This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug rtl-optimization/2962] inefficient code with and
- From: "herbert at gondor dot apana dot org dot au" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 13 Sep 2004 04:25:53 -0000
- Subject: [Bug rtl-optimization/2962] inefficient code with and
- References: <20010526114559.2962.95318-quiet@bugs.debian.org>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From herbert at gondor dot apana dot org dot au 2004-09-13 04:25 -------
Subject: Re: inefficient code with and
On Mon, Sep 13, 2004 at 03:37:03AM -0000, rth at gcc dot gnu dot org wrote:
>
> Note that either movzbl or movzwl is required on i686 to avoid a
> partial register stall. Which is why things are generated this way.
> It would perhaps be possible to avoid this for earlier processors,
> but I don't believe this is worth the effort.
Could we have it optimised away with -Os though? I don't think the
stall would matter there.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=2962