I know that this isn't directly related to POV-Ray, but I was wondering
if there would be any advantage to getting service pack 3 for VC++ 5? 
Since it is around 90 megs I'd rather not bother with it unless it would
provide for better executables.

I recently figured out the reason why I was getting slow code.  I was
compiling in debug mode.  hee hee.  So I noticed there is options for
pentium and pentium pro computers.  Does service pack 3 add anything for
Pentium II CPU's?

I hope to get a decent compile of the dispersion patch after the
holidays, so I can use all the optimizations I can get. :)

Oh yeah, happy new year all!  Maybe we won't be saying that next year.

-Mike

Post a reply to this message

In article <36857A6E.8C4ED8EE@aol.com> , Mike <Ama### [at] aolcom>  wrote:

>I know that this isn't directly related to POV-Ray, but I was wondering
>if there would be any advantage to getting service pack 3 for VC++ 5? 
>Since it is around 90 megs I'd rather not bother with it unless it would
>provide for better executables.
>
>I hope to get a decent compile of the dispersion patch after the
>holidays, so I can use all the optimizations I can get. :)

Well, in case you are up for more speed the solution is easy: Throw Visual C away! 
It is everything (e.g. buggy C++ support, slow, etc.) for big slow Windows apps, but
(nearly) useless for fast, well working (and compiling) cross platform code or just
learning *good* and ISO standard conform C++, hardly any example from books by
Stroustrup will comnpile at all - or you will find one of the nice compiler errors.
For speed try using Watcom, for (still) more speed than Visual try Borland or
CodeWarrior...but unless you are a student, getting more than one IDE will be a bit
expensive :-(

>I recently figured out the reason why I was getting slow code.  I was
>compiling in debug mode.  hee hee.  So I noticed there is options for
>pentium and pentium pro computers.  Does service pack 3 add anything for
>Pentium II CPU's?

It fixes a lot of bugs (only?), especially in the C++ support. I don't know about
speed improvements or P2 support - maybe this is only available in VC 6.


     Thorsten

Post a reply to this message

I take that to mean that there's not much to be gained from getting the
service pack, unless I'm experiencing problems with the program.  That's
a relief, since I really didn't feel like downloading it anyway. :)

I'm pretty happy with VC++ though, and I don't have any reason to get
another compiler.  Programming is merely a hobby for me.  I find the
code to be fast now that I turned debugging off for the final compile. I
did a quick test on a scene with some media, height fields, fog, and a
few other things.

VC++ - 2 minutes 29 seconds
Watcom - 2 minutes 30 seconds.

I'm not pretending to know enough to claim if one is better than the
other, but I do see this as an indication that the compiler does a
satisfactory job of making the executable.

-Mike

Thorsten Froehlich wrote:
> 
> In article <36857A6E.8C4ED8EE@aol.com> , Mike <Ama### [at] aolcom>  wrote:
> 
> >I know that this isn't directly related to POV-Ray, but I was wondering
> >if there would be any advantage to getting service pack 3 for VC++ 5?
> >Since it is around 90 megs I'd rather not bother with it unless it would
> >provide for better executables.
> >
> >I hope to get a decent compile of the dispersion patch after the
> >holidays, so I can use all the optimizations I can get. :)
> 
> Well, in case you are up for more speed the solution is easy: Throw Visual C away!
> It is everything (e.g. buggy C++ support, slow, etc.) for big slow Windows apps, but
> (nearly) useless for fast, well working (and compiling) cross platform code or just
> learning *good* and ISO standard conform C++, hardly any example from books by
> Stroustrup will comnpile at all - or you will find one of the nice compiler errors.
> For speed try using Watcom, for (still) more speed than Visual try Borland or
> CodeWarrior...but unless you are a student, getting more than one IDE will be a bit
> expensive :-(
> 
> >I recently figured out the reason why I was getting slow code.  I was
> >compiling in debug mode.  hee hee.  So I noticed there is options for
> >pentium and pentium pro computers.  Does service pack 3 add anything for
> >Pentium II CPU's?
> 
> It fixes a lot of bugs (only?), especially in the C++ support. I don't know about
> speed improvements or P2 support - maybe this is only available in VC 6.
> 
>      Thorsten

Post a reply to this message

On Mon, 28 Dec 1998 01:49:21 -0600, Mike <Ama### [at] aolcom> wrote:

>I take that to mean that there's not much to be gained from getting the
>service pack, unless I'm experiencing problems with the program.  That's
>a relief, since I really didn't feel like downloading it anyway. :)

Actually, there's a really good reason for getting the service pack.
If you get it, you can turn on global optimizations for lighting.c by
taking out the misguided kludge someone on the POV-Team put in there.
If you compile that code on the original VC5, it locks up the
compiler.

