Skip to content

G
glslang
Commit 80829fed by Neil Henning
Options

Very WIP wave ops.

parent 3cf27e00
Showing with 1350 additions and 8 deletions
Test/baseResults/hlsl.wavebroadcast.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/hlsl.waveordered.comp.out 0 → 100644
hlsl.waveordered.comp
Shader version: 500
local_size = (32, 16, 1)
0:? Sequence
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 move second child to first child ( temp uint)
0:6 indirect index (layout( row_major std430) buffer uint)
0:6 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:6 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:6 Constant:
0:6 0 (const uint)
0:6 add ( temp uint)
0:6 add ( temp uint)
0:6 component-wise multiply ( temp uint)
0:6 add ( temp uint)
0:6 component-wise multiply ( temp uint)
0:6 direct index ( temp uint)
0:6 '@gl_NumWorkGroups' ( in 3-component vector of uint NumWorkGroups)
0:6 Constant:
0:6 1 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 2 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 1 (const int)
0:6 component-wise multiply ( temp uint)
0:6 '@gl_NumSubgroups' ( in uint unknown built-in variable)
0:6 direct index ( temp uint)
0:6 '@gl_NumWorkGroups' ( in 3-component vector of uint NumWorkGroups)
0:6 Constant:
0:6 0 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 0 (const int)
0:6 '@gl_SubgroupID' ( in uint unknown built-in variable)
0:6 Constant:
0:6 1 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
Linked compute stage:
Shader version: 500
local_size = (32, 16, 1)
0:? Sequence
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 move second child to first child ( temp uint)
0:6 indirect index (layout( row_major std430) buffer uint)
0:6 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:6 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:6 Constant:
0:6 0 (const uint)
0:6 add ( temp uint)
0:6 add ( temp uint)
0:6 component-wise multiply ( temp uint)
0:6 add ( temp uint)
0:6 component-wise multiply ( temp uint)
0:6 direct index ( temp uint)
0:6 '@gl_NumWorkGroups' ( in 3-component vector of uint NumWorkGroups)
0:6 Constant:
0:6 1 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 2 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 1 (const int)
0:6 component-wise multiply ( temp uint)
0:6 '@gl_NumSubgroups' ( in uint unknown built-in variable)
0:6 direct index ( temp uint)
0:6 '@gl_NumWorkGroups' ( in 3-component vector of uint NumWorkGroups)
0:6 Constant:
0:6 0 (const int)
0:6 direct index ( temp uint)
0:6 '@gl_WorkGroupID' ( in 3-component vector of uint WorkGroupID)
0:6 Constant:
0:6 0 (const int)
0:6 '@gl_SubgroupID' ( in uint unknown built-in variable)
0:6 Constant:
0:6 1 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
// Module Version 10000
// Generated by (magic number): 80001
// Id's are bound by 46
Capability Shader
Capability GroupNonUniform
1: ExtInstImport "GLSL.std.450"
MemoryModel Logical GLSL450
EntryPoint GLCompute 4 "CSMain" 17 22 30 40
ExecutionMode 4 LocalSize 32 16 1
Source HLSL 500
Name 4 "CSMain"
Name 6 "@CSMain("
Name 10 "data"
MemberName 10(data) 0 "@data"
Name 12 "data"
Name 17 "@gl_NumWorkGroups"
Name 22 "@gl_WorkGroupID"
Name 30 "@gl_NumSubgroups"
Name 40 "@gl_SubgroupID"
Decorate 9 ArrayStride 4
MemberDecorate 10(data) 0 Offset 0
Decorate 10(data) BufferBlock
Decorate 12(data) DescriptorSet 0
Decorate 17(@gl_NumWorkGroups) BuiltIn NumWorkgroups
Decorate 22(@gl_WorkGroupID) BuiltIn WorkgroupId
Decorate 30(@gl_NumSubgroups) BuiltIn NumSubgroups
Decorate 40(@gl_SubgroupID) BuiltIn SubgroupId
2: TypeVoid
3: TypeFunction 2
8: TypeInt 32 0
9: TypeRuntimeArray 8(int)
10(data): TypeStruct 9
11: TypePointer Uniform 10(data)
12(data): 11(ptr) Variable Uniform
13: TypeInt 32 1
14: 13(int) Constant 0
15: TypeVector 8(int) 3
16: TypePointer Input 15(ivec3)
17(@gl_NumWorkGroups): 16(ptr) Variable Input
18: 8(int) Constant 1
19: TypePointer Input 8(int)
22(@gl_WorkGroupID): 16(ptr) Variable Input
23: 8(int) Constant 2
30(@gl_NumSubgroups): 19(ptr) Variable Input
32: 8(int) Constant 0
40(@gl_SubgroupID): 19(ptr) Variable Input
43: TypePointer Uniform 8(int)
4(CSMain): 2 Function None 3
5: Label
45: 2 FunctionCall 6(@CSMain()
Return
FunctionEnd
6(@CSMain(): 2 Function None 3
7: Label
20: 19(ptr) AccessChain 17(@gl_NumWorkGroups) 18
21: 8(int) Load 20
24: 19(ptr) AccessChain 22(@gl_WorkGroupID) 23
25: 8(int) Load 24
26: 8(int) IMul 21 25
27: 19(ptr) AccessChain 22(@gl_WorkGroupID) 18
28: 8(int) Load 27
29: 8(int) IAdd 26 28
31: 8(int) Load 30(@gl_NumSubgroups)
33: 19(ptr) AccessChain 17(@gl_NumWorkGroups) 32
34: 8(int) Load 33
35: 8(int) IMul 31 34
36: 8(int) IMul 29 35
37: 19(ptr) AccessChain 22(@gl_WorkGroupID) 32
38: 8(int) Load 37
39: 8(int) IAdd 36 38
41: 8(int) Load 40(@gl_SubgroupID)
42: 8(int) IAdd 39 41
44: 43(ptr) AccessChain 12(data) 14 42
Store 44 18
Return
FunctionEnd
Test/baseResults/hlsl.waveordered.frag.out 0 → 100644
hlsl.waveordered.frag
ERROR: 0:3: 'WaveGetOrderedIndex' : WaveGetOrderedIndex() unsupported in a pixel/fragment shader
ERROR: 1 compilation errors. No code generated.
Shader version: 500
gl_FragCoord origin is upper left
ERROR: node is still EOpNull!
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Test condition and select ( temp void)
0:3 Condition
0:3 Compare Equal ( temp bool)
0:3 Constant:
0:3 0 (const uint)
0:3 ERROR: Bad aggregation op
( temp uint)
0:3 true case
0:? Sequence
0:5 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:3 false case
0:? Sequence
0:9 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
Linked fragment stage:
Shader version: 500
gl_FragCoord origin is upper left
ERROR: node is still EOpNull!
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Test condition and select ( temp void)
0:3 Condition
0:3 Compare Equal ( temp bool)
0:3 Constant:
0:3 0 (const uint)
0:3 ERROR: Bad aggregation op
( temp uint)
0:3 true case
0:? Sequence
0:5 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:3 false case
0:? Sequence
0:9 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
SPIR-V is not generated for failed compile or link
Test/baseResults/hlsl.waveordered2.comp.out 0 → 100644
hlsl.waveordered2.comp
ERROR: 0:7: 'GlobalOrderedCountIncrement' : GlobalOrderedCountIncrement() unsupported
ERROR: 1 compilation errors. No code generated.
Shader version: 500
local_size = (32, 16, 1)
ERROR: node is still EOpNull!
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 Sequence
0:6 move second child to first child ( temp uint)
0:6 'i' ( temp uint)
0:6 Constant:
0:6 42 (const uint)
0:7 move second child to first child ( temp uint)
0:7 indirect index (layout( row_major std430) buffer uint)
0:7 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:7 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:7 Constant:
0:7 0 (const uint)
0:7 ERROR: Bad unary op
( temp uint)
0:7 'i' ( temp uint)
0:7 Constant:
0:7 1 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
Linked compute stage:
Shader version: 500
local_size = (32, 16, 1)
ERROR: node is still EOpNull!
