On Fri, Mar 04, 2005 at 10:40:02AM +0100, Martin Husemann wrote:
> On Thu, Mar 03, 2005 at 06:18:10PM -0600, Richard Rauch wrote:
> > I was noticing that certain MonopolySoft systems compile code relatively
> > quickly.
> 
> What "certain" system are you talking about?

E.g., an 800NHz Athlon with v6 of their compiler compiled more
quickly than GCC on a 2GHz AMD64 ("3200 AMD64").  I can't compare
so well to GCC on i386, but the AMD64 seems to compile with GCC
faster than that same Athlon with GCC  (now back running NetBSD).


> The M$ compiler and linker are dog slow once you are using optimization. And
> I mean *realy* slow.

For a current project, I need to be sure that their .net compiler
can handle the code, so I'm using .net (there is no particular
constraint to use .net features, just that the code has to compile
with .net).  Since it uses OpenGL, I needed soemthing other than
the Athlon, so I picked up a low-end new computer.  I.e., I still
can't directly compare apples to apples, but the MS compiler feels
snappier than I would expect GCC to feel on the same system.  For
giggles, I tried to enable optimizations, but they were disabled.
There was a note attached saying that most optimizations conflict
with "edit and continue" and with "runtime checks".

I tried disabling the runtime checks that I could find, but could not
find any way to turn on or off the "edit and continue", and the
optimization options remained forced-off.


Thanks for the thoughts, though.  Maybe GCC is actually compiling
pretty quickly already, though the impression that I get is one of
it being slow.  Maybe if I turn the GCC warning levels off, it will
pick up speed, or if I could find out how to enable optimizations
with .net it might slow down a lot.

-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/