home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.smalltalk
- Path: sparky!uunet!cs.utexas.edu!news.uta.edu!utafll!bruce
- From: bruce@utafll.uta.edu (Bruce Samuelson)
- Subject: Latest Slopstone/Smopstone results
- Message-ID: <1993Jan23.002208.11115@utagraph.uta.edu>
- Sender: news@utagraph.uta.edu (USENET News System)
- Nntp-Posting-Host: utafll.uta.edu
- Organization: University of Texas at Arlington
- Date: Sat, 23 Jan 1993 00:22:08 GMT
- Lines: 152
-
- Here is the latest consolidated report of Slopstones and Smopstones. The bug
- fixes posted separately do not afffect any of these numbers. At the end are
- some new Smopstone runs.
-
- Digitalk does better on Smopstones than on Slopstones, especially for the DOS
- version. Low level tests favor ParcPlace more than medium level tests, and
- there are wide divergences among individual tests. Clearly Digitalk and
- ParcPlace have optimized different parts of their systems, and there is still
- room left for further optimization.
-
- I would remind anyone running these benchmarks to do a garbage collect first
- if possible. I would have coerced it before each test, but don't know how to
- do so portably. Urs Hoelzle has pointed out that although the numbers for
- individual tests fluctuate, the geometric mean is pretty stable. You can
- reduce fluctuations by running more iterations (the 'n' variable in the
- 'execute' method). Doing so for ST/V-DOS may crash it though.
-
- ven- ver- M extrn RAM slop- smop-
- tester dor sion GUI OS Ver brand model cpu FlPt Hz cache MB stone stone
- KB (low) (med)
-
- bSamuel PPS VW 1.0 Win3.1DOS 5.0 Amax none 486DX intrn 33 256 16 1.0 1.0
- bSamuel PPS 80 4.0 Win3.1DOS 5.0 Amax none 486DX intrn 33 256 16 0.995 0.973
- bSamuel PPS 80 2.5 SunVw SunOS 4.1Sun 3/50 68020 68881 16 0 12 0.067 0.102&
- bSamuel PPS 80 4.0 OpW2.0SunOS 4.1Sun 3/50 68020 68881 16 0 12 0.114 0.107
- bSamuel Dig V 2.0cnone DOS 5.0 Amax none 486DX intrn 33 256 16 0.070 0.261
- bSamuel Dig V 2.0 none DOS 3.3 cloneXT 8088 none* 5 0 640K0.002* 0.008*
- mFeldtm Dig V 2.0 PM OS2 2.01b clone 486DX intrn 33 256 16 0.411 0.71
- mFeldtm Dig V 2.0WinOS2 OS2 2.01b clone 486DX intrn 33 256 16 0.167 0.25
- mFeldtm Dig V 1.4 PM OS2 clone 486DX intrn 33 256 16 0.236 0.470
- uHoelzl PPS 80 4.0 Sun SS/2 sparc 40 64 64 1.137# 0.995
-
- * Floating point performance was extrapolated assuming an 8087.
- & Smopstones didn't include set formation benchmark.
- # uHoelzl didn't post geometric mean; I calculated it from his other numbers.
-
- -------------------------------------------------------------------------------
- bSamuel Dig V 2.0 none DOS 3.3 cloneXT 8088 none* 5 0 640K0.002* 0.008*
- -------------------------------------------------------------------------------
-
- A Smopstone run on Bruce Samuelson's XT clone:
-
- Starting benchmarks with repetition count = 1...
-
- time in smop-
- seconds stones explanation
-
- 309 0.0102 generating fractonaccis
- 146 0.0077 generating primes
- 230(*) 0.0047(*) generating and parsing streams
- 369 0.0084 generating strings
- 534 0.0022 forming sets
- 313 0.0164 sorting strings
- 345 0.0162 sorcerer's apprentice
-
- 300 0.0078(*) geometric mean
-
- (*) Floating point performance was extrapolated assuming an 8087.
-
- -------------------------------------------------------------------------------
- mFeldtm Dig V 2.0 PM OS2 2.01b clone 486DX intrn 33 256 16 0.411 0.71
- -------------------------------------------------------------------------------
-
- Some Smopstone runs on Marten Feldtmann's 486/33:
-
- Here are the results for ST/V-PM 2.0:
-
- Starting benchmarks with repetition count = 1...
-
- time in smop-
- seconds stones explanation
-
- 2,16 1,46 generating fractonaccis
- 1,03 1,09 generating primes
- 7,44 0,14 generating and parsing streams
- 4,5 0,68 generating strings
- 3,84 0,30 forming sets
- 2,34 2,19 sorting strings
- 6,5 0,86 sorcerer's apprentice
-
- 3,31 0,71 harmonic mean
-
- -------------------------------------------------------------------------------
- mFeldtm Dig V 1.4 PM OS2 clone 486DX intrn 33 256 16 0.236 0.470
- -------------------------------------------------------------------------------
-
- Here are the results for ST/V-PM 1.4:
-
- Starting benchmarks with repetition count = 1...
-
- time in smop-
- seconds stones explanation
-
- 4,25 0,743 generating fractonaccis
- 1,44 0,780 generating primes
- 5,47 0,199 generating and parsing streams
- 8,25 0,374 generating strings
- 5,63 0,207 forming sets
- 2,78 1,848 sorting strings
- 18,12 0,309 sorcerer's apprentice
-
- 5,001 0,470 harmonic mean
-
- -------------------------------------------------------------------------------
- mFeldtm Dig V 2.0WinOS2 OS2 2.01b clone 486DX intrn 33 256 16 0.167 0.25
- -------------------------------------------------------------------------------
-
- Here are the results for ST/V-Windows 2.0:
-
- Starting benchmarks with repetition count = 1...
-
- time in smop-
- seconds stones explanation
-
- 8.97 0.35 generating fractonaccis
- 3.53 0.31 generating primes
- 7.46 0.14 generating and parsing streams
- 12.78 0.24 generating strings
- 12.32 0.09 forming sets
- 9.22 0.55 sorting strings
- 17.88 0.31 sorcerer's apprentice
-
- 9.32 0.25 harmonic mean
-
- -------------------------------------------------------------------------------
- uHoelzl PPS 80 4.0 Sun SS/2 sparc 40 64 64 1.137# 0.995
- -------------------------------------------------------------------------------
-
- A Smopstone run on Urs Hoelzle's SparcStation-2:
-
- Here are the SS-2 results (the individual numbers fluctuate somewhat
- but the geomean is quite stable):
-
- Starting benchmarks with repetition count = 1...
-
- time in smop-
- seconds stones explanation
-
- 4.087 0.772449 generating fractonaccis
- 0.952 1.17962 generating primes
- 0.87 1.25402 generating and parsing streams
- 2.495 1.23888 generating strings
- 1.273 0.916732 forming sets
- 6.617 0.776636 sorting strings
- 5.866 0.954824 sorcerer's apprentice
-
- 2.36784 0.994532 harmonic mean
- --
- **********************************************************
- * Bruce Samuelson Department of Linguistics *
- * bruce@ling.uta.edu University of Texas at Arlington *
- **********************************************************
-