After several failed attempts to make UBO performance not suck on Intel
Mesa and Windows drivers, I ended up hiding the dynamic aspect under a
flag. That way it's still possible to get the proper perf in UBO
workflows that don't do light culling, and for workflows where light
culling matters the 2x slowdown might be still better than looping
through several extra lights that don't contribute anything.
"Luckily", thanks to the DRAW_COUNT=1 and MATERIAL_COUNT=1 optimizations
not everything blows up, so i don't need to skip absolutely everything,
unfortunately Phong lights are affected by this insane crapfest as well
so basically nothing from Phong UBO support is tested there. FFS.
So it's all having the same workflow. This one results in even more
saved UBO slots per-draw than in the case of Flat, and the slowdown on
Intel is as bad as expected.
While it's one additional indirection (that has an extra cost on Intel
GPUs apparently, like with Phong and MeshVisualizer and
DistanceFieldVector already), with the assumption that draws usually
share the material info it allows to cram more draws into the 16/64k UBO
limit as the per-draw data are now one vec4 smaller.
For the indirection overhead I can imagine adding a new flag which makes
material mapping implicit (materialId == drawId). That seems to put the
benchmark numbers back to the original speed. Same could be done for
other shaders.
These deliberately share the same binding (because there's very little
space), but the shader wasn't guarding that. Discovered completely by
accident when adding tests for "multidraw with all the things" -- Mesa
gives just a warning, but ANGLE straight out fails the shader
compilation, so better have an assert there.
Besides expanding the tested platform set and updating thresholds where
needed, it makes more sense to list what is tested than what is not,
because when I forget to update the list it looks like I tested while I
did not.
Interestingly, shaders that have indirect material references are about
2x slower on Intel. Not the Flat or Vector, which contain the full
material in the DrawUniform. Will probably need extra
Intel-specific optimizations (like avoiding the indirection if
MATERIAL_COUNT=1).
Took me a while (several years?) to figure out a way to benchmark this
without basically duplicating the testing effort and without new
variants being too hard to add.
It didn't really make sense to have a separate test case just to check a
bunch of extra extensions first. This makes it much easier to test the
UBO variants as well.
Plus the "invalid" tests don't actually need to test any extensions, as
they're supposed to fail before any extension-dependent code path is
called.
We're going to eventually include this class in all Application classes
(need that in order to inherit a to-be-created Configuration class) and
the <string> and <vector> would be just too much. This change caused
magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not
much, but that's I guess because there's still a lot other vectors of
strings elsewhere.
There's a lot more places to clean up, will do those in separate
commits. This change is the most atomic I could do, and it introduces a
breaking change to all APIs that returned a std::vector or a
std::string. Fortunately (or as I hope) those weren't used that much, so
it shouldn't cause build breakages for that many people.
Quite a lot of the optimization ideas is borrowed from the new Vk
library -- such as "interning" the driver workaround strings to avoid
allocating their copies.
Not that either way would be more correct than the other (this is what
three.js uses I think), but it was documented everywhere to be
1/(1 + d^2)
but the calculation instead did
1/(1 + d)^2
This now also means the analytical test equation works. I should have
paid more attention to it not matching before, because that obviously
pointed to this problem.
Point lights are now significantly brighter than before, the tests were
updated to use a larger distance to avoid issues with overflows. Does
not affect the (default) directional lights in any way.
This is a -- long overdue -- breaking change to the rendering output of
this shader, finally adding support for lights that get darker over
distance. The attenuation equation is basically what's documented in
LightData, and the distinction between directional and point lights is
made using a newly added the fourth component of position (which means
the old three-component setters are all deprecated). This allows the
shader code to be practically branchless, which I find to be nice.
This breaks basically all rendering output so all existing Phong and
MeshTools::compile() test outputs had to be regenerated.
It's needed to support the new material attributes supported by glTF.
The test output is slightly different as the normal coming from
the texture wasn't normalized before.
Vladimír Vondruš