[fpc-devel] Benchmark for FreePascal - test

[darekM]

I've done some test
there are 5 version of revcom test
1. original
2. done bigger buffer
3  use tTextSrteam

4+ 5 in place of readln i've pass const string
6 only one set string variable
7 no move- I've rem MOVE procedure

I've repeat test in program, because time of test are vary (I don't know 
why)
on FPC I use -OG3ru -XX - Xs parameters

Source attached
My conclusion:
1. on Linux MOVE implementation is very slow (or I make mistake in tests)
2. on win read via text file is two times slower than via handle, on 
Linux is not so slowly (depends on pt. 1)
 on win consume 50%, on Linux 80%



Results:
On Delphi+win+athlon2.8

  readln 144696
  readln +big buffer 146475
  ttextStream 52767                        <2 times faster
  no read file 31106
  no read file +big buffer 30935
  no read file 30994
  no set string  32243
  no move 18353               <2 times faster
  ttextStream 52306
  readln +big buffer 146588
  readln 144410


on FPC+win+athlon2.8 (the same machine)
  readln 88309
  readln +big buffer 88159
  ttextStream 45789        <2 times faster
  no read file 32341
  no read file +big buffer 24045
  no read file 23886
  no set string  25246
  no move 15031   <2 times faster
  ttextStream 45425
  readln +big buffer 88460
  readln 87379

on FPC +linux gentoo+athlon 2.6
  readln 76621
  readln +big buffer 79921
  ttextStream 73222        <no times faster
  no read file 65700
  no read file +big buffer 70429
  no read file 75157
  no set string  80239
  no move 27527    <3 times faster
  ttextStream 76936
  readln +big buffer 77679
  readln 78108


Darek

PS. Sorry for my English.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: revcom.pas
URL: <http://lists.freepascal.org/pipermail/fpc-devel/attachments/20051215/7995e552/attachment.ksh>