Commits

Branch selector

 floor_toolchain_1406

User selector

All users

All time

Commit History

Commits on Dec 5, 2025

• * SPIRVWriter: fixed int <-> uint type mismatches in CompositeInsert/VectorInsertDynamic

  a2flo

  committed
  08fc835

  Copy full SHA for 08fc835

Commits on Dec 3, 2025

• * SPIRVWriter: fixed potential int <-> uint argument type mismatches in atomic ops

  a2flo

  committed
  359d3cb

  Copy full SHA for 359d3cb

Commits on Nov 28, 2025

• * SPIRVWriter: added support for explicitly marked flat shading I/O

  a2flo

  committed
  8949617

  Copy full SHA for 8949617

Commits on Nov 25, 2025

• * SPIRVWriter: don't translate stores to nullptr or poison value stores

  Show description for c22b10f
  a2flo

  committed
  c22b10f

  Copy full SHA for c22b10f

Commits on Oct 11, 2025

• * SPIRVWriter: enable MakePointerAvailable/MakePointerVisible/NonPrivatePointer on loads/stores again when they are annotated with floor_coherent

  a2flo

  committed
  f58b4d9

  Copy full SHA for f58b4d9

Commits on Oct 9, 2025

• * SPIRVWriter: add NonUniform decoration on OpSelect between pointers (if SSBO/image)

  a2flo

  committed
  89a46d5

  Copy full SHA for 89a46d5

Commits on Oct 8, 2025

• * SPIRVWriter: will now emit (InBounds)AccessChain rather than PtrAccessChain when the pointer operand is a global value (i.e. no indirection)

  Show description for b8c11c1
  a2flo

  committed
  b8c11c1

  Copy full SHA for b8c11c1

Commits on Oct 1, 2025

• * SPIRVWriter: deal with integer signedness mismatch between the lhs operand and the binary op itself

  a2flo

  committed
  4daab3e

  Copy full SHA for 4daab3e

Commits on Sep 29, 2025

• * SPIRVWriter: more integer signedness mismatch bitcasting when translating binary ops

  a2flo

  committed
  78f0ffd

  Copy full SHA for 78f0ffd

Commits on Sep 22, 2025

• * SPIRVWriterPass: can now specify the wanted SPIR-V version when creating the pass

  Show description for 1b78b1a
  a2flo

  committed
  1b78b1a

  Copy full SHA for 1b78b1a

Commits on Sep 21, 2025

• * SPIRVInstruction: added proper SPIRVGroupNonUniformBallot instruction class

  Show description for ea28a42
  a2flo

  committed
  ea28a42

  Copy full SHA for ea28a42

• * SPIRVWriter: added support for vector types in SIMD/shuffle/GroupNonUniform ops

  a2flo

  committed
  f22b440

  Copy full SHA for f22b440

Commits on Sep 4, 2025

• * SPIRVWriter/SPIRVContainerWriterPass: implemented now optional handling of SPV_KHR_subgroup_uniform_control_flow (global floor.vulkan_subgroup_uniform_cf metadata must be set now for this to get…

  Show description for 5612563
  a2flo

  committed
  5612563

  Copy full SHA for 5612563

Commits on Jul 29, 2025

• * SPIRVRegularizeLLVM: funnel shift lowering now happens in-place rather than via emission of a helper function

  a2flo

  committed
  437ddce

  Copy full SHA for 437ddce

Commits on Jul 23, 2025

• * SPIRVWriter: disabled emission of MakePointerVisible/MakePointerAvailable/NonPrivatePointer memory operands in OpLoad/OpStore for now as they can incur a significant performance hit

  a2flo

  committed
  3dcce67

  Copy full SHA for 3dcce67

Commits on Jul 12, 2025

• * SPIRVWriter/SPIRVRegularizeLLVM: removed our own BB sorting and replaced it with stablePreDominatorTraversal() lifted from the latest Khronos SPIRV-LLVM-Translator (original: https://github.com/…

  Show description for eeac6a0
  a2flo

  committed
  eeac6a0

  Copy full SHA for eeac6a0

Commits on May 30, 2025

• * SPIRVWriter: don't add any layout decoration on variables in Private/Function memory

