magnum

Commit Graph

Author	SHA1	Message	Date
Vladimír Vondruš	c92d608aa7	Shaders: assume drawOffset is 0 if DRAW_COUNT is 1. This is always true in the single-draw case, since setDrawOffset() asserts on this. In the multi-draw case this optimization doesn't make sense, because it doesn't make sense to create a multidraw shader with just one draw. On an Intel 630 GPU this resulted in single-draw single-material Phong to go from 550 ms to 440, which is roughly a 20% improvement. For the simpler shaders the difference is even higher. The multidraw numbers stayed the same as before, obviously.	5 years ago
Vladimír Vondruš	380441cd66	doc: this page was missing the footer prev/next/up navigation.	5 years ago
Vladimír Vondruš	1f2eef7b6a	doc: document uniform buffer and multidraw support in builtin shaders.	5 years ago
Vladimír Vondruš	bf65bf859a	GL: might make sense to say what unit are the sizes in. No, it's not 64k vec4s, but 64 kB.	5 years ago
Vladimír Vondruš	007360a9c8	GL,Shaders: allow draw() and friends to be chained. It had a point originally, but with the new multidraw workflows it's just annoying.	5 years ago
Vladimír Vondruš	c785ef2e4a	Forgot to update the year here.	5 years ago
jvn	99966194cc	MeshTools: update a broken link.	5 years ago
Vladimír Vondruš	d845793b8f	Shaders: benchmark texture array variants.	5 years ago
Vladimír Vondruš	95379ace8e	Shaders: support array textures in Flat and Phong.	5 years ago
Vladimír Vondruš	a95fa9f3de	Shaders: what's up with the method order here.	5 years ago
Vladimír Vondruš	658e8109a1	Shaders: fix grouping in Phong.vert.	5 years ago
Vladimír Vondruš	c4b748ad2d	Shaders: add a generic texture layer attribute and UBO definition. The attribute will be used for instanced texture layers in texture-array-enabled shaders.	5 years ago
Vladimír Vondruš	45f115b931	Shaders: separate instanced colored and textured test cases. With everything smashed together it was quite annoying to eyeball and too easy to miss issues.	5 years ago
Vladimír Vondruš	4c4f63b5a0	Shaders: test instanced object ID in the instanced test case. Is more self-contained that way (the new multidraw tests were done this way also).	5 years ago
Vladimír Vondruš	1ce1561eb2	Shaders: expand Flat UBO construction tests a bit.	5 years ago
Vladimír Vondruš	8077681388	Shaders: supply Vector material information in a separate UBO as well. 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.	5 years ago
Vladimír Vondruš	a2f8a920da	Shaders: supply Flat material information in a separate UBO as well. 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.	5 years ago
Vladimír Vondruš	5997d446a9	Shaders: add multidraw benchmark variants.	5 years ago
Vladimír Vondruš	2c09a2a1e6	Shaders: multidraw support in all builtin shaders.	5 years ago
Vladimír Vondruš	5d7a7d4d92	Shaders: ensure the ObjectId and Bitangent attribs are not used together. 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.	5 years ago
Vladimír Vondruš	63a8df1075	CMake: make dynamic plugins dependencies of tests that use CompareImage. So when building just a particular test, the relevant plugins get implicitly built as well without having to build everything.	5 years ago
Vladimír Vondruš	a5f662072d	Shaders: update notes about what platforms got tested. 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.	5 years ago
Vladimír Vondruš	2c578cb4df	Shaders: ah so this wasn't a Mesa bug, it was my stupidity. I just put this aside when I discovered the error, thinking it was a Mesa bug. Now that ARM Mali yelled about the same, I realized it wasn't just Mesa. Note to self: Mesa has no bugs. Can you just finally accept that?!	5 years ago
Vladimír Vondruš	b109216caa	Shaders: what in the name of fuck. That feeling when you lose three hours debugging STRANGE shader compiler issues that happen only on ES, seeing stuff like "unexpected HASH_TOKEN at line 140" or "unterminated ifdef" on just any compiler you try, and then you spot THIS. FFS. Apparently this is how I was porting shaders in 2013, but not all, I was mostly sane, wrapping things in a nice ifdef EXPLICIT_UNIFORM_LOCATION, except this one case in `b9a72bd3d1` where I temporarily went full retard. No idea why.	5 years ago