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 Sequence
0:6 move second child to first child ( temp uint)
0:6 'i' ( temp uint)
0:6 Constant:
0:6 42 (const uint)
0:7 move second child to first child ( temp uint)
0:7 indirect index (layout( row_major std430) buffer uint)
0:7 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:7 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:7 Constant:
0:7 0 (const uint)
0:7 ERROR: Bad unary op
( temp uint)
0:7 'i' ( temp uint)
0:7 Constant:
0:7 1 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
SPIR-V is not generated for failed compile or link
Test/baseResults/hlsl.waveordered2.frag.out 0 → 100644
hlsl.waveordered2.frag
ERROR: 0:4: 'GlobalOrderedCountIncrement' : GlobalOrderedCountIncrement() unsupported
ERROR: 1 compilation errors. No code generated.
Shader version: 500
gl_FragCoord origin is upper left
ERROR: node is still EOpNull!
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Sequence
0:3 move second child to first child ( temp uint)
0:3 'i' ( temp uint)
0:3 Constant:
0:3 42 (const uint)
0:4 Test condition and select ( temp void)
0:4 Condition
0:4 Compare Equal ( temp bool)
0:4 Constant:
0:4 0 (const uint)
0:4 ERROR: Bad unary op
( temp uint)
0:4 'i' ( temp uint)
0:4 true case
0:? Sequence
0:6 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:4 false case
0:? Sequence
0:10 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
Linked fragment stage:
Shader version: 500
gl_FragCoord origin is upper left
ERROR: node is still EOpNull!
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Sequence
0:3 move second child to first child ( temp uint)
0:3 'i' ( temp uint)
0:3 Constant:
0:3 42 (const uint)
0:4 Test condition and select ( temp void)
0:4 Condition
0:4 Compare Equal ( temp bool)
0:4 Constant:
0:4 0 (const uint)
0:4 ERROR: Bad unary op
( temp uint)
0:4 'i' ( temp uint)
0:4 true case
0:? Sequence
0:6 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:4 false case
0:? Sequence
0:10 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
SPIR-V is not generated for failed compile or link
Test/baseResults/hlsl.waveprefix.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/hlsl.wavequad.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/hlsl.wavequery.comp.out 0 → 100644
hlsl.wavequery.comp
Shader version: 500
local_size = (32, 16, 1)
0:? Sequence
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 move second child to first child ( temp uint)
0:6 indirect index (layout( row_major std430) buffer uint)
0:6 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:6 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:6 Constant:
0:6 0 (const uint)
0:6 '@gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:6 Test condition and select ( temp uint)
0:6 Condition
0:6 subgroupElect ( temp bool)
0:6 true case
0:6 '@gl_SubgroupSize' ( in uint unknown built-in variable)
0:6 false case
0:6 Constant:
0:6 0 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
Linked compute stage:
Shader version: 500
local_size = (32, 16, 1)
0:? Sequence
0:5 Function Definition: @CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:6 move second child to first child ( temp uint)
0:6 indirect index (layout( row_major std430) buffer uint)
0:6 @data: direct index for structure (layout( row_major std430) buffer implicitly-sized array of uint)
0:6 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
0:6 Constant:
0:6 0 (const uint)
0:6 '@gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:6 Test condition and select ( temp uint)
0:6 Condition
0:6 subgroupElect ( temp bool)
0:6 true case
0:6 '@gl_SubgroupSize' ( in uint unknown built-in variable)
0:6 false case
0:6 Constant:
0:6 0 (const uint)
0:5 Function Definition: CSMain( ( temp void)
0:5 Function Parameters:
0:? Sequence
0:5 Function Call: @CSMain( ( temp void)
0:? Linker Objects
0:? 'data' (layout( row_major std430) buffer block{layout( row_major std430) buffer implicitly-sized array of uint @data})
// Module Version 10000
// Generated by (magic number): 80001
// Id's are bound by 28
Capability Shader
Capability GroupNonUniform
1: ExtInstImport "GLSL.std.450"
MemoryModel Logical GLSL450
EntryPoint GLCompute 4 "CSMain" 16 21
ExecutionMode 4 LocalSize 32 16 1
Source HLSL 500
Name 4 "CSMain"
Name 6 "@CSMain("
Name 10 "data"
MemberName 10(data) 0 "@data"
Name 12 "data"
Name 16 "@gl_SubgroupInvocationID"
Name 21 "@gl_SubgroupSize"
Decorate 9 ArrayStride 4
MemberDecorate 10(data) 0 Offset 0
Decorate 10(data) BufferBlock
Decorate 12(data) DescriptorSet 0
Decorate 16(@gl_SubgroupInvocationID) BuiltIn SubgroupLocalInvocationId
Decorate 21(@gl_SubgroupSize) BuiltIn SubgroupSize
2: TypeVoid
3: TypeFunction 2
8: TypeInt 32 0
9: TypeRuntimeArray 8(int)
10(data): TypeStruct 9
11: TypePointer Uniform 10(data)
12(data): 11(ptr) Variable Uniform
13: TypeInt 32 1
14: 13(int) Constant 0
15: TypePointer Input 8(int)
16(@gl_SubgroupInvocationID): 15(ptr) Variable Input
18: TypeBool
19: 8(int) Constant 3
21(@gl_SubgroupSize): 15(ptr) Variable Input
23: 8(int) Constant 0
25: TypePointer Uniform 8(int)
4(CSMain): 2 Function None 3
5: Label
27: 2 FunctionCall 6(@CSMain()
Return
FunctionEnd
6(@CSMain(): 2 Function None 3
7: Label
17: 8(int) Load 16(@gl_SubgroupInvocationID)
20: 18(bool) GroupNonUniformElect 19
22: 8(int) Load 21(@gl_SubgroupSize)
24: 8(int) Select 20 22 23
26: 25(ptr) AccessChain 12(data) 14 17
Store 26 24
Return
FunctionEnd
Test/baseResults/hlsl.wavequery.frag.out 0 → 100644
hlsl.wavequery.frag
Shader version: 500
gl_FragCoord origin is upper left
0:? Sequence
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Test condition and select ( temp void)
0:3 Condition
0:3 '@gl_HelperInvocation' ( in bool HelperInvocation)
0:3 true case
0:? Sequence
0:5 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:3 false case
0:? Sequence
0:9 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
Linked fragment stage:
Shader version: 500
gl_FragCoord origin is upper left
0:? Sequence
0:2 Function Definition: @PixelShaderFunction( ( temp 4-component vector of float)
