Message-Id: <m0xD6hI-000S1uC@inti.edu.ar>
Comments: Authenticated sender is <salvador AT natacha DOT inti DOT edu DOT ar>
From: "Salvador Eduardo Tropea (SET)" <salvador AT inti DOT edu DOT ar>
Organization: INTI
To: djgpp AT delorie DOT com
Date: Mon, 22 Sep 1997 14:56:03 +0000
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: Strange benchmark results
Precedence: bulk

Jan Hubicka <hubicka AT horac DOT ta DOT jcu DOT cz> wrote:

>I was making some comparsions of EMX and DJGPP. I wanted to write test
>to benchamrk speed of systems calls real interrupt and similiar stuff,
>that depends at dpmi implementation/EMX extender. I wrote following simple
>program to make 10000 times file aaaa, write there 100 bytes (unbuferred
>one write call for each byte) and close:
[sniped]
>And I was really surprised by result:
>
>         DJGPP     EMX
>cwsdpmi   5.00     0.38(rsx dpmi forced here)
>pmode     2.97     ----
>w/o DPMI  ----     0.16(emx extender)
>Win DOS   6.04     0.43
>Windows   ----     8.51(rsxwin extender)
>QDPMI     7.91     0.22(rsx dpmi forced here)

  I think a lot of us tried it, no?.
  My conclution is that Jan is wrong.
  His conclution is OK, EMX uses buffered open/write/close as Borland does.
  From the other hand DJGPP doesn't make it and I think that's GOOD!. Open
is supposed to be low level and not buffered. Buffered means quick BUT
dangerous!

  Here are the results I got in an AMD DX4 100:

MSC 5.10      4.73           <---- 1988
BC++ 3.1      4.84           <---- 1992
DJGPP _write  12.42
CWSDPMI       13.30          <---- 1996
CWSDPMI R0    13.35
WIN 3.1	     13.46
CWSDPMI+RHIDE 13.57/14.84

  BC and M$ seems to be almost 3 times faster but take a look to the same
using fopen:

DJGPP         4.01
BC++ 3.1      5.55
MSC 5.10      5.93

  Hah!, DJGPP is faster even with pmode/real mode! isn't so bad no?

  So if you want speed then use fopen and if you want low level and a little
more safe use open.
  An interesting thing that I saw while testing was that MSC was indicating
0.4 seconds but that was because the program was failing to write. Was very
hard for me to figure out the switchs for the M$ open.

SET
------------------------------------ 0 --------------------------------
Visit my home page: http://www.geocities.com/SiliconValley/Vista/6552/
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Alternative e-mail: set-sot AT usa DOT net - ICQ: 2951574
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA
TE: +(541) 759 0013