Message-ID: <382177B2.CEC4B27C@mpx.com.au>
Date: Thu, 04 Nov 1999 23:10:26 +1100
From: infinity girl <entropic AT mpx DOT com DOT au>
X-Mailer: Mozilla 4.5 [en] (Win95; I)
X-Accept-Language: en
MIME-Version: 1.0
To: Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
CC: djgpp AT delorie DOT com
Subject: Re: mcount, and gprof
References: <199911041247 DOT NAA12665 AT acp3bf DOT physik DOT rwth-aachen DOT de>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp AT delorie DOT com

hi,

thanks fr yr reply.
i remember about mcount now..i hadn't used the profiler in months, and
had forgotten it was 
used by the profiler itself.

i am using version 2.81 of djgpp so i don't think there is anything
buggy happening.
i was a bit wrong with the hundreds of thousands tho.... i have
functions being called millons of times:

Each sample counts as 0.055556 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 22.97      0.94     0.94                             mcount
 21.62      1.83     0.89  1148074     0.00     0.00  readBuf
 17.57      2.56     0.72   574804     0.00     0.00  bsPut
 10.81      3.00     0.44        1   444.44  2049.23  outputFile
  9.46      3.39     0.39  1290460     0.00     0.00  dcopy
  5.41      3.61     0.22   142388     0.00     0.00  writeBuf
  5.41      3.83     0.22        1   222.22   839.66  calcFreqs

that is a sample of some of the top ones.
as much time is spent in my function readBuf as is spent in mcount
profiling helped me alot at first, becuz __dpmi_int was being called the
most.
a tip from someone showed me that this was caused by excessive I/O, and
i wrote
buffering routines.

i am currently only using the -O2 switch, so i will try the next level
of optimisation
and see if that helps :)   usually i would rather have a smaller
executable, considering
i am struggling with a 329MB  hard drive...inlining will affect
executable size i guess?
i'm not sure if my functions are small enough to inline anyway.

anyway thanks again :)
daniel.