Is it time to start using 4.0?

Is it time to start using 4.0?

[sean darcy]

I'm using 3.4.3.  I realize 4.0 isn't released and isn't suitable fro 
production work, whatever that means.

But, does 4.0 generate sufficiently better code ( athlon ) to make it worth 
the issues that may still remain?

FWIW, there's little I cou;ld find about the expected benefits - to the end 
user - of the shift to the 4.0 architecture. Conformance? Speed of 
compilation? Quality of code produced - size? speed? Or does the modular 
nature allow for all these good things once the base works? IOW, wait for 
4.1?

sean

Re: Is it time to start using 4.0?

[Giovanni Bajo]

sean darcy <seandarcy@hotmail.com> wrote:

> FWIW, there's little I cou;ld find about the expected benefits - to the
end
> user - of the shift to the 4.0 architecture. Conformance? Speed of
> compilation? Quality of code produced - size? speed? Or does the modular
> nature allow for all these good things once the base works? IOW, wait for
> 4.1?

FYI, I am planning to include this user-level information into
http://gcc.gnu.org/gcc-4.0/changes.html as we approach the release.
-- 
Giovanni Bajo

Re: Is it time to start using 4.0?

[Mike Stump]

On Saturday, November 20, 2004, at 11:06  AM, sean darcy wrote:
> I'm using 3.4.3.  I realize 4.0 isn't released and isn't suitable fro 
> production work, whatever that means.

Depends on how close to the edge you like to live.  If you just want it 
to work, without any hassles, wait until the release.  If you like to 
test pre-production software, and like filing bug reports, or want to 
verify that the release of gcc 4.0.0 will work well with your software, 
before the 4.0.0 release, then now is a good time to test it out.  If 
you want us to tell you the benefits for your software, wait until the 
release; if you want to tell us what the benefits to your software is, 
then test it now, and let us know.

If you do decide to test it out, things to watch for would be 
compilation speed and generated code quality.

This is why, we cannot answer your question, as the question has more 
to do with you, than with us.  We expect that the quality of the 
compiler is close to release quality, generally, I'd say that we like 
to keep it near release quality, though, bear in mind, if you get 
unlucky, you may find some temporal instabilities.

Re: Is it time to start using 4.0?

[Bradley Lucier]

Re:

> We expect that the quality of the compiler is close to release 
> quality, generally, I'd say that we like to keep it near release 
> quality, though, bear in mind, if you get unlucky, you may find some 
> temporal instabilities.

Well, I must be really unlucky, since 4.0 has never correctly compiled 
C code generated by the Gambit-C 4.0 beta Scheme->C compiler.  It's 
true that about once a month somebody fixes something that gets rid of 
another code generation problem, but I'll consider 4.0 "close to 
release quality" when I can start testing code generation quality and 
compiler run times rather than code generation correctness.

Brad

PS:  If anyone can tell me which version, if any, of gdb on either x86 
linux, sparc solaris 9, or MacOS 10.3.6, can give me accurate debug 
information for code compiled by 4.0 with options

FLAGS_OBJ = -O1 -fno-math-errno -fschedule-insns2 -fno-trapping-math 
-fno-strict-aliasing -fomit-frame-pointer -fPIC

I would really appreciate it.