0:2 Function Parameters:
0:? Sequence
0:3 Test condition and select ( temp void)
0:3 Condition
0:3 '@gl_HelperInvocation' ( in bool HelperInvocation)
0:3 true case
0:? Sequence
0:5 Branch: Return with expression
0:? Constant:
0:? 1.000000
0:? 2.000000
0:? 3.000000
0:? 4.000000
0:3 false case
0:? Sequence
0:9 Branch: Return with expression
0:? Constant:
0:? 4.000000
0:? 3.000000
0:? 2.000000
0:? 1.000000
0:2 Function Definition: PixelShaderFunction( ( temp void)
0:2 Function Parameters:
0:? Sequence
0:2 move second child to first child ( temp 4-component vector of float)
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
0:2 Function Call: @PixelShaderFunction( ( temp 4-component vector of float)
0:? Linker Objects
0:? '@entryPointOutput' (layout( location=0) out 4-component vector of float)
// Module Version 10000
// Generated by (magic number): 80001
// Id's are bound by 30
Capability Shader
1: ExtInstImport "GLSL.std.450"
MemoryModel Logical GLSL450
EntryPoint Fragment 4 "PixelShaderFunction" 13 28
ExecutionMode 4 OriginUpperLeft
Source HLSL 500
Name 4 "PixelShaderFunction"
Name 9 "@PixelShaderFunction("
Name 13 "@gl_HelperInvocation"
Name 28 "@entryPointOutput"
Decorate 13(@gl_HelperInvocation) BuiltIn HelperInvocation
Decorate 28(@entryPointOutput) Location 0
2: TypeVoid
3: TypeFunction 2
6: TypeFloat 32
7: TypeVector 6(float) 4
8: TypeFunction 7(fvec4)
11: TypeBool
12: TypePointer Input 11(bool)
13(@gl_HelperInvocation): 12(ptr) Variable Input
17: 6(float) Constant 1065353216
18: 6(float) Constant 1073741824
19: 6(float) Constant 1077936128
20: 6(float) Constant 1082130432
21: 7(fvec4) ConstantComposite 17 18 19 20
24: 7(fvec4) ConstantComposite 20 19 18 17
27: TypePointer Output 7(fvec4)
28(@entryPointOutput): 27(ptr) Variable Output
4(PixelShaderFunction): 2 Function None 3
5: Label
29: 7(fvec4) FunctionCall 9(@PixelShaderFunction()
Store 28(@entryPointOutput) 29
Return
FunctionEnd
9(@PixelShaderFunction(): 7(fvec4) Function None 8
10: Label
14: 11(bool) Load 13(@gl_HelperInvocation)
SelectionMerge 16 None
BranchConditional 14 15 23
15: Label
ReturnValue 21
23: Label
ReturnValue 24
16: Label
26: 7(fvec4) Undef
ReturnValue 26
FunctionEnd
Test/baseResults/hlsl.wavereduction.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/hlsl.wavebroadcast.comp 0 → 100644
struct Types
{
uint4 u;
int4 i;
float4 f;
double4 d;
};
RWStructuredBuffer<Types> data;
[numthreads(32, 16, 1)]
void CSMain(uint3 dti : SV_DispatchThreadID)
{
data[dti.x].u = WaveReadLaneAt(data[dti.x].u, 13);
data[dti.x].u.x = WaveReadLaneAt(data[dti.x].u.x, 13);
data[dti.x].u.xy = WaveReadLaneAt(data[dti.x].u.xy, 13);
data[dti.x].u.xyz = WaveReadLaneAt(data[dti.x].u.xyz, 13);
data[dti.x].i = WaveReadLaneAt(data[dti.x].i, 13);
data[dti.x].i.x = WaveReadLaneAt(data[dti.x].i.x, 13);
data[dti.x].i.xy = WaveReadLaneAt(data[dti.x].i.xy, 13);
data[dti.x].i.xyz = WaveReadLaneAt(data[dti.x].i.xyz, 13);
data[dti.x].f = WaveReadLaneAt(data[dti.x].f, 13);
data[dti.x].f.x = WaveReadLaneAt(data[dti.x].f.x, 13);
data[dti.x].f.xy = WaveReadLaneAt(data[dti.x].f.xy, 13);
data[dti.x].f.xyz = WaveReadLaneAt(data[dti.x].f.xyz, 13);
data[dti.x].d = WaveReadFirstLane(data[dti.x].d);
data[dti.x].d.x = WaveReadFirstLane(data[dti.x].d.x);
data[dti.x].d.xy = WaveReadFirstLane(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveReadFirstLane(data[dti.x].d.xyz);
data[dti.x].u = WaveReadFirstLane(data[dti.x].u);
data[dti.x].u.x = WaveReadFirstLane(data[dti.x].u.x);
data[dti.x].u.xy = WaveReadFirstLane(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveReadFirstLane(data[dti.x].u.xyz);
data[dti.x].i = WaveReadFirstLane(data[dti.x].i);
data[dti.x].i.x = WaveReadFirstLane(data[dti.x].i.x);
data[dti.x].i.xy = WaveReadFirstLane(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveReadFirstLane(data[dti.x].i.xyz);
data[dti.x].f = WaveReadFirstLane(data[dti.x].f);
data[dti.x].f.x = WaveReadFirstLane(data[dti.x].f.x);
data[dti.x].f.xy = WaveReadFirstLane(data[dti.x].f.xy);
data[dti.x].f.xyz = WaveReadFirstLane(data[dti.x].f.xyz);
data[dti.x].d = WaveReadFirstLane(data[dti.x].d);
data[dti.x].d.x = WaveReadFirstLane(data[dti.x].d.x);
data[dti.x].d.xy = WaveReadFirstLane(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveReadFirstLane(data[dti.x].d.xyz);
}
Test/hlsl.waveordered.comp 0 → 100644
RWStructuredBuffer<uint> data;
[numthreads(32, 16, 1)]
void CSMain()
{
data[WaveGetOrderedIndex()] = 1;
}
Test/hlsl.waveordered.frag 0 → 100644
float4 PixelShaderFunction() : COLOR0
{
if (0 == WaveGetOrderedIndex())
{
return float4(1, 2, 3, 4);
}
else
{
return float4(4, 3, 2, 1);
}
}
Test/hlsl.waveordered2.comp 0 → 100644
RWStructuredBuffer<uint> data;
[numthreads(32, 16, 1)]
void CSMain()
{
uint i = 42;
data[GlobalOrderedCountIncrement(i)] = 1;
}
Test/hlsl.waveordered2.frag 0 → 100644
float4 PixelShaderFunction() : COLOR0
{
uint i = 42;
if (0 == GlobalOrderedCountIncrement(i))
{
return float4(1, 2, 3, 4);
}
else
{
return float4(4, 3, 2, 1);
}
}
Test/hlsl.waveprefix.comp 0 → 100644
struct Types
{
uint4 u;
int4 i;
float4 f;
double4 d;
};
RWStructuredBuffer<Types> data;
[numthreads(32, 16, 1)]
void CSMain(uint3 dti : SV_DispatchThreadID)
{
data[dti.x].u = WavePrefixSum(data[dti.x].u);
data[dti.x].u.x = WavePrefixSum(data[dti.x].u.x);
data[dti.x].u.xy = WavePrefixSum(data[dti.x].u.xy);
data[dti.x].u.xyz = WavePrefixSum(data[dti.x].u.xyz);
data[dti.x].i = WavePrefixSum(data[dti.x].i);
data[dti.x].i.x = WavePrefixSum(data[dti.x].i.x);
data[dti.x].i.xy = WavePrefixSum(data[dti.x].i.xy);
data[dti.x].i.xyz = WavePrefixSum(data[dti.x].i.xyz);
data[dti.x].f = WavePrefixSum(data[dti.x].f);
data[dti.x].f.x = WavePrefixSum(data[dti.x].f.x);
data[dti.x].f.xy = WavePrefixSum(data[dti.x].f.xy);
data[dti.x].f.xyz = WavePrefixSum(data[dti.x].f.xyz);
data[dti.x].d = WavePrefixSum(data[dti.x].d);
data[dti.x].d.x = WavePrefixSum(data[dti.x].d.x);
data[dti.x].d.xy = WavePrefixSum(data[dti.x].d.xy);
data[dti.x].d.xyz = WavePrefixSum(data[dti.x].d.xyz);
data[dti.x].u = WavePrefixProduct(data[dti.x].u);
