artemist/yuzu
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

astc_decoder: Reduce workgroup size

Browse source 
This reduces the amount of over dispatching when there are odd dimensions (i.e. ASTC 8x5), which rarely evenly divide into 32x32.
This commit is contained in:
ameerj 2021-07-31 23:55:20 -04:00
parent 5ab8053511
commit c439fc9be9
3 changed files with 5 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
src/video_core/host_shaders/astc_decoder.comp
View file

@ -22,7 +22,7 @@
#endif
layout(local_size_x = 32, local_size_y = 32, local_size_z = 1) in;
layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
BEGIN_PUSH_CONSTANTS
UNIFORM(1) uvec2 block_dims;

4
src/video_core/renderer_opengl/util_shaders.cpp
View file

@ -82,8 +82,8 @@ void UtilShaders::ASTCDecode(Image& image, const ImageBufferMap& map,
glFlush();
for (const SwizzleParameters& swizzle : swizzles) {
const size_t input_offset = swizzle.buffer_offset + map.offset;
const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 32U);
const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 32U);
const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 8U);
const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 8U);
const auto params = MakeBlockLinearSwizzle2DParams(swizzle, image.info);
ASSERT(params.origin == (std::array<u32, 3>{0, 0, 0}));

4
src/video_core/renderer_vulkan/vk_compute_pass.cpp
View file

@ -358,8 +358,8 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
});
for (const VideoCommon::SwizzleParameters& swizzle : swizzles) {
const size_t input_offset = swizzle.buffer_offset + map.offset;
const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 32U);
const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 32U);
const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 8U);
const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 8U);
const u32 num_dispatches_z = image.info.resources.layers;
update_descriptor_queue.Acquire();