sussyandClaude Sonnet 5.5 4d60321f1f feat(shader): emulate per-vertex parameters on the mesh path ...

A pixel shader that reads per-vertex parameters (v_interp_mov of P10 and
P20) on a device without VK_KHR_fragment_shader_barycentric now works on
the mesh path too. The mesh program stages each such parameter in
workgroup memory and writes the values of the three vertices of every
primitive as a per-primitive output, in the order of the triangle
indices; the pixel shader reads them from there.

The emulation throws when the staging, the output components or the
output memory exceed the device's mesh limits, or when the mesh
configuration uses a provoking vertex other than the first.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>

2026-10-19 17:26:37 +02:00

..

ControlFlow

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

IntermediateRepresentation

Merge pull request #6530 from milescoviello/feat/smoothing-mixed-samples

2026-10-19 06:06:26 +02:00

Optimization

Merge pull request #6762 from simonbronk/fix/stencil-op-value-export

2026-10-19 15:45:10 +02:00

RdnaDecoder

feat(shader): decode mixed SGPR and VCC buffer descriptors

2026-10-14 12:26:01 +03:00

SpirvBackend

feat(shader): emulate per-vertex parameters on the mesh path

2026-10-19 17:26:37 +02:00

tests

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

Translation

Merge pull request #6762 from simonbronk/fix/stencil-op-value-export

2026-10-19 15:45:10 +02:00

BdaAbi.hpp

Merge pull request #5859 from gjonn/perf/shader-bda-anchors

2026-10-19 02:54:44 +13:00

CacheKey.hpp

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

CMakeLists.txt

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

CompiledVariant.hpp

refactor(shader): remove resource specialization from artifact key

2026-10-08 17:26:38 +03:00

Deserialize.cpp

perf(recompiler): keep LDS loads branched and drop the LDS sentinel

2026-10-17 12:19:09 +13:00

FragmentOutputSpecialization.hpp

feat(agc): implement 11_11_10 float color targets through 32-bit uint attachments

2026-10-17 17:29:21 -04:00

ImageTableAbi.hpp

feat(agc): plan table columns whose descriptor spans two loads of a record

2026-10-14 18:20:24 +02:00

PatchSpirvTools.cmake

fix(build): keep bundled SPIRV validator tables valid at exit

2026-10-15 12:38:32 -04:00

PipelineSpecialization.hpp

fix(agc): move workgroup offset specialization IDs to 8241-8243

2026-10-18 22:54:44 -05:00

Recompiler.cpp

feat(shader): emulate per-vertex parameters on the mesh path

2026-10-19 17:26:37 +02:00

Recompiler.hpp

feat(shader): emulate per-vertex parameters on the mesh path

2026-10-19 17:26:37 +02:00

RuntimeAbi.hpp

Merge pull request #7238 from gjonn/fix/recompiler-dynamic-storage-mip-slots

2026-10-18 18:58:30 +13:00

ShaderCacheDirectory.hpp

fix(shader): write cache files of 2 GiB and more in chunks

2026-10-17 14:48:47 +02:00

ShaderCachePack.cpp

feat(recompiler): load prebuilt shader packs before the writable cache

2026-10-15 05:44:34 +11:00

ShaderCachePack.hpp

feat(recompiler): load prebuilt shader packs before the writable cache

2026-10-15 05:44:34 +11:00

ShaderCacheSourceVersion.hpp

build(recompiler): generate the shader cache version as a source file

2026-10-14 05:38:36 -04:00

ShaderCacheVersion.cmake

fix(recompiler): key shader caches on compiled sources

2026-10-19 04:27:37 +02:00

ShaderDiskCache.cpp

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

ShaderDiskCache.hpp

feat(shader): emulate smooth barycentrics on the mesh path

2026-10-19 17:26:37 +02:00

StencilReplay.cpp

fix(agc): replace with a stencil operation value the pixel shader exports

2026-10-19 12:53:13 +02:00

StencilReplay.hpp

fix(agc): replace with a stencil operation value the pixel shader exports

2026-10-19 12:53:13 +02:00

ThreadOwned.hpp

fix(agc): free per-thread driver state at thread exit without emutls

2026-10-04 19:08:54 -05:00

UserDataUse.hpp

perf(recompiler): report which user-data words a resource capture only reads through

2026-10-15 09:26:58 -05:00

VertexInputSpecialization.hpp

perf(shader): map cached SPIR-V from the shader cache files

2026-10-14 10:36:24 +04:00