data[dti.x].u.x = WavePrefixProduct(data[dti.x].u.x);
data[dti.x].u.xy = WavePrefixProduct(data[dti.x].u.xy);
data[dti.x].u.xyz = WavePrefixProduct(data[dti.x].u.xyz);
data[dti.x].i = WavePrefixProduct(data[dti.x].i);
data[dti.x].i.x = WavePrefixProduct(data[dti.x].i.x);
data[dti.x].i.xy = WavePrefixProduct(data[dti.x].i.xy);
data[dti.x].i.xyz = WavePrefixProduct(data[dti.x].i.xyz);
data[dti.x].f = WavePrefixProduct(data[dti.x].f);
data[dti.x].f.x = WavePrefixProduct(data[dti.x].f.x);
data[dti.x].f.xy = WavePrefixProduct(data[dti.x].f.xy);
data[dti.x].f.xyz = WavePrefixProduct(data[dti.x].f.xyz);
data[dti.x].d = WavePrefixProduct(data[dti.x].d);
data[dti.x].d.x = WavePrefixProduct(data[dti.x].d.x);
data[dti.x].d.xy = WavePrefixProduct(data[dti.x].d.xy);
data[dti.x].d.xyz = WavePrefixProduct(data[dti.x].d.xyz);
}
Test/hlsl.wavequad.comp 0 → 100644
struct Types
{
uint4 u;
int4 i;
float4 f;
double4 d;
};
RWStructuredBuffer<Types> data;
[numthreads(32, 16, 1)]
void CSMain(uint3 dti : SV_DispatchThreadID)
{
data[dti.x].u = QuadReadLaneAt(data[dti.x].u, 0);
data[dti.x].u.x = QuadReadLaneAt(data[dti.x].u.x, 0);
data[dti.x].u.xy = QuadReadLaneAt(data[dti.x].u.xy, 0);
data[dti.x].u.xyz = QuadReadLaneAt(data[dti.x].u.xyz, 0);
data[dti.x].i = QuadReadLaneAt(data[dti.x].i, 0);
data[dti.x].i.x = QuadReadLaneAt(data[dti.x].i.x, 0);
data[dti.x].i.xy = QuadReadLaneAt(data[dti.x].i.xy, 0);
data[dti.x].i.xyz = QuadReadLaneAt(data[dti.x].i.xyz, 0);
data[dti.x].f = QuadReadLaneAt(data[dti.x].f, 0);
data[dti.x].f.x = QuadReadLaneAt(data[dti.x].f.x, 0);
data[dti.x].f.xy = QuadReadLaneAt(data[dti.x].f.xy, 0);
data[dti.x].f.xyz = QuadReadLaneAt(data[dti.x].f.xyz, 0);
data[dti.x].d = QuadReadLaneAt(data[dti.x].d, 0);
data[dti.x].d.x = QuadReadLaneAt(data[dti.x].d.x, 0);
data[dti.x].d.xy = QuadReadLaneAt(data[dti.x].d.xy, 0);
data[dti.x].d.xyz = QuadReadLaneAt(data[dti.x].d.xyz, 0);
data[dti.x].u = QuadReadLaneAt(data[dti.x].u, 1);
data[dti.x].u.x = QuadReadLaneAt(data[dti.x].u.x, 1);
data[dti.x].u.xy = QuadReadLaneAt(data[dti.x].u.xy, 1);
data[dti.x].u.xyz = QuadReadLaneAt(data[dti.x].u.xyz, 1);
data[dti.x].i = QuadReadLaneAt(data[dti.x].i, 1);
data[dti.x].i.x = QuadReadLaneAt(data[dti.x].i.x, 1);
data[dti.x].i.xy = QuadReadLaneAt(data[dti.x].i.xy, 1);
data[dti.x].i.xyz = QuadReadLaneAt(data[dti.x].i.xyz, 1);
data[dti.x].f = QuadReadLaneAt(data[dti.x].f, 1);
data[dti.x].f.x = QuadReadLaneAt(data[dti.x].f.x, 1);
data[dti.x].f.xy = QuadReadLaneAt(data[dti.x].f.xy, 1);
data[dti.x].f.xyz = QuadReadLaneAt(data[dti.x].f.xyz, 1);
data[dti.x].d = QuadReadLaneAt(data[dti.x].d, 1);
data[dti.x].d.x = QuadReadLaneAt(data[dti.x].d.x, 1);
data[dti.x].d.xy = QuadReadLaneAt(data[dti.x].d.xy, 1);
data[dti.x].d.xyz = QuadReadLaneAt(data[dti.x].d.xyz, 1);
data[dti.x].u = QuadReadLaneAt(data[dti.x].u, 2);
data[dti.x].u.x = QuadReadLaneAt(data[dti.x].u.x, 2);
data[dti.x].u.xy = QuadReadLaneAt(data[dti.x].u.xy, 2);
data[dti.x].u.xyz = QuadReadLaneAt(data[dti.x].u.xyz, 2);
data[dti.x].i = QuadReadLaneAt(data[dti.x].i, 2);
data[dti.x].i.x = QuadReadLaneAt(data[dti.x].i.x, 2);
data[dti.x].i.xy = QuadReadLaneAt(data[dti.x].i.xy, 2);
data[dti.x].i.xyz = QuadReadLaneAt(data[dti.x].i.xyz, 2);
data[dti.x].f = QuadReadLaneAt(data[dti.x].f, 2);
data[dti.x].f.x = QuadReadLaneAt(data[dti.x].f.x, 2);
data[dti.x].f.xy = QuadReadLaneAt(data[dti.x].f.xy, 2);
data[dti.x].f.xyz = QuadReadLaneAt(data[dti.x].f.xyz, 2);
data[dti.x].d = QuadReadLaneAt(data[dti.x].d, 2);
data[dti.x].d.x = QuadReadLaneAt(data[dti.x].d.x, 2);
data[dti.x].d.xy = QuadReadLaneAt(data[dti.x].d.xy, 2);
data[dti.x].d.xyz = QuadReadLaneAt(data[dti.x].d.xyz, 2);
data[dti.x].u = QuadReadLaneAt(data[dti.x].u, 3);
data[dti.x].u.x = QuadReadLaneAt(data[dti.x].u.x, 3);
data[dti.x].u.xy = QuadReadLaneAt(data[dti.x].u.xy, 3);
data[dti.x].u.xyz = QuadReadLaneAt(data[dti.x].u.xyz, 3);
data[dti.x].i = QuadReadLaneAt(data[dti.x].i, 3);
data[dti.x].i.x = QuadReadLaneAt(data[dti.x].i.x, 3);
data[dti.x].i.xy = QuadReadLaneAt(data[dti.x].i.xy, 3);
data[dti.x].i.xyz = QuadReadLaneAt(data[dti.x].i.xyz, 3);
data[dti.x].f = QuadReadLaneAt(data[dti.x].f, 3);
data[dti.x].f.x = QuadReadLaneAt(data[dti.x].f.x, 3);
data[dti.x].f.xy = QuadReadLaneAt(data[dti.x].f.xy, 3);
data[dti.x].f.xyz = QuadReadLaneAt(data[dti.x].f.xyz, 3);
data[dti.x].d = QuadReadLaneAt(data[dti.x].d, 3);
data[dti.x].d.x = QuadReadLaneAt(data[dti.x].d.x, 3);
data[dti.x].d.xy = QuadReadLaneAt(data[dti.x].d.xy, 3);
data[dti.x].d.xyz = QuadReadLaneAt(data[dti.x].d.xyz, 3);
data[dti.x].u = QuadSwapX(data[dti.x].u);
data[dti.x].u.x = QuadSwapX(data[dti.x].u.x);
data[dti.x].u.xy = QuadSwapX(data[dti.x].u.xy);
data[dti.x].u.xyz = QuadSwapX(data[dti.x].u.xyz);
data[dti.x].i = QuadSwapX(data[dti.x].i);
data[dti.x].i.x = QuadSwapX(data[dti.x].i.x);
data[dti.x].i.xy = QuadSwapX(data[dti.x].i.xy);
data[dti.x].i.xyz = QuadSwapX(data[dti.x].i.xyz);
data[dti.x].f = QuadSwapX(data[dti.x].f);
data[dti.x].f.x = QuadSwapX(data[dti.x].f.x);
data[dti.x].f.xy = QuadSwapX(data[dti.x].f.xy);
data[dti.x].f.xyz = QuadSwapX(data[dti.x].f.xyz);
data[dti.x].d = QuadSwapX(data[dti.x].d);
data[dti.x].d.x = QuadSwapX(data[dti.x].d.x);
data[dti.x].d.xy = QuadSwapX(data[dti.x].d.xy);
data[dti.x].d.xyz = QuadSwapX(data[dti.x].d.xyz);
data[dti.x].u = QuadSwapY(data[dti.x].u);
data[dti.x].u.x = QuadSwapY(data[dti.x].u.x);
data[dti.x].u.xy = QuadSwapY(data[dti.x].u.xy);
data[dti.x].u.xyz = QuadSwapY(data[dti.x].u.xyz);
data[dti.x].i = QuadSwapY(data[dti.x].i);
data[dti.x].i.x = QuadSwapY(data[dti.x].i.x);
data[dti.x].i.xy = QuadSwapY(data[dti.x].i.xy);
data[dti.x].i.xyz = QuadSwapY(data[dti.x].i.xyz);
data[dti.x].f = QuadSwapY(data[dti.x].f);
data[dti.x].f.x = QuadSwapY(data[dti.x].f.x);
data[dti.x].f.xy = QuadSwapY(data[dti.x].f.xy);
data[dti.x].f.xyz = QuadSwapY(data[dti.x].f.xyz);
data[dti.x].d = QuadSwapY(data[dti.x].d);
data[dti.x].d.x = QuadSwapY(data[dti.x].d.x);
data[dti.x].d.xy = QuadSwapY(data[dti.x].d.xy);
data[dti.x].d.xyz = QuadSwapY(data[dti.x].d.xyz);
}
Test/hlsl.wavequery.comp 0 → 100644
