This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patches] Re: SSE min/max instruction support
On Sun, Feb 25, 2001 at 12:33:45AM +0100, Jan Hubicka wrote:
> OK, I am investigating ifcvt.c - proper place seems to be noce_emit_cmove.
No. You want a new function.
> In order to do canonicalization, I should know about constantness of the
> operands. On RISC architectures I don't think I see it - do you see
> as good solution to call init_alias_analysis and canon_exp on the
> operands before doing the simplifications?
Huh? No, you don't need to mess with alias analysis at all.
RISC machines will have loaded an fp constant into a register,
which is where it will need to be for the fmax insn anyway.
r~