Board index » delphi » TMT pascal lite: SLOW!!

TMT pascal lite: SLOW!!

so I wrote a directory recursion routine. runs through all the files while
displaying them on screen. BP's compilation run fast. TMT pascals 386
code was a lot slower. I thought their string units might be slow but even
without screen display it run much slower than the BP compilation. WHY?. I
even disabled addition of filesizes and some dos interrupts to get
clustersizes in the TMT pascal version. I thought 386 code was much faster
even when running under windows95.

 

Re:TMT pascal lite: SLOW!!


On 27 Aug 1996, Selom Ofori wrote:

Quote
> so I wrote a directory recursion routine. runs through all the files while
> displaying them on screen. BP's compilation run fast. TMT pascals 386
> code was a lot slower. I thought their string units might be slow but even
> without screen display it run much slower than the BP compilation. WHY?. I
> even disabled addition of filesizes and some dos interrupts to get
> clustersizes in the TMT pascal version. I thought 386 code was much faster
> even when running under windows95.

TMT lite doesn't "optimize" code..  They have a version that does, but you
aren't using it..

           \|/           Peace can not exist without war, nor can war
       \\\\/|\////          exist without peace. Think about it...
        \\\\|////             ----------------------------------
            V                 KnightBird - knght...@teleport.com

Re:TMT pascal lite: SLOW!!


In article <Pine.SUN.3.92.960828021338.20003S-100...@linda.teleport.com>, KnightBird <knght...@teleport.com> wrote:

Quote
>On 27 Aug 1996, Selom Ofori wrote:

>> so I wrote a directory recursion routine. runs through all the files while
>> displaying them on screen. BP's compilation run fast. TMT pascals 386
>> code was a lot slower. I thought their string units might be slow but even
>> without screen display it run much slower than the BP compilation. WHY?. I
>> even disabled addition of filesizes and some dos interrupts to get
>> clustersizes in the TMT pascal version. I thought 386 code was much faster
>> even when running under windows95.

>TMT lite doesn't "optimize" code..  They have a version that does, but you
>aren't using it..

But how does one know if one should use TMT instead of BP for speed, BTW, I had some
problems with TMT which where very strange, had to do with Videomem.

something like this:
                            var
                                videomem : dword;
                           begin
                                videomem:=_zerot+$a0000;
                             end;

Sometimes it works, but sometimes it just quits at this point (It normally had something
to do with where I put it in the source, BUT that shouldn't be a problem).

Andre Jakobs
  MicroBrain Technologies Inc.
    The Netherlands

Re:TMT pascal lite: SLOW!!


In article: <Pine.SUN.3.92.960828021338.20003S-100...@linda.teleport.com>  

Quote
KnightBird <knght...@teleport.com> writes:

>On 27 Aug 1996, Selom Ofori wrote:

>> so I wrote a directory recursion routine. runs through all the files while
>> displaying them on screen. BP's compilation run fast. TMT pascals 386
>> code was a lot slower. I thought their string units might be slow but even
>> without screen display it run much slower than the BP compilation. WHY?. I
>> even disabled addition of filesizes and some dos interrupts to get
>> clustersizes in the TMT pascal version. I thought 386 code was much faster
>> even when running under windows95.

>TMT lite doesn't "optimize" code..  They have a version that does, but you
>aren't using it..

Really? Well if I were a compiler company trying to establish a user base I
would release the best compiler I had, and limit it in some other way, such as a
nag screen or something. TMT would surely mention in the docs that their "real"
compiler was optimizing as well, but I can find any mention of this.

I'm not saying the TMT compiler is rubbish, or that it won't get better, I'm
just saying I don't beleive they've got an optimizing compiler tucked up their
jumper; at least not bug-free enough to be of any use (yet). I eagerly await for
the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT (that
at least means it includes a TD-like de{*word*81} and preferably an SVGA IDE as
well) that can really cut it, whether it comes from TMT or someone else.

-- Jay

 --------------------------------------------------------------------------
| Jason Burgon - author of Graphic Vision, TV-Like GUI for 256 Colour SVGA |
| g...@jayman.demon.co.uk   ftp://SimTel/msdos/turbopas/gv4svga1.zip         |
| ***NEW VERSION OF GV AVAILABLE FROM*** -> http://www.jayman.demon.co.uk  |
 --------------------------------------------------------------------------

Re:TMT pascal lite: SLOW!!


Quote
Jason Burgon <Ja...@jayman.demon.co.uk> wrote:
>In article: <Pine.SUN.3.92.960828021338.20003S-100...@linda.teleport.com>  
>KnightBird <knght...@teleport.com> writes:

>>On 27 Aug 1996, Selom Ofori wrote:

>>> so I wrote a directory recursion routine. runs through all the files while
>>> displaying them on screen. BP's compilation run fast. TMT pascals 386
>>> code was a lot slower. I thought their string units might be slow but even
>>> without screen display it run much slower than the BP compilation. WHY?. I
>>> even disabled addition of filesizes and some dos interrupts to get
>>> clustersizes in the TMT pascal version. I thought 386 code was much faster
>>> even when running under windows95.

>>TMT lite doesn't "optimize" code..  They have a version that does, but you
>>aren't using it..

Actually, the observed slowness has nothing to do whatsoever with the
optimization in the compiler. A program that does a directory search
would normally spend 99% of its time in disk access. A dos-extended
program, like the one compiled by TMT Pascal would always perform
slower than a real mode program, since most of the time will go into
the mode switching. If you are interested in performance measurements,
you should try a program which is CPU- or MEMORY-intensive, not
DISK-intensive.