RWStructuredBuffer<uint> data;
[numthreads(32, 16, 1)]
void CSMain()
{
data[WaveGetLaneIndex()] = (WaveOnce()) ? WaveGetLaneCount() : 0;
}
Test/hlsl.wavequery.frag 0 → 100644
float4 PixelShaderFunction() : COLOR0
{
if (WaveIsHelperLane())
{
return float4(1, 2, 3, 4);
}
else
{
return float4(4, 3, 2, 1);
}
}
Test/hlsl.wavereduction.comp 0 → 100644
struct Types
{
uint4 u;
int4 i;
float4 f;
double4 d;
};
RWStructuredBuffer<Types> data;
[numthreads(32, 16, 1)]
void CSMain(uint3 dti : SV_DispatchThreadID)
{
data[dti.x].u = WaveAllSum(data[dti.x].u);
data[dti.x].u.x = WaveAllSum(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllSum(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllSum(data[dti.x].u.xyz);
data[dti.x].i = WaveAllSum(data[dti.x].i);
data[dti.x].i.x = WaveAllSum(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllSum(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllSum(data[dti.x].i.xyz);
data[dti.x].f = WaveAllSum(data[dti.x].f);
data[dti.x].f.x = WaveAllSum(data[dti.x].f.x);
data[dti.x].f.xy = WaveAllSum(data[dti.x].f.xy);
data[dti.x].f.xyz = WaveAllSum(data[dti.x].f.xyz);
data[dti.x].d = WaveAllSum(data[dti.x].d);
data[dti.x].d.x = WaveAllSum(data[dti.x].d.x);
data[dti.x].d.xy = WaveAllSum(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveAllSum(data[dti.x].d.xyz);
data[dti.x].u = WaveAllProduct(data[dti.x].u);
data[dti.x].u.x = WaveAllProduct(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllProduct(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllProduct(data[dti.x].u.xyz);
data[dti.x].i = WaveAllProduct(data[dti.x].i);
data[dti.x].i.x = WaveAllProduct(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllProduct(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllProduct(data[dti.x].i.xyz);
data[dti.x].f = WaveAllProduct(data[dti.x].f);
data[dti.x].f.x = WaveAllProduct(data[dti.x].f.x);
data[dti.x].f.xy = WaveAllProduct(data[dti.x].f.xy);
data[dti.x].f.xyz = WaveAllProduct(data[dti.x].f.xyz);
data[dti.x].d = WaveAllProduct(data[dti.x].d);
data[dti.x].d.x = WaveAllProduct(data[dti.x].d.x);
data[dti.x].d.xy = WaveAllProduct(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveAllProduct(data[dti.x].d.xyz);
data[dti.x].u = WaveAllMin(data[dti.x].u);
data[dti.x].u.x = WaveAllMin(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllMin(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllMin(data[dti.x].u.xyz);
data[dti.x].i = WaveAllMin(data[dti.x].i);
data[dti.x].i.x = WaveAllMin(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllMin(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllMin(data[dti.x].i.xyz);
data[dti.x].f = WaveAllMin(data[dti.x].f);
data[dti.x].f.x = WaveAllMin(data[dti.x].f.x);
data[dti.x].f.xy = WaveAllMin(data[dti.x].f.xy);
data[dti.x].f.xyz = WaveAllMin(data[dti.x].f.xyz);
data[dti.x].d = WaveAllMin(data[dti.x].d);
data[dti.x].d.x = WaveAllMin(data[dti.x].d.x);
data[dti.x].d.xy = WaveAllMin(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveAllMin(data[dti.x].d.xyz);
data[dti.x].u = WaveAllMax(data[dti.x].u);
data[dti.x].u.x = WaveAllMax(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllMax(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllMax(data[dti.x].u.xyz);
data[dti.x].i = WaveAllMax(data[dti.x].i);
data[dti.x].i.x = WaveAllMax(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllMax(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllMax(data[dti.x].i.xyz);
data[dti.x].f = WaveAllMax(data[dti.x].f);
data[dti.x].f.x = WaveAllMax(data[dti.x].f.x);
data[dti.x].f.xy = WaveAllMax(data[dti.x].f.xy);
data[dti.x].f.xyz = WaveAllMax(data[dti.x].f.xyz);
data[dti.x].d = WaveAllMax(data[dti.x].d);
data[dti.x].d.x = WaveAllMax(data[dti.x].d.x);
data[dti.x].d.xy = WaveAllMax(data[dti.x].d.xy);
data[dti.x].d.xyz = WaveAllMax(data[dti.x].d.xyz);
data[dti.x].u = WaveAllBitAnd(data[dti.x].u);
data[dti.x].u.x = WaveAllBitAnd(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllBitAnd(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllBitAnd(data[dti.x].u.xyz);
data[dti.x].i = WaveAllBitAnd(data[dti.x].i);
data[dti.x].i.x = WaveAllBitAnd(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllBitAnd(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllBitAnd(data[dti.x].i.xyz);
data[dti.x].u = WaveAllBitOr(data[dti.x].u);
data[dti.x].u.x = WaveAllBitOr(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllBitOr(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllBitOr(data[dti.x].u.xyz);
data[dti.x].i = WaveAllBitOr(data[dti.x].i);
data[dti.x].i.x = WaveAllBitOr(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllBitOr(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllBitOr(data[dti.x].i.xyz);
data[dti.x].u = WaveAllBitXor(data[dti.x].u);
data[dti.x].u.x = WaveAllBitXor(data[dti.x].u.x);
data[dti.x].u.xy = WaveAllBitXor(data[dti.x].u.xy);
data[dti.x].u.xyz = WaveAllBitXor(data[dti.x].u.xyz);
data[dti.x].i = WaveAllBitXor(data[dti.x].i);
data[dti.x].i.x = WaveAllBitXor(data[dti.x].i.x);
data[dti.x].i.xy = WaveAllBitXor(data[dti.x].i.xy);
data[dti.x].i.xyz = WaveAllBitXor(data[dti.x].i.xyz);
}
Test/hlsl.wavevote.comp 0 → 100644
RWStructuredBuffer<uint64_t> data;
[numthreads(32, 16, 1)]
void CSMain(uint3 dti : SV_DispatchThreadID)
{
data[dti.x] = WaveBallot(WaveAnyTrue(dti.x == 0));
data[dti.y] = WaveBallot(WaveAllTrue(dti.y == 0));
data[dti.z] = WaveBallot(WaveAllEqual(dti.z == 0));
}
glslang/Include/intermediate.h
...@@ -711,6 +711,14 @@ enum TOperator { ...@@ -711,6 +711,14 @@ enum TOperator {
// matrix // matrix
EOpMatrixSwizzle, // select multiple matrix components (non-column) EOpMatrixSwizzle, // select multiple matrix components (non-column)
// SM6 wave ops
EOpWaveGetLaneCount, // Will decompose to gl_SubgroupSize.
EOpWaveGetLaneIndex, // Will decompose to gl_SubgroupInvocationID.
EOpWaveIsHelperLane, // Will decompose to gl_HelperInvocation.
EOpWaveBallot, // Will decompose to subgroupBallot.
EOpWaveGetOrderedIndex, // Will decompose to an equation containing gl_SubgroupID.
EOpGlobalOrderedCountIncrement, // Will nice error.
