gcc vs Microsoft and Borland

gcc vs Microsoft and Borland

Post by Seth Van Oor » Mon, 07 Jun 1999 04:00:00



Anybody know how gcc does on optimizing c code for speed versus
Microsoft and Borland compilers, especially for pentium and pentium II?

Seth

 
 
 

gcc vs Microsoft and Borland

Post by Art S. Kage » Tue, 08 Jun 1999 04:00:00



> Anybody know how gcc does on optimizing c code for speed versus
> Microsoft and Borland compilers, especially for pentium and pentium II?

Check out the DJGPP Home pages at www.delorie.com.  DJGPP is a 32bit
port of gcc to the MS-DOS environment.  There should be information
there about comparisons between DJGPP and MSCPP and BCPP.

Art S. Kagel

 
 
 

gcc vs Microsoft and Borland

Post by K Le » Wed, 09 Jun 1999 04:00:00



: Anybody know how gcc does on optimizing c code for speed versus
: Microsoft and Borland compilers, especially for pentium and pentium II?

check out pgcc.  i don't have the link handy, but just do a search for
them http://gcc.ml.org.  

Steve

 
 
 

gcc vs Microsoft and Borland

Post by K Le » Wed, 09 Jun 1999 04:00:00



: : Anybody know how gcc does on optimizing c code for speed versus
: : Microsoft and Borland compilers, especially for pentium and pentium II?

: check out pgcc.  i don't have the link handy, but just do a search for
: them http://gcc.ml.org.  

what the hell was I smoking?  I say I don't have the link handy, and then
give you the exact url... anyway, check it out :)

Steve

 
 
 

gcc vs Microsoft and Borland

Post by Art S. Kage » Wed, 09 Jun 1999 04:00:00


The benchmark author just send me an update via email that the Linux
Games Development Center has posted his results in their WEB Site:

//sunsite.auc.dk/linuxgames/articles/tropea_compilers.html

It basically show that GCC is not as good on Pentium as PGCC which is
not as good as EGCS and all are better than Watcom.  The entire
benchmark is scaled versus MS C so you can see the comparison with that
immediately.  Bottom line is the GCC variants are better at floating
point, and especially FP emulation and calculations, but not as good
otherwise.  It is interesting to note that with optimizations invoked
the GCC variants blow Watcom off and that is a well respected compiler.  
It also raises the question: If Watcom and GCC cannot beat MS what
optimizations for Intel CPUs does MS know about that us mortals do not?



> > Anybody know how gcc does on optimizing c code for speed versus
> > Microsoft and Borland compilers, especially for pentium and pentium II?

> Check out the DJGPP Home pages at www.delorie.com.  DJGPP is a 32bit
> port of gcc to the MS-DOS environment.  There should be information
> there about comparisons between DJGPP and MSCPP and BCPP.

Art S. Kagel
 
 
 

gcc vs Microsoft and Borland

Post by Seth Van Oor » Wed, 09 Jun 1999 04:00:00


And all of this implies that linux could probably be quite a bit faster.

Seth


> The benchmark author just send me an update via email that the Linux
> Games Development Center has posted his results in their WEB Site:

> //sunsite.auc.dk/linuxgames/articles/tropea_compilers.html

> It basically show that GCC is not as good on Pentium as PGCC which is
> not as good as EGCS and all are better than Watcom.  The entire
> benchmark is scaled versus MS C so you can see the comparison with that
> immediately.  Bottom line is the GCC variants are better at floating
> point, and especially FP emulation and calculations, but not as good
> otherwise.  It is interesting to note that with optimizations invoked
> the GCC variants blow Watcom off and that is a well respected compiler.
> It also raises the question: If Watcom and GCC cannot beat MS what
> optimizations for Intel CPUs does MS know about that us mortals do not?



> > > Anybody know how gcc does on optimizing c code for speed versus
> > > Microsoft and Borland compilers, especially for pentium and pentium II?

