On Mon, Feb 11, 2019 at 10:07:44AM +0100, Eric Botcazou wrote: > > You make an arbitrary distinction between certain CPUs and duplicate > > patterns based on that. > > Sure, somewhat arbitrary, but that's a proof of concept and IMO better than > treating every processor the same way. The alternative would be to > complicate > further the existing patterns by means of the "enabled" attribute or > somesuch, > but I can try it too.
No, we should allow both integer and floating point insns for integer stores always. We just get the cost estimates slightly wrong now, apparently. Segher