}; };
class TIntermTraverser; class TIntermTraverser;
......
glslang/MachineIndependent/intermOut.cpp
...@@ -697,11 +697,61 @@ bool TOutputTraverser::visitAggregate(TVisit /* visit */, TIntermAggregate* node ...@@ -697,11 +697,61 @@ bool TOutputTraverser::visitAggregate(TVisit /* visit */, TIntermAggregate* node
case EOpWorkgroupMemoryBarrier: out.debug << "WorkgroupMemoryBarrier"; break; case EOpWorkgroupMemoryBarrier: out.debug << "WorkgroupMemoryBarrier"; break;
case EOpWorkgroupMemoryBarrierWithGroupSync: out.debug << "WorkgroupMemoryBarrierWithGroupSync"; break; case EOpWorkgroupMemoryBarrierWithGroupSync: out.debug << "WorkgroupMemoryBarrierWithGroupSync"; break;
case EOpSubgroupBarrier: out.debug << "subgroupBarrier"; break; case EOpSubgroupBarrier: out.debug << "subgroupBarrier"; break;
case EOpSubgroupMemoryBarrier: out.debug << "subgroupMemoryBarrier"; break; case EOpSubgroupMemoryBarrier: out.debug << "subgroupMemoryBarrier"; break;
case EOpSubgroupMemoryBarrierBuffer: out.debug << "subgroupMemoryBarrierBuffer"; break; case EOpSubgroupMemoryBarrierBuffer: out.debug << "subgroupMemoryBarrierBuffer"; break;
case EOpSubgroupMemoryBarrierImage: out.debug << "subgroupMemoryBarrierImage"; break; case EOpSubgroupMemoryBarrierImage: out.debug << "subgroupMemoryBarrierImage"; break;
case EOpSubgroupMemoryBarrierShared: out.debug << "subgroupMemoryBarrierShared"; break; case EOpSubgroupMemoryBarrierShared: out.debug << "subgroupMemoryBarrierShared"; break;
case EOpSubgroupElect: out.debug << "subgroupElect"; break;
case EOpSubgroupAll: out.debug << "subgroupAll"; break;
case EOpSubgroupAny: out.debug << "subgroupAny"; break;
case EOpSubgroupAllEqual: out.debug << "subgroupAllEqual"; break;
case EOpSubgroupBroadcast: out.debug << "subgroupBroadcast"; break;
case EOpSubgroupBroadcastFirst: out.debug << "subgroupBroadcastFirst"; break;
case EOpSubgroupBallot: out.debug << "subgroupBallot"; break;
case EOpSubgroupInverseBallot: out.debug << "subgroupInverseBallot"; break;
case EOpSubgroupBallotBitExtract: out.debug << "subgroupBallotBitExtract"; break;
case EOpSubgroupBallotBitCount: out.debug << "subgroupBallotBitCount"; break;
case EOpSubgroupBallotInclusiveBitCount: out.debug << "subgroupBallotInclusiveBitCount"; break;
case EOpSubgroupBallotExclusiveBitCount: out.debug << "subgroupBallotExclusiveBitCount"; break;
case EOpSubgroupBallotFindLSB: out.debug << "subgroupBallotFindLSB"; break;
case EOpSubgroupBallotFindMSB: out.debug << "subgroupBallotFindMSB"; break;
case EOpSubgroupShuffle: out.debug << "subgroupShuffle"; break;
case EOpSubgroupShuffleXor: out.debug << "subgroupShuffleXor"; break;
case EOpSubgroupShuffleUp: out.debug << "subgroupShuffleUp"; break;
case EOpSubgroupShuffleDown: out.debug << "subgroupShuffleDown"; break;
case EOpSubgroupAdd: out.debug << "subgroupAdd"; break;
case EOpSubgroupMul: out.debug << "subgroupMul"; break;
case EOpSubgroupMin: out.debug << "subgroupMin"; break;
case EOpSubgroupMax: out.debug << "subgroupMax"; break;
case EOpSubgroupAnd: out.debug << "subgroupAnd"; break;
case EOpSubgroupOr: out.debug << "subgroupOr"; break;
case EOpSubgroupXor: out.debug << "subgroupXor"; break;
case EOpSubgroupInclusiveAdd: out.debug << "subgroupInclusiveAdd"; break;
case EOpSubgroupInclusiveMul: out.debug << "subgroupInclusiveMul"; break;
case EOpSubgroupInclusiveMin: out.debug << "subgroupInclusiveMin"; break;
case EOpSubgroupInclusiveMax: out.debug << "subgroupInclusiveMax"; break;
case EOpSubgroupInclusiveAnd: out.debug << "subgroupInclusiveAnd"; break;
case EOpSubgroupInclusiveOr: out.debug << "subgroupInclusiveOr"; break;
case EOpSubgroupInclusiveXor: out.debug << "subgroupInclusiveXor"; break;
case EOpSubgroupExclusiveAdd: out.debug << "subgroupExclusiveAdd"; break;
case EOpSubgroupExclusiveMul: out.debug << "subgroupExclusiveMul"; break;
case EOpSubgroupExclusiveMin: out.debug << "subgroupExclusiveMin"; break;
case EOpSubgroupExclusiveMax: out.debug << "subgroupExclusiveMax"; break;
case EOpSubgroupExclusiveAnd: out.debug << "subgroupExclusiveAnd"; break;
case EOpSubgroupExclusiveOr: out.debug << "subgroupExclusiveOr"; break;
case EOpSubgroupExclusiveXor: out.debug << "subgroupExclusiveXor"; break;
case EOpSubgroupClusteredAdd: out.debug << "subgroupClusteredAdd"; break;
case EOpSubgroupClusteredMul: out.debug << "subgroupClusteredMul"; break;
case EOpSubgroupClusteredMin: out.debug << "subgroupClusteredMin"; break;
case EOpSubgroupClusteredMax: out.debug << "subgroupClusteredMax"; break;
case EOpSubgroupClusteredAnd: out.debug << "subgroupClusteredAnd"; break;
case EOpSubgroupClusteredOr: out.debug << "subgroupClusteredOr"; break;
case EOpSubgroupClusteredXor: out.debug << "subgroupClusteredXor"; break;
case EOpSubgroupQuadBroadcast: out.debug << "subgroupQuadBroadcast"; break;
case EOpSubgroupQuadSwapHorizontal: out.debug << "subgroupQuadSwapHorizontal"; break;
case EOpSubgroupQuadSwapVertical: out.debug << "subgroupQuadSwapVertical"; break;
case EOpSubgroupQuadSwapDiagonal: out.debug << "subgroupQuadSwapDiagonal"; break;
default: out.debug.message(EPrefixError, "Bad aggregation op"); default: out.debug.message(EPrefixError, "Bad aggregation op");
} }
......
gtests/Hlsl.FromFile.cpp
...@@ -282,6 +282,17 @@ INSTANTIATE_TEST_CASE_P( ...@@ -282,6 +282,17 @@ INSTANTIATE_TEST_CASE_P(
{"hlsl.type.identifier.frag", "main"}, {"hlsl.type.identifier.frag", "main"},
{"hlsl.typeGraphCopy.vert", "main"}, {"hlsl.typeGraphCopy.vert", "main"},
{"hlsl.typedef.frag", "PixelShaderFunction"}, {"hlsl.typedef.frag", "PixelShaderFunction"},
{"hlsl.wavequery.comp", "CSMain"},
{"hlsl.wavequery.frag", "PixelShaderFunction"},
{"hlsl.wavevote.comp", "CSMain"},
{"hlsl.wavebroadcast.comp", "CSMain"},
{"hlsl.wavereduction.comp", "CSMain"},
{"hlsl.waveprefix.comp", "CSMain"},
{"hlsl.wavequad.comp", "CSMain"},
{"hlsl.waveordered.comp", "CSMain"},
{"hlsl.waveordered2.comp", "CSMain"},
{"hlsl.waveordered.frag", "PixelShaderFunction"},
{"hlsl.waveordered2.frag", "PixelShaderFunction"},
{"hlsl.whileLoop.frag", "PixelShaderFunction"}, {"hlsl.whileLoop.frag", "PixelShaderFunction"},
{"hlsl.void.frag", "PixelShaderFunction"}, {"hlsl.void.frag", "PixelShaderFunction"},
}), }),
......
gtests/main.cpp
...@@ -54,7 +54,7 @@ int main(int argc, char** argv) ...@@ -54,7 +54,7 @@ int main(int argc, char** argv)
glslangtest::GlobalTestSettings.updateMode = true; glslangtest::GlobalTestSettings.updateMode = true;
} }
if (std::string("--test-root") == argv[i]) { if (std::string("--test-root") == argv[i]) {
// Allow the user set the tets root directory. This is useful // Allow the user set the test root directory. This is useful
// for testing with files from another source tree. // for testing with files from another source tree.
if (i + 1 < argc) { if (i + 1 < argc) {
glslangtest::GlobalTestSettings.testRoot = argv[i + 1]; glslangtest::GlobalTestSettings.testRoot = argv[i + 1];
......
