This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Stepanov, PowerPC, life_analysis
- To: mark at codesourcery dot com (Mark Mitchell)
- Subject: Re: Stepanov, PowerPC, life_analysis
- From: Joe Buck <jbuck at synopsys dot COM>
- Date: Sat, 9 Jun 2001 11:40:21 -0700 (PDT)
- Cc: gcc at gcc dot gnu dot org, rth at cygnus dot com (Richard Henderson), dje at watson dot ibm dot com (David Edelsohn), jason at cygnus dot com (Jason Merrill)
> [ blow away some register copies before removing ADDRESSOF ]
> In any case, I'm looking for input as to people think that:
>
> - This is the right idea, and if so, whether the details are
> right.
Internals experts like Richard will have to tell us whether it's safe
and the details are right. Conceptually, it seems OK, but that might
just reflect my ignorance.
> - Whether or not we need to fix this for GCC 3.0.
By our original release criteria, it fixes an important performance
regression, so it would have been required for the release. We've relaxed
requirements on that, but this kind of poor Stepanov performance is likely
to mean poor C++ performance across the board, so I'd like to see it go in
unless the gurus consider it unsafe. Of course, I'm assuming no
regressions on primary release platforms.