de5@ornl.gov (Dave Sill) (12/15/90)
Disclaimer: This information is provided as a public service. It is
unofficial and incomplete, and may even be wrong. For complete SPEC
benchmark results, contact:
Systems Performance Evaluation Cooperative (SPEC)
c/o Waterside Associates
39510 Paseo Padre Parkway
Suite 350
Fremont, CA 94538
415/792-3334
(After January 1, 1991)
SPEC
c/o Waterside Associates
39138 Fremont Blvd.
Fremont, CA 94538
415/792-3334
Machine GCC Esp Spice Dod Nasa7 Li Eqn Mat Fppp Tom Mean
------- --- --- ----- --- ----- -- --- --- ---- --- ----
DECstation 5000 83 122 1747 103 889 284 60 266 138 153 209
SPARCserver 490 69 138 1458 133 1012 318 63 201 162 215 220
Harris Night Hawk 65 108 1782 156 1240 313 65 211 163 182 223
SPARCstation 330 108 196 2096 196 1439 553 88 308 232 324 327
DECsystem 5400 135 161 2633 146 1569 510 81 434 227 268 328
DECsystem 5810 153 143 2876 128 2047 529 73 545 216 291 342
DECstation 3100 136 189 2523 165 1525 473 98 460 243 268 343
DECstation 2100 196 254 3418 224 2068 641 125 627 340 363 467
SPARCstation 1 139 254 2915 369 1968 690 113 410 389 445 459
Sun 4/260 150 292 5017 551 5878 683 133 485 1030 628 678
VAX 11/780 1482 2266 23951 1863 20093 6206 1101 4525 3038 2649 3868
--
Dave Sill (de5@ornl.gov) It will be a great day when our schools have
Martin Marietta Energy Systems all the money they need and the Air Force
Workstation Support has to hold a bake sale to buy a new bomber.de5@ornl.gov (Dave Sill) (01/19/91)
Here's the latest compilation of SPEC results. Send additions,
corrections, and comments to me at de5@ornl.gov.
How does one compute a geometric mean? How *exactly* is the SPECmark
computed? This table shows elapsed time in seconds for each of the
ten programs. Would SPECratios be preferable? E.g., which is better:
the 83 second GCC figure, or the 17.3 GCC SPECratio? I'm not sure how
easy it would be to fit the ratios in 80 columns.
Disclaimer: This information is provided as a public service. It is
unofficial, incomplete, and may even have a typo or two. For complete
SPEC benchmark results, contact:
Systems Performance Evaluation Cooperative (SPEC)
c/o Waterside Associates
39138 Fremont Blvd.
Fremont, CA 94538
415/792-3334
Machine GCC Esp Spice Dod Nasa7 Li Eqn Mat Fppp Tom Notes
------- --- --- ----- --- ----- -- --- --- ---- --- -----
Alacron AL860 100 105 1323 109 351 286 60 161 117 61 5
Apollo DN10010 123 176 3093 81 985 926 141 208 95 133 1
CDC 4340 97 131 1761 118 1110 306 64 333 174 187 7
CDC 4360 77 126 1774 109 1092 260 60 340 152 157 7
CDC 4380 79 125 1774 107 1092 260 60 338 151 155 7
CDC 4680 BETA 33 52 635 49 507 138 31 86 58 72 7
CDC 910B-621 82 136 2395 127 1395 381 68 466 189 245 7
CDC 920B-450 111 163 2260 168 1595 400 83 457 249 248 7
DEC 6000/410 291 349 3863 270 2450 862 164 696 405 358 7
DEC DS2100 196 254 3372 278 2230 646 125 1002 476 352 1
DEC DS2100 196 254 3418 224 2068 641 125 627 340 363 4,7
DEC DS3100 136 189 2523 165 1525 473 98 460 243 268 4,7
DEC DS3100 145 194 2500 208 1646 480 99 749 292 260 1
DEC DS5000/200 83 122 1747 103 889 284 60 266 138 153 4,7
DEC DS5400 135 161 2633 146 1569 510 81 434 227 268 4,7