hlsl/hlslGrammar.cpp
...@@ -1457,6 +1457,10 @@ bool HlslGrammar::acceptType(TType& type, TIntermNode*& nodeList) ...@@ -1457,6 +1457,10 @@ bool HlslGrammar::acceptType(TType& type, TIntermNode*& nodeList)
new(&type) TType(EbtUint, EvqTemporary, 4); new(&type) TType(EbtUint, EvqTemporary, 4);
break; break;
case EHTokUint64:
new(&type) TType(EbtUint64);
break;
case EHTokBool: case EHTokBool:
new(&type) TType(EbtBool); new(&type) TType(EbtBool);
break; break;
......
hlsl/hlslParseHelper.cpp
...@@ -3600,6 +3600,22 @@ void HlslParseContext::decomposeIntrinsic(const TSourceLoc& loc, TIntermTyped*& ...@@ -3600,6 +3600,22 @@ void HlslParseContext::decomposeIntrinsic(const TSourceLoc& loc, TIntermTyped*&
return imageAggregate != nullptr && imageAggregate->getOp() == EOpImageLoad; return imageAggregate != nullptr && imageAggregate->getOp() == EOpImageLoad;
}; };
const auto lookupBuiltinVariable = [&](const char* name, TBuiltInVariable builtin, TType& type) -> TIntermTyped* {
TSymbol* symbol = symbolTable.find(name);
if (nullptr == symbol) {
type.getQualifier().builtIn = builtin;
TVariable* variable = new TVariable(new TString(name), type);
symbolTable.insert(*variable);
symbol = symbolTable.find(name);
assert(symbol && "Inserted symbol could not be found!");
}
return intermediate.addSymbol(*(symbol->getAsVariable()), loc);
};
// HLSL intrinsics can be pass through to native AST opcodes, or decomposed here to existing AST // HLSL intrinsics can be pass through to native AST opcodes, or decomposed here to existing AST
// opcodes for compatibility with existing software stacks. // opcodes for compatibility with existing software stacks.
static const bool decomposeHlslIntrinsics = true; static const bool decomposeHlslIntrinsics = true;
...@@ -4126,7 +4142,104 @@ void HlslParseContext::decomposeIntrinsic(const TSourceLoc& loc, TIntermTyped*& ...@@ -4126,7 +4142,104 @@ void HlslParseContext::decomposeIntrinsic(const TSourceLoc& loc, TIntermTyped*&
break; break;
} }
case EOpWaveGetLaneCount:
{
// Mapped to gl_SubgroupSize builtin (We preprend @ to the symbol
// so that it inhabits the symbol table, but has a user-invalid name
// in-case some source HLSL defined the symbol also).
TType type(EbtUint, EvqVaryingIn);
node = lookupBuiltinVariable("@gl_SubgroupSize", EbvSubgroupSize2, type);
break;
}
case EOpWaveGetLaneIndex:
{
// Mapped to gl_SubgroupInvocationID builtin (We preprend @ to the
// symbol so that it inhabits the symbol table, but has a
// user-invalid name in-case some source HLSL defined the symbol
// also).
TType type(EbtUint, EvqVaryingIn);
node = lookupBuiltinVariable("@gl_SubgroupInvocationID", EbvSubgroupInvocation2, type);
break;
}
case EOpWaveIsHelperLane:
{
// Mapped to gl_HelperInvocation builtin (We preprend @ to the symbol
// so that it inhabits the symbol table, but has a user-invalid name
// in-case some source HLSL defined the symbol also).
TType type(EbtBool, EvqVaryingIn);
node = lookupBuiltinVariable("@gl_HelperInvocation", EbvHelperInvocation, type);
break;
}
case EOpWaveBallot:
{
// Mapped to subgroupBallot() builtin (NOTE: if an IHV has
// a subgroup size > 64 these wave ops will not work for them!)
// uvec4 type.
TType uvec4Type(EbtUint, EvqTemporary, 4);
// Get the uvec4 return from subgroupBallot().
TIntermTyped* res = intermediate.addBuiltInFunctionCall(loc,
EOpSubgroupBallot, true, arguments, uvec4Type);
// And extract a uvec2 for the two highest components.
TIntermTyped* xy = handleDotDereference(loc, res, "xy");
// uint64_t type.
TType uint64Type(EbtUint64, EvqTemporary);
// And bitcast the result for a uint64_t
node = intermediate.addBuiltInFunctionCall(loc,
EOpPackUint2x32, true, xy, uint64Type);
break;
}
case EOpWaveGetOrderedIndex:
{
if (language == EShLangFragment) {
// NOTE: For HLSL SM6.0 this should work for PS too, but the current GLSL extensions don't allow this.
error(loc, "WaveGetOrderedIndex() unsupported in a pixel/fragment shader", "WaveGetOrderedIndex", "");
break;
}
TType uintType(EbtUint, EvqVaryingIn);
TIntermTyped* subgroupID = lookupBuiltinVariable("@gl_SubgroupID", EbvSubgroupID, uintType);
TIntermTyped* numSubgroups = lookupBuiltinVariable("@gl_NumSubgroups", EbvNumSubgroups, uintType);
TType uvec3Type(EbtUint, EvqVaryingIn, 3);
TIntermTyped* numWorkGroups = lookupBuiltinVariable("@gl_NumWorkGroups", EbvNumWorkGroups, uvec3Type);
TIntermTyped* workGroupID = lookupBuiltinVariable("@gl_WorkGroupID", EbvWorkGroupId, uvec3Type);
//x & y components of gl_NumWorkGroups
TIntermTyped* numWorkGroupsX = handleDotDereference(loc, numWorkGroups, "x");
TIntermTyped* numWorkGroupsY = handleDotDereference(loc, numWorkGroups, "y");
// x & y components of globalSize
TIntermTyped* globalSizeX = handleBinaryMath(loc, "mul", EOpMul, numSubgroups, numWorkGroupsX);
TIntermTyped* globalSizeY = numWorkGroupsY;
// x, y & z components of gl_WorkGroupID
TIntermTyped* workGroupX = handleDotDereference(loc, workGroupID, "x");
TIntermTyped* workGroupY = handleDotDereference(loc, workGroupID, "y");
TIntermTyped* workGroupZ = handleDotDereference(loc, workGroupID, "z");
// We're going to build up the following variables to get a uniquely ordered ID:
// (globalSize.y * gl_WorkGroupID.z + gl_WorkGroupID.y) * globalSize.x + gl_WorkGroupID.x + gl_SubgroupID
node = handleBinaryMath(loc, "mul", EOpMul, globalSizeY, workGroupZ);
node = handleBinaryMath(loc, "add", EOpAdd, node, workGroupY);
node = handleBinaryMath(loc, "mul", EOpMul, node, globalSizeX);
node = handleBinaryMath(loc, "add", EOpAdd, node, workGroupX);
node = handleBinaryMath(loc, "add", EOpAdd, node, subgroupID);
break;
}
case EOpGlobalOrderedCountIncrement:
{
// NOTE: For HLSL SM6.0 this should work, but the current GLSL extensions don't allow this.
error(loc, "GlobalOrderedCountIncrement() unsupported", "GlobalOrderedCountIncrement", "");
break;
}
default: default:
break; // most pass through unchanged break; // most pass through unchanged
} }
......
hlsl/hlslParseables.cpp
...@@ -250,6 +250,8 @@ glslang::TString& AppendTypeName(glslang::TString& s, const char* argOrder, cons ...@@ -250,6 +250,8 @@ glslang::TString& AppendTypeName(glslang::TString& s, const char* argOrder, cons
case 'D': s += "double"; break; case 'D': s += "double"; break;
case 'I': s += "int"; break; case 'I': s += "int"; break;
case 'U': s += "uint"; break; case 'U': s += "uint"; break;
case 'L': s += "int64_t"; break;
case 'M': s += "uint64_t"; break;
case 'B': s += "bool"; break; case 'B': s += "bool"; break;
case 'S': s += "sampler"; break; case 'S': s += "sampler"; break;
case 's': s += "SamplerComparisonState"; break; case 's': s += "SamplerComparisonState"; break;
...@@ -524,7 +526,7 @@ void TBuiltInParseablesHlsl::initialize(int /*version*/, EProfile /*profile*/, c ...@@ -524,7 +526,7 @@ void TBuiltInParseablesHlsl::initialize(int /*version*/, EProfile /*profile*/, c
// orderKey can be: // orderKey can be:
// S = scalar, V = vector, M = matrix, - = void // S = scalar, V = vector, M = matrix, - = void
// typekey can be: // typekey can be:
// D = double, F = float, U = uint, I = int, B = bool, S = sampler, s = shadowSampler // D = double, F = float, U = uint, I = int, B = bool, S = sampler, s = shadowSampler, M = uint64_t, L = int64_t
// An empty order or type key repeats the first one. E.g: SVM,, means 3 args each of SVM. // An empty order or type key repeats the first one. E.g: SVM,, means 3 args each of SVM.
