marten@feki.toppoint.de (Marten Feldtmann) (05/26/91)
First of all, I would like to get some benchmark result of ST80-R4
implementations. Here are my numbers for a PC-386/33 MHz with Cache,
8MB-RAM and VGA-card (800x600, 16 colors, TSENG 4000) under Windows 3.0
with static-swapfile:
System Benchmarks (Rates)
-------------------------
Individual benchmark results (5 iterations):
Benchmark Minimum Maximum Median
--------- ------- ------- ------
AllCallsOn 86.2482 93.6524 91.047
AllImplementors 131.313 168.831 131.579
ClassOrganizer 315.104 440.0 367.781
Compiler 203.097 239.014 225.48
Decompiler 225.455 375.758 322.078
Inspect 330.909 554.878 551.515
KeyboardLookAhead 220.779 518.293 309.091
KeyboardSingle 330.804 396.364 360.927
PrintDefinition 257.576 258.359 257.576
PrintHierarchy 456.621 909.091 606.061
TextDisplay 232.727 259.109 233.151
TextEditing 256.068 274.382 264.909
TextFormatting 696.97 1045.45 696.97
TextScanning 94.5455 94.8328 94.5455
Benchmark suite results (5 iterations):
Rating Type Minimum Maximum Median
----------- ------- ------- ------
H-Mean 219.598 228.081 223.089
And then there're some bugs(?) I would like to mention...
a) file-list
Try to use the file-list utility, select a volume different to the
standard one. In the middle view you see the contents of the selected
directory. Try to make a new directory there with "make directory" and
wonderful - it's added to the view. Use another file-list to verify the
result -- the new directory is not there! The new directory was created,
but always in the directory, where st80.exe is --- grrrrrrr! The middle
view sometimes doesn't get the info from the upper view - the actual
path. How can one correct this?
b) Hierarchy Category Browser from System Full Browser (Objectkit ST)
Start the "System Full Browser", select "spawn hierarchy" and one can
get an instance of "**Hierarchy ** Category Browser". Inside this new
browser select a method and wait for the source-code of this method --
you won't get it, because nothing happens. By the way: it works
wonderful from the normal "System Browser". How can one correct this?
Greetings,
Marten
--
Marten Feldtmann, Kieler Str. 29, 2300 Kiel-14, Germany, +49 431-731916
*** please remember: I'v to pay for incomming mail ***jhgillespie@ucdavis.edu (05/29/91)
In article <91.146.12:14:27@feki.toppoint.de> marten@feki.toppoint.de (Marten Feldtmann) writes: >First of all, I would like to get some benchmark result of ST80-R4 >implementations.System Benchmarks (Rates) Here are the results for a Mac IIci (25Mhz 68030) with a Daystar cache card. ------------------------- Individual benchmark results (5 iterations): Benchmark Minimum Maximum Median --------- ------- ------- ------ AllCallsOn 48.7013 50.5192 49.14 AllImplementors 65.6566 72.9517 67.9206 ClassOrganizer 116.458 229.167 222.018 Compiler 96.5368 109.906 104.744 Decompiler 96.3481 119.231 105.802 Inspect 220.874 306.397 290.735 KeyboardLookAhead 84.493 234.16 166.016 KeyboardSingle 258.907 269.802 264.563 PrintDefinition 143.098 172.065 166.016 PrintHierarchy 275.482 336.7 275.482 TextDisplay 277.056 287.64 287.64 TextEditing 258.262 260.977 258.42 TextFormatting 158.402 580.808 534.884 TextScanning 189.091 210.811 189.091 Benchmark suite results (5 iterations): Rating Type Minimum Maximum Median ----------- ------- ------- ------ H-Mean 132.868 148.232 139.941
johnson@m2.ti.com (Doug Johnson) (05/30/91)
In article <91.146.12:14:27@feki.toppoint.de> marten@feki.toppoint.de (Marten Feldtmann) writes: >First of all, I would like to get some benchmark result of ST80-R4 >implementations. Sparcstation 2GX, 64 meg RAM. System Benchmarks (Rates) ------------------------- Individual benchmark results (5 iterations): Benchmark Minimum Maximum Median --------- ------- ------- ------ AllCallsOn 169.014 177.515 177.34 AllImplementors 242.537 269.709 262.097 ClassOrganizer 542.601 952.756 883.212 Compiler 404.719 536.058 433.01 Decompiler 464.419 826.667 733.728 Inspect 1058.14 1358.21 1213.33 KeyboardLookAhead 317.164 923.913 696.721 KeyboardSingle 614.084 734.007 685.535 PrintDefinition 534.591 685.484 582.192 PrintHierarchy 1136.36 1369.86 1333.33 TextDisplay 815.287 941.176 876.712 TextEditing 605.452 680.645 652.241 TextFormatting 1493.51 1982.76 1916.67 TextScanning 557.143 588.679 588.679 Benchmark suite results (5 iterations): Rating Type Minimum Maximum Median ----------- ------- ------- ------ H-Mean 511.233 546.574 529.792 -- Doug
warner@scubed.com (Ken Warner) (05/31/91)
In article <9048@ucdavis.ucdavis.edu> jhgillespie@ucdavis.edu writes: >In article <91.146.12:14:27@feki.toppoint.de> marten@feki.toppoint.de (Marten >Feldtmann) writes: >>First of all, I would like to get some benchmark result of ST80-R4 [stuff deleted] >Here are the results for a Mac IIci (25Mhz 68030) with a Daystar cache card. [stuff deleted] >Individual benchmark results (5 iterations): >Benchmark Minimum Maximum Median >AllCallsOn 48.7013 50.5192 49.14 >Benchmark suite results (5 iterations): [stuff deleted] >H-Mean 132.868 148.232 139.941 What do these numbers mean? I'm very interested in the IIci benchmark; but which cache card was used? Does anyone (hello PP) have any relative benchmarks relating various Mac's Sun's and '{3,4}86 machines? Ken Warner warner@scubed.scubed.com
jhgillespie@ucdavis.edu (05/31/91)
In article <737@scubed.SCUBED.COM> warner@scubed.com (Ken Warner) writes: >What do these numbers mean? I'm very interested in the IIci benchmark; >but which cache card was used? It was a DayStar 32K cache card. The benchmark is the one in PP ObjectKit.
marten@feki.toppoint.de (Marten Feldtmann) (06/01/91)
warner@scubed.com (Ken Warner) writes: [ some benchmarks numbers.... ] >What do these numbers mean? I'm very interested in the IIci benchmark; They're just numbers to compare the performances of machine against each other. As ParcPlace wrote on page 57 in their "Advanced Programming User Guide": "The median harmonic mean of the SystemBenchmark default test suite is the standard benchmark score used by ParcPlace personnel when comparing system performance in different operating configurations." And(!!!!!): "The test suite differs from the suite used in prior releases of Objectworks \ Smalltalk, so the scores cannot be compared across versions meaningfully." SystemBenchmark is part of the Smalltalk/Objectkit package made by ParcPlace. Marten -- Marten Feldtmann, Kieler Str. 29, 2300 Kiel-14, Germany, +49 431-731916 * PLEASE REMEMBER: I'v to pay for incomming mail -- keep your answer short *