Since the last TOP500 [3] report the list has changed on quite some points. In this section we indicate the systems that have dropped out and we describe the architectures that have come in newly. The new list is given in Table 1.1
Table 1.1: Number of Systems Installed
N | first | medium | last | |||
SP2 | 114 | 851.3 | 1265.4 | 6 | 269 | 491 |
Power Chall | 109 | 362.2 | 465.5 | 240 | 378 | 475 |
C90 | 52 | 480.2 | 538.1 | 60 | 179 | 500 |
T3D | 32 | 666.2 | 998.2 | 4 | 113 | 388 |
CM5 | 23 | 313.7 | 685.5 | 10 | 157 | 295 |
Paragon | 21 | 293.7 | 431.3 | 2 | 174 | 377 |
VPP500 | 19 | 690.2 | 872.8 | 1 | 89 | 203 |
SX3 | 18 | 194.5 | 216.8 | 39 | 175 | 422 |
Po.Ch.-Array* | 16 | 160.9 | 250.0 | 30 | 144 | 390 |
VP2000 | 14 | 55.4 | 70.0 | 273 | 286 | 332 |
GC/PP | 10 | 49.3 | 94.7 | 128 | 283 | 478 |
SPP1000 | 10 | 33.1 | 92.8 | 354 | 359 | 363 |
S-3800 | 9 | 157.7 | 176.0 | 27 | 86 | 156 |
iPSC/860 | 8 | 20.8 | 41.0 | 492 | 496 | 499 |
Triton90* | 8 | 38.8 | 48.6 | 191 | 245 | 419 |
SPP1200* | 7 | 25.1 | 55.7 | 298 | 327 | 485 |
CM200 | 5 | 34.6 | 70.0 | 119 | 183 | 226 |
CS | 5 | 20.6 | 86.4 | 221 | 327 | 484 |
Paragon SMP* | 5 | 203.8 | 246.4 | 3 | 55 | 102 |
C4/XA | 3 | 8.7 | 9.7 | 423 | 424 | 425 |
KSR2 | 3 | 12.6 | 16.6 | 235 | 269 | 287 |
CM2 | 2 | 10.4 | 28.0 | 216 | 217 | 217 |
SP1 | 2 | 9.6 | 32.0 | 233 | 234 | 234 |
SX4* | 2 | 76.2 | 80.0 | 9 | 31 | 52 |
AlphaServer | 1 | 5.0 | 7.2 | 227 | 227 | 227 |
CM5E | 1 | 7.7 | 20.0 | 135 | 135 | 135 |
KSR1 | 1 | 3.4 | 5.1 | 348 | 348 | 348 |
Total | 500 | 4785.4 | 6903.8 |
The following systems have dropped from the list:
For the description of the various types of architectures one is referred to [3] or [2]. In this section we only describe the machines that appear for the first time in Table 1.1 in alphabetical order. For the descriptions of the other systems are given in [3]. Apart from the systems parameter list that is given for each system also the values of and as available from [1] are quoted together with the number of nodes associated with these values. The reason that the system parameter list is not sufficient is that the maximal theoretical peak performance is given for the maximal configuration of a system. Such a system may not actually exist.