On Tue, 20 Mar 2018, Jakub Jelinek wrote: > On Tue, Mar 20, 2018 at 11:41:01AM +0100, Richard Biener wrote: > > If the precision matches the mode maybe. But I thought we try to > > avoid using %al (HImode) or %ax (QImode) operands in the end? > > Yes, but we try to do that only on the level of the emitted assembly, > sometimes we simply emit a QImode or HImode instruction that does > movl %eax, %ebx > and similar. That is also zero cost compared to what it would cost to do it > in SImode.
Sure - but that doesn't actually truncate? That said, even sign-conversions might cost sth like unsigned HImode to signed HImode might end up doing a HI->SImode sign-extension to have the reg in promoted form. Richard.
