writes:
> GCN uses V64BImode to represent vector masks in the middle-end, and DImode
> bit-masks to represent them in the back-end. These must be converted at
> expand
> time and the most convenient way is to simply use a SUBREG.
>
> This works fine except that simplify_subreg needs to be able t
On Tue, Sep 11, 2018 at 4:36 PM Andrew Stubbs wrote:
>
> On 05/09/18 13:43, Richard Biener wrote:
> > No. You might want to look into the x86 backend if there's maybe more
> > tweaks
> > needed when using non-vector mask modes.
>
> I tracked it down to the vector alignment configuration.
>
> App
On 05/09/18 13:43, Richard Biener wrote:
No. You might want to look into the x86 backend if there's maybe more tweaks
needed when using non-vector mask modes.
I tracked it down to the vector alignment configuration.
Apparently the vectorizer likes to build a "truth" vector, but is
perfectly
On Wed, Sep 5, 2018 at 2:40 PM Andrew Stubbs wrote:
>
> On 05/09/18 13:05, Richard Biener wrote:
> > On Wed, Sep 5, 2018 at 1:51 PM wrote:
> >>
> >>
> >> GCN uses V64BImode to represent vector masks in the middle-end, and DImode
> >> bit-masks to represent them in the back-end. These must be con
On 05/09/18 13:05, Richard Biener wrote:
On Wed, Sep 5, 2018 at 1:51 PM wrote:
GCN uses V64BImode to represent vector masks in the middle-end, and DImode
bit-masks to represent them in the back-end. These must be converted at expand
time and the most convenient way is to simply use a SUBREG.
On Wed, Sep 5, 2018 at 1:51 PM wrote:
>
>
> GCN uses V64BImode to represent vector masks in the middle-end, and DImode
> bit-masks to represent them in the back-end. These must be converted at
> expand
> time and the most convenient way is to simply use a SUBREG.
x86 with AVX512 uses SImode in
On Wed, Sep 05, 2018 at 12:50:25PM +0100, a...@codesourcery.com wrote:
> 2018-09-05 Andrew Stubbs
>
> gcc/
> * simplify-rtx.c (convert_packed_vector): New function.
> (simplify_immed_subreg): Recognised Boolean vectors and call
> convert_packed_vector.
> ---
> + in