Yes, in case anyone cares, I was QUITE upset to see that there.  
Kludging the code to work around a no-longer-relevant bug in a single
compiler is not the right way to program.  Perhaps a big comment at
the top that says "If you can't compile this, get SP3 or upgrade to
VC6" would be a bit more friendly to those of us who have done so.

Post a reply to this message

Mike <Ama### [at] aolcom> wrote:

>I'm pretty happy with VC++ though, and I don't have any reason to get
>another compiler.  Programming is merely a hobby for me.  I find the
>code to be fast now that I turned debugging off for the final compile. I
>did a quick test on a scene with some media, height fields, fog, and a
>few other things.
>
>VC++ - 2 minutes 29 seconds
>Watcom - 2 minutes 30 seconds.

try to do a longer test, as parse time can skew the results. our observations
is that VC or Borland are significantly faster at parsing than Watcom (probably
due to some slow routine in the Watcom run-time library, maybe malloc or some
such). but the actual render should be significantly faster using watcom.

Post a reply to this message

In article <368### [at] aolcom> , Mike <Ama### [at] aolcom>  wrote:
>I take that to mean that there's not much to be gained from getting the
>service pack, unless I'm experiencing problems with the program.  That's
>a relief, since I really didn't feel like downloading it anyway. :)

You got the point, only if you have had one of the problems - 90 MB is a lot (well, I
don't care about such things, the PC running VC are not my own ones, nor am a
responsible for them :-)

>I'm pretty happy with VC++ though, and I don't have any reason to get
>another compiler.  Programming is merely a hobby for me.

Yes, and as long as you don't want (very good) C++ code or need cross plattform
support it is reasonable choice for Windows.

>I find the
>code to be fast now that I turned debugging off for the final compile. I
>did a quick test on a scene with some media, height fields, fog, and a
>few other things.
>
>VC++ - 2 minutes 29 seconds
>Watcom - 2 minutes 30 seconds.
>
>I'm not pretending to know enough to claim if one is better than the
>other, but I do see this as an indication that the compiler does a
>satisfactory job of making the executable.

Yes, code speed, stability and portablity are hard to compare at all, there is no
compiler for every job (yet?) - while sometimes a difference is hard to measure at
all, sometimes it may be well over 50% (or even more) just depending on the compiler.
 I usually use CodeWarrior (because I program mostly for the Macintosh) and I never
compiled POV-Ray for Windows yet, but I have by now done some (cross plattform)
"algorithm only" programs and sometimes found big differences, sometimes none at all.
The point is that if the compiled version is faster for whatever you use it for it is
fine, but don't be surprised if you get other results with other scenes.  --  Good
luck playing with the POV-Ray source code!


    Thorsten

Post a reply to this message

povray.org admin team wrote:
> 
> Mike <Ama### [at] aolcom> wrote:
> 
> >I'm pretty happy with VC++ though, and I don't have any reason to get
> >another compiler.  Programming is merely a hobby for me.  I find the
> >code to be fast now that I turned debugging off for the final compile. I
> >did a quick test on a scene with some media, height fields, fog, and a
> >few other things.
> >
> >VC++ - 2 minutes 29 seconds
> >Watcom - 2 minutes 30 seconds.
> 
> try to do a longer test, as parse time can skew the results. our observations
> is that VC or Borland are significantly faster at parsing than Watcom (probably
> due to some slow routine in the Watcom run-time library, maybe malloc or some
> such). but the actual render should be significantly faster using watcom.

I havn't ran tests with 3.1a but when i compiled 3.01 with VC++ 5, (I
created my
own work around then, which eludes me on what i did) I got
slower parse time and faster render times. But I set the compiler to
optimize for a Pentium Pro.

But since I use superpatch anyways now, I'm not completely set up to
easily test 3.1a.

Post a reply to this message