I'm not knowledgeable about programming, but I managed to modify the makefile for MSVC to work with Digital Mars and compile the MP3 encoder, LAME.

http://www.mp3dev.org/mp3/

I tried every listed optimization switch with the DM compiler and the following results were obtained in converting a test WAV file:

lame - downloaded version - 15 seconds (compiler Unknown, but I think MS VC
5.0)
lame - MS Visual C 6.0       - 16 seconds
lame - DM 8.25                   - 25 seconds
lame - Borland Free            - 34 seconds

As speed when encoding, (especially using the DLL) is quite important. I am hoping someone can enlighten me as to how I can get Digital Mars to make faster code. I have tried:

CC_OPTS    = -Nc -Jm -o /DNDEBUG

plus I tried all the various options for the optimiser and the 25 seconds is the fastest I can get.

Would the code run faster if comiled as DOSX? (I haven't been able to do this yet). Is there some flag I can set? Is it because the application is a Windows program?

(Note: I am also linking in the NASM assembler routines in both DM and VC).

Try -ff (fast floating point). You also might consider -gt to turn on profiling. This should point you to which routine is consuming all the time.


"Andrew Logan" <andrew.logan@home.com> wrote in message news:9vjnrd$2utf$1@digitaldaemon.com...
> I'm not knowledgeable about programming, but I managed to modify the makefile for MSVC to work with Digital Mars and compile the MP3 encoder, LAME.
>
> http://www.mp3dev.org/mp3/
>
> I tried every listed optimization switch with the DM compiler and the following results were obtained in converting a test WAV file:
>
> lame - downloaded version - 15 seconds (compiler Unknown, but I think MS
VC
> 5.0)
> lame - MS Visual C 6.0       - 16 seconds
> lame - DM 8.25                   - 25 seconds
> lame - Borland Free            - 34 seconds
>
> As speed when encoding, (especially using the DLL) is quite important. I
am
> hoping someone can enlighten me as to how I can get Digital Mars to make faster code. I have tried:
>
> CC_OPTS    = -Nc -Jm -o /DNDEBUG
>
> plus I tried all the various options for the optimiser and the 25 seconds
is
> the fastest I can get.
>
> Would the code run faster if comiled as DOSX? (I haven't been able to do this yet). Is there some flag I can set? Is it because the application is
a
> Windows program?
>
> (Note: I am also linking in the NASM assembler routines in both DM and
VC).
>
>
>