[patches] Re: SSE min/max instruction support

[Jan Hubicka]

What about the name for function? Best one I can come with
is emit_condtional_move, but it is already used :)
> 
> 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.
I am looking for recognizing as often as possible the other cases.
For instance many conditional moves may turn out to be set flags
instructions and similar.
For this I want to know if operands are constant.
Also such function seems to be only resonable place to do optimizations
like (a==b?a:b) to "a".
I've just checked that such tests (if written using if, not ?) get
as far as to ifcvt, where they seems to get converted to conditional move
that is later zapped.

Honza

> 
> 
> r~