DEC DS5810 153 143 2876 128 2047 529 73 545 216 291 4
DG AV310 150 173 2886 270 2161 460 105 415 366 319 7
DG AV410 154 177 2921 278 2258 481 107 439 380 331 7
DG AV5010 136 170 2753 262 2115 450 103 408 349 298 7
DG AV6200 111 133 2177 203 1688 355 81 321 279 241 7
HP 9000/340 478 985 13306 1694 16744 1881 500 4525 3038 2943 7
HP 9000/370 274 472 7044 582 5581 955 250 1616 921 757 7
HP 9000/834 145 255 2753 214 2208 530 109 431 338 312 7
HP DN10010 116 176 2013 81 980 559 99 206 95 133 7
HP9000/340 485 982 13360 1756 16617 1910 505 4625 3090 2839 1
HP9000/370 273 471 7134 590 5576 954 248 1625 923 766 1
HP9000/834 146 256 2761 214 2218 530 109 432 339 311 1
Harris Night Hawk 65 108 1782 156 1240 313 65 211 163 182 6
IBM RS/6000-320 108 139 1210 86 755 398 62 259 71 47 7
IBM RS/6000-520 109 139 1210 87 753 398 61 259 71 47 7
IBM RS/6000-530 83 109 868 67 566 313 48 208 56 35 7
IBM RS/6000-540 71 91 721 56 463 262 41 171 46 29 2,7
IBM RS/6000-730 85 109 868 67 549 313 49 202 56 35 7
IBM RS/6000-930 85 109 868 67 566 313 48 208 56 35 7
Intel 486 107 186 2722 321 3464 369 100 476 434 616 7
Intel STAR860 120 113 1618 119 447 351 62 210 139 74 5
MIPS 6280 Beta 33 52 635 49 507 138 31 86 58 72 2,7
MIPS M/120-5 119 185 2424 170 1571 404 92 764 250 238 1,7
MIPS M/2000 77 125 1997 106 1198 263 62 541 151 158 1
MIPS M/2000 78 125 1979 106 1092 261 60 340 149 150 7
MIPS RC2030 169 198 2720 190 1679 448 98 793 644 248 1
MIPS RC3240 96 128 1979 117 1110 304 64 328 171 191 7
MIPS RC3260 80 126 2013 108 1104 266 62 345 152 153 7
MIPS Rx2030 172 192 2661 188 1511 437 96 471 620 241 7
MIPS Rx3230 81 126 1687 110 957 275 59 294 192 143 7
Moto 8608 193 197 3350 295 3187 458 129 520 488 509 1,7
Moto 8612 81 99 1618 153 1148 260 53 210 199 178 7
Moto 8864DP 138 205 3609 308 3508 478 135 597 537 540 1
Moto 8864SP 85 117 1916 184 1322 300 69 246 207 228 7
Moto 8864SP 106 146 2395 230 1647 376 86 308 260 285 7
Moto 8864SP 120 195 3279 290 3172 448 133 547 504 487 1,7
Moto MPC-100 est 106 146 2395 230 1647 376 86 308 260 285 7
Moto MPC-200 est 85 117 1916 184 1322 300 69 246 207 228 7
SGI 4D/210S 82 136 2395 127 1395 381 68 466 189 245 7
SGI 4D/25S 111 163 2260 168 1595 400 83 457 249 248 7
SGI 4D/320S 69 96 1565 94 1104 243 55 340 123 161 5
Solbourne 5/801 76 139 1607 161 1248 341 65 200 170 224 7
Stardent 3010 83 112 1629 95 319 343 60 42 104 43 2,7
Sun 4/260 150 292 5017 551 5878 683 133 485 1030 628 3
Sun SS SLC est 139 255 3373 433 2283 690 114 476 453 509 7
Sun SS1 138 254 2875 374 2308 689 113 409 387 470 1
Sun SS1 139 254 2915 369 1968 690 113 410 389 445 3,7
Sun SS1+ 117 206 2395 305 1661 653 92 354 320 373 7
Sun SS330 107 196 2152 225 1800 552 88 315 323 352 1
Sun SS330 108 196 2096 196 1439 553 88 308 232 324 3,7
Sun SS490 69 138 1458 133 1012 318 63 201 162 215 3,7
VAX 11/780 ref 1482 2266 23951 1863 20093 6206 1101 4525 3038 2649 1
Notes:
1. SPEC Newsletter, Volume 1, Issue 1, Fall 1989
2. 39283@mips.mips.COM (mash@mips.COM (John Mashey))
3. SPEC Newsletter, Volume 2, Issue 1, Winter 1990
4. "Digital's RISC Family Graphics and CPU Performance Summary"
