I know that fragments are rasterized in a 2x2 quad. That's why the seam between two triangles can be rasterized twice. At least that article states this. I wonder is it still true today on modern hardware?
$\begingroup$
$\endgroup$
6
-
$\begingroup$ Exactly how do you "know" that? Reference maybe? $\endgroup$– AndreasCommented May 17, 2016 at 10:16
-
$\begingroup$ @Andreas the linked article states this, although doesn't provide a reference. However, it does provide evidence that this affects performance, unless that can be explained in terms of some other cause. $\endgroup$– trichoplax is on Codidact nowCommented May 17, 2016 at 10:44
-
$\begingroup$ @trichoplax My bad. Did not read question properly. $\endgroup$– AndreasCommented May 17, 2016 at 10:47
-
$\begingroup$ Also I saw several times that instead of fullscreen quad people use big triangle that covers the screen. $\endgroup$– nikitablackCommented May 17, 2016 at 10:48
-
1$\begingroup$ @nikitablack Oh I see - I thought you meant 2 triangles instead of a quad - but you mean 1 oversized triangle that encompasses the whole screen. That's interesting. I apologise - that is very much related to this question :) $\endgroup$– trichoplax is on Codidact nowCommented May 17, 2016 at 11:20
|
Show 1 more comment
2 Answers
$\begingroup$
$\endgroup$
It is true to on today's GPUs, unless one or more vendors has implemented a technique such as this: http://graphics.stanford.edu/papers/fragmerging/shade_sig10.pdf
To clarify, rasterization is not being done multiple times, or redundantly - rather it is the shading work that is done redundantly. Each 2x2 quad will be shaded once for every triangle that covers at least one sub pixel sample in that 2x2 quad. So, for 4x MSAA, that could be as many as 16 times if every sample covers a different (very small) triangle.
$\begingroup$
$\endgroup$
1
That's why the seam between two triangles can be rasterized twice.
The seam won't be rasterized twice because of the OpenGL or D3D fill/rasterisation rules.
What I think Humus is alluding to is that a fragment shader may be executed multiple times on pixels spanning the shared edge, which just implies wasted effort.
The moral of the story is "avoid small or long/thin triangles if you want efficient rendering".
-
1$\begingroup$ Enabling GL_POLYGON_SMOOTH would cause multiple rasterization. $\endgroup$– AndreasCommented May 17, 2016 at 16:55