This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Add -mno-branch-likely for mips
- To: law at redhat dot com
- Subject: Re: Add -mno-branch-likely for mips
- From: Mike Stump <mrs at windriver dot com>
- Date: Tue, 30 Jan 2001 17:59:28 -0800 (PST)
- Cc: gcc-patches at gcc dot gnu dot org
> From: Mike Stump <mrs@windriver.com>
> Date: Mon, 29 Jan 2001 16:30:45 -0800 (PST)
> To: law@redhat.com
> > To: Mike Stump <mrs@windriver.com>
> > From: Jeffrey A Law <law@redhat.com>
> > Date: Fri, 26 Jan 2001 11:45:09 -0700
> > > Add -mno-branch-likely for mips.
> >
> > So why do you really need this?
> I've suggested implementing support for the CPU variants he has in
> mind, and having them select the likely win by cpu type. I'll see
> what we says to that.
Here is what he said:
> Except that we're usually not targeting individual CPUs either (with
> the exception of the Vr5400 little-endian library, which does use
> the -mcpu= field in the compiler command line options). The -mcpu=X
> flags are too specific for our needs, and the -mipsX flags are
> usually way too general.
So, the case is, we want to tune a little bit, for a feature, but not
for an entire specific CPU. A case he mentioned was the assembler
working around silcon bugs, due to a specific CPU type and not wanting
to get those bits.
Do you have any strong objections to this? I have a weak objection,
but I don't think the cost of the flag is so onerous as to argue for
not putting in this work. What's your feeling?