[Mesa-dev] [PATCH v4 (part2) 15/59] i965/vec4: Implement unsized array's length calculation

[Samuel Iglesias Gonsálvez]

On 28/08/15 08:48, Jordan Justen wrote:
> On 2015-08-27 23:07:26, Samuel Iglesias Gonsálvez wrote:
>>
>> On 28/08/15 08:06, Samuel Iglesias Gonsálvez wrote:
>>>
>>> On 28/08/15 02:21, Jordan Justen wrote:
>>>> On 2015-08-05 01:30:12, Iago Toral Quiroga wrote:
>>>>> From: Samuel Iglesias Gonsalvez <siglesias at igalia.com>
>>>>>
>>>>> Notice that Skylake needs to include a header in the sampler message
>>>>> so it will need some tweaks to work there.
>>>>>
>>>>> Signed-off-by: Samuel Iglesias Gonsalvez <siglesias at igalia.com>
>>>>> ---
>>>>>  src/mesa/drivers/dri/i965/brw_defines.h          |  3 ++
>>>>>  src/mesa/drivers/dri/i965/brw_shader.cpp         |  3 ++
>>>>>  src/mesa/drivers/dri/i965/brw_vec4.cpp           |  1 +
>>>>>  src/mesa/drivers/dri/i965/brw_vec4.h             |  6 ++++
>>>>>  src/mesa/drivers/dri/i965/brw_vec4_generator.cpp | 31 ++++++++++++++++
>>>>
>>>>>  src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp   | 46 ++++++++++++++++++++++--
>>>>
>>>> Shouldn't this be in brw_vec4_nir.cpp instead?
>>>>
>>>
>>> I wrote this patch before vec4 NIR backend was upstreamed. I have a
>>> separate patch that adds a similar code to vec4 nir backend.
> 
> Did this make it into v4?
> 

Nope, I have written it this week because nir vec4 backend was pushed to
master while I was on holidays.

>>> Or do you suggest to delete this patch?
>>>
>>
>> I mean: delete brw_vec4_visitor.cpp changes from this patch and add
>> brw_vec4_nir.cpp changes to it.
> 
> I think the nir path is the priority. I skimmed the series a bit and
> it didn't seem like this array length part was enabled on the nir
> path.
> 
> I don't think there's a need to enable the non-nir path. But, if you
> somehow find it useful for debugging, then you can leave the non-nir
> path.
> 

OK

> I think that going forward we should only implement new features on
> the nir path, and I assume sometime before the next release branch
> we'll delete the non-nir vec4 path. (That's how it happened with the
> scalar side anyway.)
> 

OK, I will rewrite this patch in order to keep vec4 backend specific
bits in a separate patch so we can drop it if non-nir vec4 path is
removed before this series reaches master.

Thanks,

Sam

> -Jordan
>