  Show description for cda682a
  a2flo

  committed
  cda682a

  Copy full SHA for cda682a

Commits on May 29, 2025

• * SPIRVWriter: when translating GEP or libfloor array load indices, it is now ensured that indices are always unsigned integers (if <= 32-bit or <= 16-bit for work-group memory) and have at least …

  Show description for 51d790e
  a2flo

  committed
  51d790e

  Copy full SHA for 51d790e

Commits on May 23, 2025

• * SPIRVWriter: added SIMD/sub-group barrier support (floor.barrier.simd)

  a2flo

  committed
  ccd0496

  Copy full SHA for ccd0496

Commits on May 21, 2025

• * added support for SPV_KHR_maximal_reconvergence and SPV_KHR_subgroup_uniform_control_flow

  Show description for 17c83be
  a2flo

  committed
  17c83be

  Copy full SHA for 17c83be

Commits on May 20, 2025

• * SPIRVWriter: handle signed/unsigned mismatches in select instructions

  Show description for dc0c26c
  a2flo

  committed
  dc0c26c

  Copy full SHA for dc0c26c

Commits on May 9, 2025

• * SPIRVWriter: extended sub-group shuffle and sub-group add/min/max algorithm support to 8-bit, 16-bit and 64-bit types

  a2flo

  committed
  8554373

  Copy full SHA for 8554373

Commits on Apr 14, 2025

• * SPIRVWriter: add EarlyFragmentTests execution mode when function metadata is set

  a2flo

  committed
  5c98161

  Copy full SHA for 5c98161

Commits on Apr 11, 2025

• * SPIRVWriter: we also need to remove the DereferenceableOrNull SSBO check here as well

  a2flo

  committed
  d779a41

  Copy full SHA for d779a41

• * SPIRVWriter: added base vertex/instance index handling (handle builtins + set DrawParameters capability in vertex shaders)

  a2flo

  committed
  4018d85

  Copy full SHA for 4018d85

Commits on Apr 9, 2025

• * SPIRVWriter: we must also modify the result type of a vector shuffle even if all operand types match, but the result type doesn't

  a2flo

  committed
  ece98b8

  Copy full SHA for ece98b8

Commits on Apr 6, 2025

• * updated copyright year

  a2flo

  committed
  9991278

  Copy full SHA for 9991278

Commits on Apr 4, 2025

• * SPIRVWriter/OCLUtil: added image LOD query support

  a2flo

  committed
  1667040

  Copy full SHA for 1667040

Commits on Apr 2, 2025

• * work around broken Intel code to fix clang 20 compilation

  a2flo

  committed
  ac8a2c0

  Copy full SHA for ac8a2c0

Commits on Feb 16, 2025

• * SPIRVWriter: added handling for new floor.exit function call -> this translates to either an OpKill (discard) for fragment shaders or OpReturn for vertex shaders and compute kernels

  a2flo

  committed
  d396181

  Copy full SHA for d396181

Commits on Feb 15, 2025

• * SPIRVWriter: deal with new (actually correct) argument attributes -> only emit an IUB/SSBO variable if the metadata says so + assert when we get an IUB inside the SSBO array branch

  Show description for 6e5c36b
  a2flo

  committed
  6e5c36b

  Copy full SHA for 6e5c36b

Commits on Feb 5, 2025

• * added SPV_KHR_workgroup_memory_explicit_layout support (required now), with capabilities being adding when workgroup/local memory is being used

  Show description for 3b5f7c1
  a2flo

  committed
  3b5f7c1

  Copy full SHA for 3b5f7c1

Commits on Dec 2, 2024

• * initial Vulkan 1.4 support

  a2flo

  committed
  aed2003

  Copy full SHA for aed2003

Commits on Oct 16, 2024

• * SPIRVWriter: in OpenCL, when using OpImageWrite with an LOD argument, add the cl_khr_mipmap_image and cl_khr_mipmap_image_writes source extensions (still doesn't pass validation, but this is cor…

  Show description for 93ee967
  a2flo

  committed
  93ee967

  Copy full SHA for 93ee967

Commits on Oct 4, 2024

• * SPIRVWriter: vector shuffle translation also handles rewriting the result type now when necessary

  Show description for 649d8a9
  a2flo

  committed
  649d8a9

  Copy full SHA for 649d8a9

Pagination

Previous

Next