The hardware doesn't support 16-bit integer types, so we need to implement
these using 32-bit integer instructions and then convert the result back
to 16-bit.
---
src/intel/Makefile.sources| 1 +
src/intel/compiler/brw_nir.c | 2 +
src/intel/compile
It may be useful to just use nir_algebraic for this. We already do for
trig workarounds. It's more painful from a build-system perspective but,
in general, the fewer hand-rolled algebraic lowering passes we have, the
better.
On Wed, Apr 11, 2018 at 12:20 AM, Iago Toral Quiroga
wrote:
> The har
side-note, not sure if it really effects what you are doing here, but
karol ran into some cases, like 8bit signed imax, which needs to be
"lowered" to 16b (or 32b) and converted back for hw that doesn't
support smaller than 16b (or 32b). I think I have the same case with
ir3, which also has 16b bu
On Tue, Apr 24, 2018 at 7:38 AM, Rob Clark wrote:
> side-note, not sure if it really effects what you are doing here, but
> karol ran into some cases, like 8bit signed imax, which needs to be
> "lowered" to 16b (or 32b) and converted back for hw that doesn't
> support smaller than 16b (or 32b).
On Tue, Apr 24, 2018 at 5:45 PM, Jason Ekstrand wrote:
> On Tue, Apr 24, 2018 at 7:38 AM, Rob Clark wrote:
>>
>> side-note, not sure if it really effects what you are doing here, but
>> karol ran into some cases, like 8bit signed imax, which needs to be
>> "lowered" to 16b (or 32b) and converted
On April 24, 2018 18:38:33 Rob Clark wrote:
On Tue, Apr 24, 2018 at 5:45 PM, Jason Ekstrand wrote:
On Tue, Apr 24, 2018 at 7:38 AM, Rob Clark wrote:
side-note, not sure if it really effects what you are doing here, but
karol ran into some cases, like 8bit signed imax, which needs to be
"lower
On Tue, 2018-04-24 at 23:11 -0400, Jason Ekstrand wrote:
> On April 24, 2018 18:38:33 Rob Clark wrote:
>
> On Tue, Apr 24, 2018 at 5:45 PM, Jason Ekstrand > wrote:
> On Tue, Apr 24, 2018 at 7:38 AM, Rob Clark
> wrote:
>
> side-note, not sure if it really effects what you are doing here, but
>
Some of these comments may be duplicates of ones I made the first time
through.
On Wed, Apr 11, 2018 at 12:20 AM, Iago Toral Quiroga
wrote:
> The hardware doesn't support 16-bit integer types, so we need to implement
> these using 32-bit integer instructions and then convert the result back
> to
On Wed, 2018-04-25 at 07:05 -0700, Jason Ekstrand wrote:
> Some of these comments may be duplicates of ones I made the first
> time through.
>
> On Wed, Apr 11, 2018 at 12:20 AM, Iago Toral Quiroga om> wrote:
> > The hardware doesn't support 16-bit integer types, so we need to
> > implement
> >