// '>' as first letter of order creates an output parameter // '>' as first letter of order creates an output parameter
// '<' as first letter of order creates an input parameter // '<' as first letter of order creates an input parameter
...@@ -875,6 +877,32 @@ void TBuiltInParseablesHlsl::initialize(int /*version*/, EProfile /*profile*/, c ...@@ -875,6 +877,32 @@ void TBuiltInParseablesHlsl::initialize(int /*version*/, EProfile /*profile*/, c
{ "DecrementCounter", nullptr, nullptr, "-", "-", EShLangAll, true }, { "DecrementCounter", nullptr, nullptr, "-", "-", EShLangAll, true },
{ "Consume", nullptr, nullptr, "-", "-", EShLangAll, true }, { "Consume", nullptr, nullptr, "-", "-", EShLangAll, true },
// SM 6.0
{ "WaveOnce", "S", "B", "-", "-", EShLangPSCS, false},
{ "WaveGetLaneCount", "S", "U", "-", "-", EShLangPSCS, false},
{ "WaveGetLaneIndex", "S", "U", "-", "-", EShLangPSCS, false},
{ "WaveIsHelperLane", "S", "B", "-", "-", EShLangPS, false},
{ "WaveAnyTrue", "S", "B", "S", "B", EShLangPSCS, false},
{ "WaveAllTrue", "S", "B", "S", "B", EShLangPSCS, false},
{ "WaveAllEqual", "S", "B", "S", "B", EShLangPSCS, false},
{ "WaveBallot", "S", "M", "S", "B", EShLangPSCS, false},
{ "WaveReadLaneAt", nullptr, nullptr, "SV,S", "DFUI,U", EShLangPSCS, false},
{ "WaveReadFirstLane", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllSum", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllProduct", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllMin", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllMax", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllBitAnd", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllBitOr", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveAllBitXor", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WavePrefixSum", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WavePrefixProduct", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "QuadReadLaneAt", nullptr, nullptr, "SV,S", "DFUI,U", EShLangPSCS, false},
{ "QuadSwapX", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "QuadSwapY", nullptr, nullptr, "SV", "DFUI", EShLangPSCS, false},
{ "WaveGetOrderedIndex", "S", "U", "-", "-", EShLangPSCS, false},
{ "GlobalOrderedCountIncrement", "S", "U", "S", "U", EShLangPSCS, false},
// Mark end of list, since we want to avoid a range-based for, as some compilers don't handle it yet. // Mark end of list, since we want to avoid a range-based for, as some compilers don't handle it yet.
{ nullptr, nullptr, nullptr, nullptr, nullptr, 0, false }, { nullptr, nullptr, nullptr, nullptr, nullptr, 0, false },
}; };
...@@ -1211,6 +1239,32 @@ void TBuiltInParseablesHlsl::identifyBuiltIns(int /*version*/, EProfile /*profil ...@@ -1211,6 +1239,32 @@ void TBuiltInParseablesHlsl::identifyBuiltIns(int /*version*/, EProfile /*profil
// GS methods // GS methods
symbolTable.relateToOperator(BUILTIN_PREFIX "Append", EOpMethodAppend); symbolTable.relateToOperator(BUILTIN_PREFIX "Append", EOpMethodAppend);
symbolTable.relateToOperator(BUILTIN_PREFIX "RestartStrip", EOpMethodRestartStrip); symbolTable.relateToOperator(BUILTIN_PREFIX "RestartStrip", EOpMethodRestartStrip);
// Wave ops
symbolTable.relateToOperator("WaveOnce", EOpSubgroupElect);
symbolTable.relateToOperator("WaveGetLaneCount", EOpWaveGetLaneCount);
symbolTable.relateToOperator("WaveGetLaneIndex", EOpWaveGetLaneIndex);
symbolTable.relateToOperator("WaveIsHelperLane", EOpWaveIsHelperLane);
symbolTable.relateToOperator("WaveAnyTrue", EOpSubgroupAny);
symbolTable.relateToOperator("WaveAllTrue", EOpSubgroupAll);
symbolTable.relateToOperator("WaveAllEqual", EOpSubgroupAllEqual);
symbolTable.relateToOperator("WaveBallot", EOpWaveBallot);
symbolTable.relateToOperator("WaveReadLaneAt", EOpSubgroupShuffle);
symbolTable.relateToOperator("WaveReadFirstLane", EOpSubgroupBroadcastFirst);
symbolTable.relateToOperator("WaveAllSum", EOpSubgroupAdd);
symbolTable.relateToOperator("WaveAllProduct", EOpSubgroupMul);
symbolTable.relateToOperator("WaveAllMin", EOpSubgroupMin);
symbolTable.relateToOperator("WaveAllMax", EOpSubgroupMax);
symbolTable.relateToOperator("WaveAllBitAnd", EOpSubgroupAnd);
symbolTable.relateToOperator("WaveAllBitOr", EOpSubgroupOr);
symbolTable.relateToOperator("WaveAllBitXor", EOpSubgroupXor);
symbolTable.relateToOperator("WavePrefixSum", EOpSubgroupInclusiveAdd);
symbolTable.relateToOperator("WavePrefixProduct", EOpSubgroupInclusiveMul);
symbolTable.relateToOperator("QuadReadLaneAt", EOpSubgroupQuadBroadcast);
symbolTable.relateToOperator("QuadSwapX", EOpSubgroupQuadSwapHorizontal);
symbolTable.relateToOperator("QuadSwapY", EOpSubgroupQuadSwapVertical);
symbolTable.relateToOperator("WaveGetOrderedIndex", EOpWaveGetOrderedIndex);
symbolTable.relateToOperator("GlobalOrderedCountIncrement", EOpGlobalOrderedCountIncrement);
} }
// //
......
hlsl/hlslScanContext.cpp
...@@ -143,6 +143,7 @@ void HlslScanContext::fillInKeywordMap() ...@@ -143,6 +143,7 @@ void HlslScanContext::fillInKeywordMap()
(*KeywordMap)["bool"] = EHTokBool; (*KeywordMap)["bool"] = EHTokBool;
(*KeywordMap)["int"] = EHTokInt; (*KeywordMap)["int"] = EHTokInt;
(*KeywordMap)["uint"] = EHTokUint; (*KeywordMap)["uint"] = EHTokUint;
(*KeywordMap)["uint64_t"] = EHTokUint64;
(*KeywordMap)["dword"] = EHTokDword; (*KeywordMap)["dword"] = EHTokDword;
(*KeywordMap)["half"] = EHTokHalf; (*KeywordMap)["half"] = EHTokHalf;
(*KeywordMap)["float"] = EHTokFloat; (*KeywordMap)["float"] = EHTokFloat;
...@@ -651,6 +652,7 @@ EHlslTokenClass HlslScanContext::tokenizeIdentifier() ...@@ -651,6 +652,7 @@ EHlslTokenClass HlslScanContext::tokenizeIdentifier()
case EHTokBool: case EHTokBool:
case EHTokInt: case EHTokInt:
case EHTokUint: case EHTokUint:
case EHTokUint64:
case EHTokDword: case EHTokDword:
case EHTokHalf: case EHTokHalf:
case EHTokFloat: case EHTokFloat:
......
hlsl/hlslTokens.h
...@@ -95,6 +95,7 @@ enum EHlslTokenClass { ...@@ -95,6 +95,7 @@ enum EHlslTokenClass {
EHTokBool, EHTokBool,
EHTokInt, EHTokInt,
EHTokUint, EHTokUint,
EHTokUint64,
EHTokDword, EHTokDword,
EHTokHalf, EHTokHalf,
EHTokFloat, EHTokFloat,
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment