Re: Alternatives to gcc (was Re: gcc 4.3: when will it become standard compiler?)

2009/1/13 Maxim Sobolev <sobomax_at_freebsd.org>:

> Well, this is workaround not a solution. Sooner or later FreeBSD will hit
> some principal limitation of the current compiler, like for example it was
> in the old days of gcc 2.xx, when FreeBSD had stuck with version that was
> outdated by few years resulting in inability to use any more or less modern
> C++ code with the system compiler. Existing processors develop all the time
> (SSE 4.2 for example) and the new architectures emerge (Cell for example),
> so that it's just matter of time when it happens again.

So have people actually done some tests with the latest gcc and the
freebsd world/kernel and  -demonstrated- a speedup with that?

I'd be happy with a crappy but fast and standard compiler in /usr/src
if it build the world and kernel and the kernel was within 5% or so of
the hyper-optimised very-latest compiler.

But then, I seem to have falled square in the "compilers can't do all
the magic; stop writing crappy code" school who believes you should
only need a magically awesome compiler for about 1% of your codebase,
and the rest should just be well-written to start with.

So I re-iterate. Why all of the discussion having the default compiler
be something new and shiny, when those who need the performance gains
can just install -that- compiler as a port and use that?



Adrian