Vladimír Vondruš	2ca8d0ebb2	Shaders: clarify what the heck.	5 years ago
Vladimír Vondruš	b62f1903eb	Shaders: ensure the structures are 4-byte aligned.	5 years ago
Vladimír Vondruš	24f6c33069	Shaders: ye, SwiftShader, thanks for reminding me of the harsh reality. 256 vectors in total is really damn low, guess I'll have to optimize the remaining uniform structures to deduplicate the redundant material info.	5 years ago
Vladimír Vondruš	4aa5217947	Shaders: a crutch for MSVC's neverending fails. Hope this is enough?	5 years ago
Vladimír Vondruš	0512edf754	Shaders: Clang 3.8, WHY are you runing my nice tidy paddings.	5 years ago
Vladimír Vondruš	0863d9e811	Shaders: reassign MeshVisualizer uniforms to be contiguous for UBOs too.	5 years ago
Vladimír Vondruš	0ea727f21e	Shaders: ah, GCC, and your optimizations.	5 years ago
Vladimír Vondruš	79d70f1e3a	Shaders: add UBO variants to the benchmark. 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).	5 years ago
Vladimír Vondruš	ef9da0ec96	Shaders: add UBO support to all shaders.	5 years ago
Vladimír Vondruš	37c8cfd384	Shaders: MeshVisualizer color map binding was done in a wrong branch. It probably didn't matter as much as the only platform without ARB_explicit_uniform_location is Mac, which doesn't have ARB_shading_language_420pack either.	5 years ago
Vladimír Vondruš	73ba2b9d64	Shaders: benchmark all shader variants. 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.	5 years ago
Vladimír Vondruš	666fe17603	package/ci: don't even attempt to run benchmarks on SwiftShader. The damn thing has more bugs than features and doesn't support timer queries anyway, so there's no point in trying anything.	5 years ago
Vladimír Vondruš	18dcfae51e	package/archlinux: don't run benchmarks in check(). Takes too long and is useless, as we don't see the output anyway. Also deduplicate the env var setters and export them just once at the start.	5 years ago
Vladimír Vondruš	d343be1d8c	Shaders: this accidentally didn't actually test the thing. ...but was the same as the "blending" case above. Fortunately there was no bug hidden underneath.	5 years ago
Vladimír Vondruš	1e6da607c9	Shaders: minor.	5 years ago
Vladimír Vondruš	87b7be26f9	Shaders: forgot to rename this.	5 years ago
Vladimír Vondruš	d5623b0386	Shaders: wrongly copypasted variable name in a test.	5 years ago
Vladimír Vondruš	9bf1d4c1bd	Shaders: whoops, tested with a wrong shader here as well. But the output is the same because the default 3D projection is orthographic.	5 years ago
Vladimír Vondruš	1279d3fe5d	Shaders: merge all MeshVisualizer constructor tests together. 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.	5 years ago
Vladimír Vondruš	52c6c30fdf	Shaders: this SKIP was too early.	5 years ago
Vladimír Vondruš	ae3fd92fd3	Shaders: uh oh? good thing there was no bug hidden underneath.	5 years ago
Vladimír Vondruš	96a0ecd87e	Shaders: doc++	5 years ago
Vladimír Vondruš	47e7de47c8	Shaders: remove the AbstractVector base. 167 additions, 341 deletions. I think that speaks for itself about benefits of this base class.	5 years ago
Vladimír Vondruš	94f0dcb421	Shaders: this thing is ... not used anywhere.	5 years ago
Vladimír Vondruš	d3e11765ac	Shaders: EXPLICIT_TEXTURE_LAYER is very misleading. I went through renaming this on many places quite some time ago, but this one slipped through. Now that UBOs will be a thing, rename to EXPLICIT_BINDING instead of EXPLICIT_UNIFORM_BINDING.	5 years ago
Vladimír Vondruš	ee82113590	GL: this isn't needed on Apple platforms either.	5 years ago

1 2 3 4 5 ...

8922 Commits (c92d608aa712a7344b8f715bbf0de80f5ea2ddee) All Branches Search

8922 Commits (c92d608aa712a7344b8f715bbf0de80f5ea2ddee)

All Branches