On 01/09/2018 11:28 AM, Richard Biener wrote:
Note that the vectorizer gives up on loops with gathers with no target
support for
gathers. It could simply open-code the gather though (and properly cost that
open-coded variant), that's probably the way to go here.
Man, I wish I had made this
On Tue, Jan 9, 2018 at 11:58 AM, Jan Hubicka wrote:
>> On Tue, Jan 9, 2018 at 11:26 AM, Jan Hubicka wrote:
>> > Hi,
>> > gather instructions are rather hard to implement in hardware and except for
>> > skylake+ chips (i.e. haswell and Zen) they seems to be rather
> On Tue, Jan 9, 2018 at 11:26 AM, Jan Hubicka wrote:
> > Hi,
> > gather instructions are rather hard to implement in hardware and except for
> > skylake+ chips (i.e. haswell and Zen) they seems to be rather slow; to the
> > degree I did not find real world loop where gather would
On Tue, Jan 9, 2018 at 11:26 AM, Jan Hubicka wrote:
> Hi,
> gather instructions are rather hard to implement in hardware and except for
> skylake+ chips (i.e. haswell and Zen) they seems to be rather slow; to the
> degree I did not find real world loop where gather would help on
Hi,
gather instructions are rather hard to implement in hardware and except for
skylake+ chips (i.e. haswell and Zen) they seems to be rather slow; to the
degree I did not find real world loop where gather would help on Zen.
This patch simply adds a knob to disable its autogeneration (builtin