[Mesa-dev] [PATCH 09/13] SQUASH: i965/fs: Rework fs_visitor::lower_load_payload

[Jason Ekstrand]

On Tue, May 5, 2015 at 4:20 PM, Matt Turner <mattst88 at gmail.com> wrote:
> On Tue, May 5, 2015 at 4:13 PM, Jason Ekstrand <jason at jlekstrand.net> wrote:
>> On Wed, Apr 15, 2015 at 6:44 PM, Matt Turner <mattst88 at gmail.com> wrote:
>>> On Wed, Apr 15, 2015 at 5:13 PM, Jason Ekstrand <jason at jlekstrand.net> wrote:
>>>> On Wed, Apr 15, 2015 at 1:31 PM, Matt Turner <mattst88 at gmail.com> wrote:
>>>>> On Wed, Apr 1, 2015 at 6:19 PM, Jason Ekstrand <jason at jlekstrand.net> wrote:
>>>>>> Instead of the complicated and broken-by-design pile of heuristics we had
>>>>>> before, we now have a straightforward lowering:
>>>>>>
>>>>>>  1) All header sources are copied directly using force_writemask_all and,
>>>>>>     since they are guaranteed to be a single register, there are no
>>>>>>     force_sechalf issues.
>>>>>>
>>>>>>  2) All non-header sources are copied using the exact same force_sechalf
>>>>>>     and saturate modifiers as the LOAD_PAYLOAD operation itself.
>>>>>
>>>>> Let's not do this. Nothing puts a saturate modifier on LOAD_PAYLOAD
>>>>> today, and it is kind of confusing about what it means. Can't we have
>>>>> fbwrites that write depth as well. I wouldn't think we wanted to
>>>>> saturate that.
>>>>
>>>> Sure.  I can drop saturate and just assert that it's not set.  We do
>>>> want to keep force_sechalf and force_writemask_all though.
>>>
>>> I didn't think about those before, but I don't know how a load_payload
>>> could have force_writemask_all set. Have I missed something?
>>
>> No, no one (to my knowlege) sets force_writemask_all on it but I see
>> no reason why it shouldn't be respected.  As for saturate, we do for
>> fb_writes when key->clamp_fragment_color is set.
>> --Jason
>
> In your last email in this thread, I thought we'd agreed not to do
> anything (i.e., not allow) saturate on load_payload. Handling it seems
> confusing and moreover, unnecessary.

Right.  It'll be a bit of a pain, but I can get that fixed...
--Jason