> > Check out the DJGPP Home pages at www.delorie.com.  DJGPP is a 32bit
> > port of gcc to the MS-DOS environment.  There should be information
> > there about comparisons between DJGPP and MSCPP and BCPP.

> Art S. Kagel

 
 
 

gcc vs Microsoft and Borland

Post by Seth Van Oor » Wed, 09 Jun 1999 04:00:00


One thing they didn't seem to make clear was whether they took the best
performance over all the optimiziation levels listed in parentheses or
what. At least in pgcc, the '-fomit-frame-pointer' doesn't get turned on
till around -O5, and this is one of the biggest optimizations you can
make.

Seth


> The benchmark author just send me an update via email that the Linux
> Games Development Center has posted his results in their WEB Site:

> //sunsite.auc.dk/linuxgames/articles/tropea_compilers.html

> It basically show that GCC is not as good on Pentium as PGCC which is
> not as good as EGCS and all are better than Watcom.  The entire
> benchmark is scaled versus MS C so you can see the comparison with that
> immediately.  Bottom line is the GCC variants are better at floating
> point, and especially FP emulation and calculations, but not as good
> otherwise.  It is interesting to note that with optimizations invoked
> the GCC variants blow Watcom off and that is a well respected compiler.
> It also raises the question: If Watcom and GCC cannot beat MS what
> optimizations for Intel CPUs does MS know about that us mortals do not?



> > > Anybody know how gcc does on optimizing c code for speed versus
> > > Microsoft and Borland compilers, especially for pentium and pentium II?

> > Check out the DJGPP Home pages at www.delorie.com.  DJGPP is a 32bit
> > port of gcc to the MS-DOS environment.  There should be information
> > there about comparisons between DJGPP and MSCPP and BCPP.

> Art S. Kagel

 
 
 

gcc vs Microsoft and Borland

Post by Paul F. Ku » Thu, 10 Jun 1999 04:00:00



Quote:> > Anybody know how gcc does on optimizing c code for speed versus
> > Microsoft and Borland compilers, especially for pentium and pentium II?

   YMMV, but for our simulation program compiled with VC++ 6.0 on NT
and egcs 1.1.2 (-O2) on Linux, runs at about the same speed.

--

Stanford Linear Accelerator Center, Stanford University
Voice: (650) 926-2884   Fax: (650) 926-4335

 
 
 

gcc vs Microsoft and Borland

Post by Christopher Mahmoo » Fri, 11 Jun 1999 04:00:00



Quote:> It also raises the question: If Watcom and GCC cannot beat MS what
> optimizations for Intel CPUs does MS know about that us mortals do not?

I don't know if it's that simple though.  It seems like the gcc people understand
that there's more to a compiler than the speed of the object code it creates. I
haven't used MS stuff for a few years, but from what I remember the error
messages (to take just one example) really sucked.  And about the only
machine I can't use gcc on around me make coffee...
-ckm
 
 
 

1. Microsoft vs Borland

Did anyone hear any specifics about a settlement between Borland and
Microsoft?  The accusation was that Borland felt Microsoft was aggressively
recruiting their senior programmers and diluting their staff.  Using my
intuition -- I would assume this false because I have not seen any evidence
of improvement in Microsoft products recently.  Any thoughts?

2. Solaris 1.1.2: Routing, redirects

3. Is there aIntegrated Development Tool like Borland C++, Borland J++ for linux?

4. PPP doesn't redial

5. Who will win? Borland or Microsoft or Programmers?

6. problems with forwarding

7. WABI & Microsoft C++/Borland C++

8. screwed permissions! how to fix???

9. They both suck! (was: Borland or Microsoft compilers ?)

10. A soluation for Microsoft vs. DOJ was: Microsoft, a Public Utility

11. Linux vs OS2 vs NT vs Win95 vs Multics vs PDP11 vs BSD geeks

12. Matrox Mystique ands X.

13. Apogee C/C++ vs gcc vs ...