5. "Supercomputing Review", September, 1990
6. tom@hcx2.ssd.csd.harris.com (Tom Horsley)
7. Calculated from "Your Mileage May Vary", John R. Mashey
--
Dave Sill (de5@ornl.gov) It will be a great day when our schools have
Martin Marietta Energy Systems all the money they need and the Air Force
Workstation Support has to hold a bake sale to buy a new bomber.mccalpin@perelandra.cms.udel.edu (John D. McCalpin) (01/24/91)
> On 20 Jan 91 00:29:51 GMT, mccalpin@perelandra.cms.udel.edu I said: >>> On 18 Jan 91 19:11:35 GMT, de5@ornl.gov (Dave Sill) said: Dave> Here's the latest compilation of SPEC results. Send additions, Dave> corrections, and comments to me at de5@ornl.gov. me> For those interested, I have placed the information that Dave posted me> in an 'sc' spreadsheet file. This is an ascii file in 'sc' internal me> format for reading into a spreadsheet, *not* a version for printing. I have updated the file to include some interesting calculations. The original raw data has been hidden (using the sc 'z' command) so that only the interesting stuff is shown by default. You can recover the original data using the 's' command of sc. The current table has the following columns: (1) machine name (2) total SPECmark (3) integer "SPECmark" (*) (4) floating-point "SPECmark" (*) (5) ratio: FP SPEC/Integer SPEC (6) ratio: Best SPEC/Worst SPEC (*) Please note that these are entirely *unofficial* results and categories and are intended only as an easy reference. The "Integer SPECmark" is the geometric mean of the SPEC ratios for the four integer tests: gcc, espresso, li, eqntott. The "Floating-point SPECmark" is the geometric mean of the other 6 tests. The file is available by anonymous ftp from perelandra.cms.udel.edu in pub/bench/spec.sc. Please send updates and corrections to me and I will keep the file up-to-date, though if I get over 100 machines to list I will have to either restructure the calculations or build a bigger version of sc! Any ideas for other interesting calculated quantities to stick into the table are welcome..... -- John D. McCalpin mccalpin@perelandra.cms.udel.edu Assistant Professor mccalpin@brahms.udel.edu College of Marine Studies, U. Del. J.MCCALPIN/OMNET
era@era.scd.ucar.edu (Ed Arnold) (02/02/91)
In article <1991Jan18.191135.1135@cs.utk.edu> de5@ornl.gov writes: >Here's the latest compilation of SPEC results. Send additions, >corrections, and comments to me at de5@ornl.gov. Thanks for posting this info. Is this something you update on a regular basis? >Disclaimer: This information is provided as a public service. It is >unofficial, incomplete, and may even have a typo or two. For complete >SPEC benchmark results, contact: > Systems Performance Evaluation Cooperative (SPEC) > c/o Waterside Associates > 39138 Fremont Blvd. > Fremont, CA 94538 > 415/792-3334 Unfortunately, it appears that SPEC wants to restrict the availability of info from them, because the price has significantly increased. Last year, the newsletter was $150, and the newsletter+tape $450. This year, it's $399 and $699. That 166% price increase still gets you just four issues. -- era@ncar.ucar.edu