Re: GCC 2.x vs 3.x for Linux kernel

["Marcel Cox" <marcel_cox at hotmail dot com>]

Let me just give a few comments on it:

1) When people want feedback from GCC people on their discussion on GCC,
they should simply CC their discussion to this list, rather than tell after
the fact "Hey, we had a GCC bashing discussion. Please take a look at it at
..."

2) GCC compilation speed is considered a serious issue, and you just have to
take a quick look at this mailing list to see that there are several
discussions on this topic and that the issue is taken seriously, but not
easy to solve

3) A linux kernel is not a good example to measure code quality of a
compiler as it is code very little suited to optimizations. However to
really see if the new compiler gives an advatnage, you should always try to
generate code for the best matching processor, rather than telling that you
are not allowed to schedule for this new processor just because the old
compiler does not support that processor model. That would be denying one of
the improvements of the new compiler version.

Marcel

"Anuradha Ratnaweera" <anuradha@gnu.org> wrote in message
1044865310.22564.2.camel@anuradha">news:1044865310.22564.2.camel@anuradha...
>
> Please ignore this mail if this has already been discussed.
>
>     http://www.uwsg.iu.edu/hypermail/linux/kernel/0302.0/0356.html
>
> Please CC me when replying.  I am not on the GCC list.
>
> Anuradha
>
>
>