Age | Commit message (Collapse) | Author |
|
Rename ConstFunctionPointerINTEL to ConstantFunctionPointerINTEL
|
|
Add SpecConstantSubgroupMaxSize to the clspv reflection non-semantic instruction set
|
|
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
instruction set
Signed-off-by: Kévin Petit <kpet@free.fr>
|
|
|
|
For details refer to https://gitlab.khronos.org/spirv/SPIR-V/-/issues/639
As part of the commit following changes have been introduced:
1. Added a separate section in spirv xml to reserve vendor specific bit masks.
2. Added a new image operand bit mask to support non constant offsets in textureGatherOffsets as defined in GL_NV_gpu_shader5
|
|
The use of these flags (the FPFastMath decoration) is already protected
by the capability, so it isn't needed to protect the individual values
as well.
|
|
An identical declaration of `OpArbitraryFloatPowNINTEL` exists just
above, with the exact same opcode and operands.
|
|
Fix minor details in SPV_INTEL_optnone extension
|
|
|
|
Spec: https://github.com/intel/llvm/pull/3198
|
|
Spec: https://github.com/intel/llvm/pull/3198
|
|
It is described in the spec as being the "same as LocalSizeHint mode,
but using <id> operands instead of literals", but the grammar had a
single <id> operand instead of the 3 literals for LocalSizeHint.
|
|
Fix grammar for PackedVectorFormat
|
|
OverflowModes enumerants
|
|
PackedVectorFormat4x8BitKHR should be enabled by the SPV_KHR_integer_dot_product
extension that first introduced it and not the DotProductInput4x8BitPackedKHR
as per the extension specification.
See http://htmlpreview.github.io/?https://github.com/KhronosGroup/SPIRV-Registry/blob/master/extensions/KHR/SPV_KHR_integer_dot_product.html
Signed-off-by: Kevin Petit <kevin.petit@arm.com>
|
|
Upstream SPV_INTEL_debug_module extension
|
|
Spec https://github.com/intel/llvm/pull/3976
|
|
|
|
Add header changes for SPV_EXT_shader_atomic_float16_add
|
|
Signed-off-by: David Neto <dneto@google.com>
Signed-off-by: Kevin Petit <kevin.petit@arm.com>
Change-Id: Icd243bb9c2a6f8a40713db215a6ca5946ea7abb3
|
|
|
|
|
|
Support SPV_KHR_subgroup_uniform_control_flow
|
|
Add CPP_for_OpenCL to grammar
|
|
Upstream ac_fixed and hls_float Intel extensions
|
|
|
|
|
|
Signed-off-by: Artem Gindinson <artem.gindinson@intel.com>
|
|
Fix OpTypeBufferSurfaceINTEL token description
|
|
According to spec: https://github.com/intel/llvm/pull/1612
|
|
SPV_INTEL_arbitrary_precision_floating_point and
SPV_INTEL_arbitrary_precision_fixed_point extensions are
being upstreamed.
Specs:
https://github.com/intel/llvm/blob/2f6e965e686354fbb25f9c177a667a646de302eb/sycl/doc/extensions/SPIRV/SPV_INTEL_arbitrary_precision_fixed_point.asciidoc
https://github.com/intel/llvm/blob/bd86b218f749ea0e20ddc18c42db491faf54014a/sycl/doc/extensions/SPIRV/SPV_INTEL_arbitrary_precision_floating_point.asciidoc
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
|
|
|
|
|
|
* The grammar was not updated when revision 3 of SPV_KHR_shader_clock
was published
* That revision renamed the *Execution* operand to *Scope*
|
|
add None as a possible value for DebugInfoFlags
|
|
Header generator: Check enumerant ordering
|
|
|
|
|
|
In the grammar, enforce ordering rules:
- Instructions must appear in order of their opcode
- Non-instructions: each successive enumerant within a single kind must
appear in order
- Reorder enumerants Subgroup*MaskKHR enums to satisfy the rule.
|
|
|
|
|
|
|
|
This is a cosmetic change for the benefit of generating the SPIR-V spec.
It reorders the "FP Denorm Mode" and "FP Operation Mode" so they are
the last sections in chapter 3 before the instruction listing.
They become 3.37 and 3.38. The idea is to preserve the section numbering
for earlier sections. For example, keep 3.31 as the Capability section.
|
|
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|
|
Spec:
https://github.com/KhronosGroup/SPIRV-Registry/blob/7d96a31cf56c60de76a6ae7a26ace3c7bfd999bf/extensions/INTEL/SPV_INTEL_fpga_cluster_attributes.asciidoc
https://github.com/KhronosGroup/SPIRV-Registry/blob/7d96a31cf56c60de76a6ae7a26ace3c7bfd999bf/extensions/INTEL/SPV_INTEL_fp_fast_math_mode.asciidoc
Signed-off-by: Dmitry Sidorov <dmitry.sidorov@intel.com>
|