[Mesa-dev] [PATCH 0/5] R600, GCN: Guard Band support

[Roland Scheidegger]

Am 11.04.2016 um 15:58 schrieb Marek Olšák:
> On Mon, Apr 11, 2016 at 2:41 PM, Roland Scheidegger <sroland at vmware.com> wrote:
>> Am 11.04.2016 um 00:34 schrieb Marek Olšák:
>>> Hi,
>>>
>>> This patch series adds Guard Band support into r600g and radeonsi.
>>>
>>> It first implements the Guard Band in radeonsi, then it moves all radeonsi scissor & viewport code into gallium/radeon, and then r600g is switched to it and its original scissor & viewport code is deleted.
>>>
>>> The differences between the R600 and GCN code are almost none.
>>>
>>> This should improve performance if clipping is the bottleneck.
>>>
>>> Grigori Goronzy started this originally.
>>>
>>> Please review.
>>>
>>> Marek
>>
>> If this changes rendering of (large) points, you should probably mention
>> this. But I wouldn't know if it actually does (it's just what I'd expect).
> 
> Since points are either trivially accepted or trivially discarded, the
> guard band has no effect on them. The guard band only affects
> primitives that cross the viewport boundary and must be clipped.
Yes, but I thought this essentially replaced viewport with guardband +
scissoring. In which case it will affect points which are inside the vp
but extend to outside it as now parts of them will get scissored away.
But maybe the hw doesn't work like that ;-).


> 
>> (I'm not entirely sure if some other cases wouldn't be affected by this
>> too, like fill mode line.)
> 
> Yes, but the OpenGL definition is pretty bad already. See the
> GL_ARB_viewport_array spec and check out the paragraph containing the
> word "shitty". (and have fun)
> 

Yes, it's non-ideal (just saying if it really is different, it would be
worth mentioning it in the commit).

Roland