Future 186 5.5 Parsytec GmbH: 187 5.5.1 The Company 187 5.5.2 System Overview 188 Plug-in Boards 188 Mid-range Systems 189 The Stand-alone Systems 191 5.5.3 Applications and Performance 192 5.5.4 The Software 193 5.5.5 The Market 194 5.5.6 The Future 195 6 New Machines For New Niches 201 6.1 Cogent Research Incorporated: 202 6.1.1 The Company 202 6.1.2 The Hardware 202 6.1.3 The Software 204 6.1.4 The Market Niche 205 6.1.5 The Future 206 6.2 Silicon Graphics Incorporated: 207 6.2.1 The Company 207 6.2.2 The Hardware 208 The Personal Iris 209 The SGI POWER Range 210 Graphics Facilities and Add-on Hardware 210 6.2.3 The Software 211 The Development Machine 211 Graphics and Third-party Software 212 6.2.4 The Market Niche 213 6.2.5 The Future 214 6.3 Stardent Computer Incorporated: 215 6.3.1 The Company 215 Stellar Computer Incorporated 215 Ardent Computer Corporation 217 The Partnership with Kubota 218 6.3.2 The Hardware 219 Stardent 3000 219 6.3.3 The Software 221 Operating System and Compilers 221 Visualisation Systems 221 Third Party Software 222 6.3.4 The Market 222 6.3.5 The Future 223 6.4 Teradata Corporation: 225 6.4.1 The Company 225 6.4.2 The Hardware 226 6.4.3 Interface Software 228 6.4.4 Internal Software 229 6.4.5 Support for Critical Environments 230 6.4.6 The Marketplace 230 6.4.7 The Future 231 7 Vector Supercomputers: It's Never too Late to Parallelise 239 7.1 Cray Research Inc.: 240 7.1.1 The Company 240 7.1.2 Cray X-MP -> Cray Y-MP -> C90 241 Cray X-MP 241 Cray Y-MP 242 The C90 243 7.1.3 Cray-2 -> Cray-3 243 Cray-2 243 Cray-3 and Cray-4 245 7.1.4 The Software 246 7.1.5 The Market Niche 248 7.1.6 The Competition 249 7.1.7 The Future 250 7.2 NEC: 252 7.2.1 The Company 252 7.2.2 The Hardware 252 7.2.3 The Software 254 7.2.4 The Market 254 7.3 Supercomputer Systems Inc: 256 7.3.1 The Company 256 7.3.2 Who is Steve Chen? 257 7.3.3 SSI's Options 257 8 The Giants: Biding Their Time 261 8.1 Digital Equipment Corporation: 262 8.1.1 The Company 262 8.1.2 The Hardware 263 The Cm* 263 ASMP and SMP 263 Vax 9000 264 Firefly 265 8.1.3 The Software 265 Vax Software 265 Firefly Software 266 8.1.4 The Market 267 8.1.5 The Future 267 8.2 Fujitsu: 269 8.2.1 The Company 269 8.2.2 Vector Computers 269 8.2.3 The CAP-C5 Array Processor 272 8.2.4 The AP-1000 Array Processor 275 8.2.5 The Future 277 8.3 IBM Corporation: 279 8.3.1 The Company 279 8.3.2 The ES/9000 280 8.3.3 Operating Systems 283 MVS 283 VM/CMS 284 8.3.4 Programming Languages 284 Parallel Fortran 284 Other Software 285 8.3.5 The Market 285 8.3.6 The Future 286 9 Software: Efficiency vs. Portability? 291 9.1 Express: 292 9.1.1 The Express Family of Products 292 9.1.2 History and Philosophy 293 9.1.3 Express 293 9.1.4 Cubix and Plotix 295 9.1.5 Other Utilities 296 9.2 Helios: 297 9.2.1 Perihelion and Helios 297 9.2.2 The Helios Model 297 9.2.3 The Helios Network 298 9.2.4 Helios Programming Environment 300 9.2.5 Helios as an Environment 303 9.2.6 The Market for Helios 303 9.3 Linda: 304 9.3.1 Linda operations 305 out(t) 305 in(s) 305 rd(s) 305 inp(s) and rdp(s) 306 eval() 306 9.3.2 Linda Examples 306 Client-Server 306 Task Farm 306 9.3.3 The Future 307 9.4 Strand88: 309 9.4.1 History 309 9.4.2 Design 310 Parallel Semantics 310 Portability 310 Foreign Language Interface 311 9.4.3 Markets for Strand88 312 9.4.4 The Competition 314 9.4.5 The Future 315 10 Machines Past 317 10.1 BiiN: 318 10.1.1 The Company 318 10.1.2 The Hardware 318 10.1.3 The Software 320 10.1.4 The Market 321 10.1.5 BiiN's Demise 321 10.2 ETA Systems: 323 10.2.1 The Company 323 10.2.2 The Hardware 324 10.2.3 The Software 325 10.2.4 Costs and Sales 326 10.2.5 What Went Wrong? 327 The Stock Market and Financing 327 ETA Management 328 Marketing 328 CDC Management 329 10.3 Multiflow: 330 10.3.1 Going Up 330 10.3.2 Going Down 332 10.3.3 Why Did Multiflow Fail? 332 10.3.4 A Future for VLIW technology? 333 10.4 Myrias Research Corporation: 335 10.4.1 The Company 335 10.4.2 User-Level Model 335 10.4.3 Claims About the User-Level Model 337 10.4.4 Compromises in the User-Level Model 338 10.4.5 The Virtual Machine and Its Implementation 338 10.4.6 The Product Range 339 10.4.7 Sales 341 10.4.8 The End 341 10.5 Symult Incorporated: 342 10.5.1 Ametek and Symult 342 10.5.2 The Symult S2010 342 10.5.3 The Communications Network 343 10.5.4 Why did the S2010 fail? 345 11 Machines Future 353 A Technology 359 A.1 Cache Memory: 360 A.1.1 Performance 360 A.1.2 Cache Update Policies 360 A.1.3 Multicache Consistency 361 A.2 Message Passing Systems: 361 A.2.1 Addressing of Messages 362 Numeric Task and Type Identifiers 362 Ports 363 Channels 363 Operating System 363 A.2.2 Deadlock and Related Issues 364 A.2.3 Controlling Message Transfer 364 A.2.4 Delivering Messages 365 A.2.5 Message Selection 366 A.2.6 Placing Tasks 366 A.3 Visualisation: 367 A.3.1 Doing It In Hardware 367 A.4 Parallel Compilers: 368 A.4.1 Vectorising Compilers 369 A.4.2 Parallelising Compilers 370 VLIW Technology and Trace Scheduling 370 Trace Scheduling 371 B Units 373 C Glossary 375 D Trademarks 383 Index 387