RFA: pervasive SSE codegen inefficiency

[Richard Henderson]

On Mon, Sep 19, 2005 at 05:19:20PM -0700, Dale Johannesen wrote:
> (Although just which subregs are safe to look under will require more
> attention than I've given it, if we want this in.)

Look at MODES_TIEABLE or something.

> Really I don't think this is an RA problem at all.

You don't?!?  Please defend that with something besides the current
structure of our RA.  Which is *known* to have major deficiencies.

> (define_insn "*addmixed<mode>3"
>   [(set (match_operand:V2DI 0 "register_operand" "=x")
> 	(subreg:V2DI (plus:SSEMODE124
> 	  (match_operand:SSEMODE124 2 "nonimmediate_operand" "xm")
> 	  (subreg:SSEMODE124 (match_operand:V2DI 1 "nonimmediate_operand" "%0") 0)) 0))]

I absolutely will not allow you do add 5000 of these patterns.
Which is what you'll need if you think you'll be able to solve
the problem this way.


r~