Re: Use builtin_widen_mult_even/odd in expand_vector_divmod

2012-06-29 Thread Richard Henderson
On 06/29/2012 12:43 PM, Jakub Jelinek wrote: > 2012-06-29 Jakub Jelinek > > * tree-vect-generic.c (expand_vector_divmod): For even/odd > widening multiply, put even always as first argument to > VEC_PERM_EXPR. Ok. r~

Re: Use builtin_widen_mult_even/odd in expand_vector_divmod

2012-06-29 Thread Jakub Jelinek
On Fri, Jun 29, 2012 at 11:33:39AM -0700, H.J. Lu wrote: > On Fri, Jun 29, 2012 at 9:50 AM, Richard Henderson wrote: > > We use it everywhere else, but it got forgotten here.  Saves two shuffles > > on Altivec. > > > > Although with all of this duplication it makes me wonder if we shouldn't > >

Re: Use builtin_widen_mult_even/odd in expand_vector_divmod

2012-06-29 Thread H.J. Lu
On Fri, Jun 29, 2012 at 9:50 AM, Richard Henderson wrote: > We use it everywhere else, but it got forgotten here.  Saves two shuffles on > Altivec. > > Although with all of this duplication it makes me wonder if we shouldn't just > give up > on the idea of auto-generating MULT_HIGHPART from othe

Use builtin_widen_mult_even/odd in expand_vector_divmod

2012-06-29 Thread Richard Henderson
We use it everywhere else, but it got forgotten here. Saves two shuffles on Altivec. Although with all of this duplication it makes me wonder if we shouldn't just give up on the idea of auto-generating MULT_HIGHPART from other operations (particularly given the extra type frobbing involved).