Quote

>Really? Well if I were a compiler company trying to establish a user base I
>would release the best compiler I had, and limit it in some other way, such as a
>nag screen or something. TMT would surely mention in the docs that their "real"
>compiler was optimizing as well, but I can find any mention of this.

>I'm not saying the TMT compiler is rubbish, or that it won't get better, I'm
>just saying I don't beleive they've got an optimizing compiler tucked up their
>jumper; at least not bug-free enough to be of any use (yet). I eagerly await for
>the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT (that
>at least means it includes a TD-like de{*word*81} and preferably an SVGA IDE as
>well) that can really cut it, whether it comes from TMT or someone else.

Coming.

- Show quoted text -

Quote

>-- Jay

> --------------------------------------------------------------------------
>| Jason Burgon - author of Graphic Vision, TV-Like GUI for 256 Colour SVGA |
>| g...@jayman.demon.co.uk   ftp://SimTel/msdos/turbopas/gv4svga1.zip         |
>| ***NEW VERSION OF GV AVAILABLE FROM*** -> http://www.jayman.demon.co.uk  |
> --------------------------------------------------------------------------

Re:TMT pascal lite: SLOW!!


In article: <322b3cda.12411...@166.84.0.220>  supp...@tmt.com (TMT Development

Quote
Corporation) writes:

>Jason Burgon <Ja...@jayman.demon.co.uk> wrote:

>>I'm not saying the TMT compiler is rubbish, or that it won't get better, I'm
>>just saying I don't beleive they've got an optimizing compiler tucked up their
>>jumper; at least not bug-free enough to be of any use (yet). I eagerly await
>>for the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT
>>(that at least means it includes a TD-like de{*word*81} and preferably an SVGA IDE
>>as well) that can really cut it, whether it comes from TMT or someone else.

>Coming.

Great! I hope your de{*word*81}, like TD, will be able to work on the secondary
(mono) monitor, as this is almost a pre-requisite for debugging graphics
applications, especially SVGA ones. I don't mind if the IDE works in text or
graphics modes, just so long as it goes beyond 80 flippin' columns! :-)

-- Jay

 --------------------------------------------------------------------------
| Jason Burgon - author of Graphic Vision, TV-Like GUI for 256 Colour SVGA |
| g...@jayman.demon.co.uk   ftp://SimTel/msdos/turbopas/gv4svga1.zip         |
| ***NEW VERSION OF GV AVAILABLE FROM*** -> http://www.jayman.demon.co.uk  |
 --------------------------------------------------------------------------

Re:TMT pascal lite: SLOW!!


: I eagerly await for
: the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT
: (that at least means it includes a TD-like de{*word*81} and preferably an
: SVGA IDE as well) that can really cut it, whether it comes from TMT or
: someone else.

An IDE for FPK-Pascal is under development (actually it works
but it's uncertain if Borland will allow using TurboVision,
seems we will have to wait for FreeVision, also under development).

For OS/2 such a thing already exists: Virtual Pascal.
A Dos version of Virtual Pascal may be done by LSX-Power.

Links to all mentioned compilers are on my web page.

Klaus
--
Klaus Hartnegg, Institut fuer Biophysik, Hansa-Strasse 9a, D-79104 Freiburg
hartn...@uni-freiburg.de   http://www.brain.uni-freiburg.de/~klaus/

Re:TMT pascal lite: SLOW!!


In article: <50j83o$...@n.ruf.uni-freiburg.de>  

Quote
hartn...@sun2.ruf.uni-freiburg.de (Klaus Hartnegg) writes:

>: I eagerly await for
>: the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT
>: (that at least means it includes a TD-like de{*word*81} and preferably an
>: SVGA IDE as well) that can really cut it, whether it comes from TMT or
>: someone else.

>An IDE for FPK-Pascal is under development (actually it works
>but it's uncertain if Borland will allow using TurboVision,
>seems we will have to wait for FreeVision, also under development).

Just release it as an .EXE for now then. Someone at FPK owns a copy of TP7 or
BP7 don't they?

Quote
>For OS/2 such a thing already exists: Virtual Pascal.
>A Dos version of Virtual Pascal may be done by LSX-Power.

But like FPK, LSX hasn't got a de{*word*81}. It also needs a memory manager that
provides 32-bit DPMI, and VP plus LSX starts getting expensive. :-(

-- Jay

 --------------------------------------------------------------------------
| Jason Burgon - author of Graphic Vision, TV-Like GUI for 256 Colour SVGA |
| g...@jayman.demon.co.uk   ftp://SimTel/msdos/turbopas/gv4svga1.zip         |
| ***NEW VERSION OF GV AVAILABLE FROM*** -> http://www.jayman.demon.co.uk  |
 --------------------------------------------------------------------------

Re:TMT pascal lite: SLOW!!


Quote
>>>for the first reasonably priced 32-bit DPMI Pascal compiler DEVELOPMENT SUIT
>>>(that at least means it includes a TD-like de{*word*81} and preferably an SVGA IDE
>>>as well) that can really cut it, whether it comes from TMT or someone else.

In the DOS32 distribution there's a file debug.obj which, when linked
in, allows a interactive debugging session.

I tried to contact TMT-support about how to link this file in, but all
they said was to 'uses debug', but that's only a runtime-error reporting
unit :-(

Any ideas on how to link it in ?

Greetings, Mark
--
Mark Bijster      'No-one is perfect in this unperfect world'
ma...@sci.kun.nl                          -Ziggy Marley-
University of Nijmegen, The Netherlands
http://www.sci.kun.nl/cgi-bin-thalia/smoelfind?1990/markb.html

Other Threads