Matthew Wilson has written an article for CUJ which does some performance comparisons for various C++ compilers at http://www.cuj.com/documents/s=8943/cujexp0312wilson/. Of special interest is this chart http://www.cuj.com/documents/s=8943/cujexp0312wilson/wilson_f2.htm.

Congratulations Walter!

Phill.


"Walter" <walter@digitalmars.com> wrote in message news:braq3g$2vtn$1@digitaldaemon.com...
> Matthew Wilson has written an article for CUJ which does some performance comparisons for various C++ compilers at http://www.cuj.com/documents/s=8943/cujexp0312wilson/. Of special interest is this chart http://www.cuj.com/documents/s=8943/cujexp0312wilson/wilson_f2.htm.
>
>

I've already seen another comparison table about performance (on dm site I think). But I found that with the METAKIT library, digitalmars executable is 3 times slower than VC5 !!

search in 50000 records database with vc5 : 24.931111 ms/search
search in 50000 records database with dm8.38: 65.156000 ms/search

and this is a problem particularly because I use the metakit library for quick searches...

Walter wrote:

> Matthew Wilson has written an article for CUJ which does some performance
> comparisons for various C++ compilers at
> http://www.cuj.com/documents/s=8943/cujexp0312wilson/. Of special interest
> is this chart
> http://www.cuj.com/documents/s=8943/cujexp0312wilson/wilson_f2.htm.
> 
> 

-- 
Riccardo Cohen

Articque
Les Roches
37230 Fondettes
France
web = http://www.articque.com
tel: +33 02 47 49 90 49
fax: +33 02 47 49 91 49

Which version of the compiler was it? 8.38 brought in some significant improvements in the template optimisation

"Riccardo Cohen" <rcohen@articque.com> wrote in message news:3FD9963F.60008@articque.com...
> I've already seen another comparison table about performance (on dm site I
think). But I found that
> with the METAKIT library, digitalmars executable is 3 times slower than
VC5 !!
>
> search in 50000 records database with vc5 : 24.931111 ms/search search in 50000 records database with dm8.38: 65.156000 ms/search
>
> and this is a problem particularly because I use the metakit library for
quick searches...
>
> Walter wrote:
>
> > Matthew Wilson has written an article for CUJ which does some
performance
> > comparisons for various C++ compilers at http://www.cuj.com/documents/s=8943/cujexp0312wilson/. Of special
interest
> > is this chart http://www.cuj.com/documents/s=8943/cujexp0312wilson/wilson_f2.htm.
> >
> >
>
> -- 
> Riccardo Cohen
>
> Articque
> Les Roches
> 37230 Fondettes
> France
> web = http://www.articque.com
> tel: +33 02 47 49 90 49
> fax: +33 02 47 49 91 49
>

I used 8.38. I checked by typing ...dm\bin\dmc.exe :

Digital Mars Compiler Version 8.38n
Copyright (C) Digital Mars 2000-2003.  All Rights Reserved.

Matthew Wilson wrote:
> Which version of the compiler was it? 8.38 brought in some significant
> improvements in the template optimisation
> 
> "Riccardo Cohen" <rcohen@articque.com> wrote in message
> news:3FD9963F.60008@articque.com...
> 
>>I've already seen another comparison table about performance (on dm site I
> 
> think). But I found that
> 
>>with the METAKIT library, digitalmars executable is 3 times slower than
> 
> VC5 !!
> 
>>search in 50000 records database with vc5 : 24.931111 ms/search
>>search in 50000 records database with dm8.38: 65.156000 ms/search
>>
>>and this is a problem particularly because I use the metakit library for
> 
> quick searches...
> 
>>Walter wrote:
>>
>>
>>>Matthew Wilson has written an article for CUJ which does some
> 
> performance
> 
>>>comparisons for various C++ compilers at
>>>http://www.cuj.com/documents/s=8943/cujexp0312wilson/. Of special
> 
> interest
> 
>>>is this chart
>>>http://www.cuj.com/documents/s=8943/cujexp0312wilson/wilson_f2.htm.
>>>
>>>
>>
>>-- 
>>Riccardo Cohen
>>
>>Articque
>>Les Roches
>>37230 Fondettes
>>France
>>web = http://www.articque.com
>>tel: +33 02 47 49 90 49
>>fax: +33 02 47 49 91 49
>>
> 
> 
> 

-- 
Riccardo Cohen

Articque
Les Roches
37230 Fondettes
France
web = http://www.articque.com
tel: +33 02 47 49 90 49
fax: +33 02 47 49 91 49

"Matthew Wilson" <matthew.hat@stlsoft.dot.org> wrote in news:brc5j6$1ta8$1@digitaldaemon.com:

> Which version of the compiler was it?

BTW, Matthew, afaik there is nothing like Watcom 12.0 listed in your article. There is Open Watcom 1.0 and 1.1 (1.2 RC for braves). They internally have __WATCOMC__ predefined with value adjusted to mark continuation of development but using it for naming whole product is like calling DMC Symantec C++ just because it has __SC__ predefined.

Any chance for such interesting article about strong FPU/double operations?

ABX

> > Which version of the compiler was it?
>
> BTW, Matthew, afaik there is nothing like Watcom 12.0 listed in your article. There is Open Watcom 1.0 and 1.1 (1.2 RC for braves). They internally have __WATCOMC__ predefined with value adjusted to mark continuation of development but using it for naming whole product is like calling DMC Symantec C++ just because it has __SC__ predefined.

I'm still stuck in the old Watcom mode. I should start talking about OW 1.0 etc.

Please feel free to point it out if I do it again. :)

> Any chance for such interesting article about strong FPU/double
operations?

There is a strong chance, once I'm past the book deadline (2nd Feb). By all means write to me again after that time and remind me

Cheers

Matthew