Skip to content

G
glslang
Commit 83170e3f by Daniel Koch
Options

Enable GL_KHR_shader_subgroup properly

(and GL_NV_shader_subgroup_partitioned) based on GL/ES version instead of predicating it on vulkan SPV generation Also add AST testing. The glsl.450.subgroup* files are largely the same as the spv.subgroup* The glsl.es320.subgroup* files are the same as the 450 versions, but modified to be ES compatible.
parent 5efb004d
Showing with 2116 additions and 10 deletions
Test/baseResults/glsl.450.subgroup.frag.out 0 → 100644
glsl.450.subgroup.frag
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:4 Function Definition: main( ( global void)
0:4 Function Parameters:
0:6 Sequence
0:6 move second child to first child ( temp 4-component vector of uint)
0:6 'data' (layout( location=0) out 4-component vector of uint)
0:6 Construct uvec4 ( temp 4-component vector of uint)
0:6 'gl_SubgroupSize' ( flat in uint unknown built-in variable)
0:6 'gl_SubgroupInvocationID' ( flat in uint unknown built-in variable)
0:6 Constant:
0:6 0 (const uint)
0:6 Constant:
0:6 0 (const uint)
0:? Linker Objects
0:? 'data' (layout( location=0) out 4-component vector of uint)
Linked fragment stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:4 Function Definition: main( ( global void)
0:4 Function Parameters:
0:6 Sequence
0:6 move second child to first child ( temp 4-component vector of uint)
0:6 'data' (layout( location=0) out 4-component vector of uint)
0:6 Construct uvec4 ( temp 4-component vector of uint)
0:6 'gl_SubgroupSize' ( flat in uint unknown built-in variable)
0:6 'gl_SubgroupInvocationID' ( flat in uint unknown built-in variable)
0:6 Constant:
0:6 0 (const uint)
0:6 Constant:
0:6 0 (const uint)
0:? Linker Objects
0:? 'data' (layout( location=0) out 4-component vector of uint)
Test/baseResults/glsl.450.subgroup.geom.out 0 → 100644
glsl.450.subgroup.geom
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
invocations = -1
max_vertices = 1
input primitive = points
output primitive = points
0:? Sequence
0:10 Function Definition: main( ( global void)
0:10 Function Parameters:
0:12 Sequence
0:12 move second child to first child ( temp 4-component vector of uint)
0:12 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:12 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:12 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:12 Constant:
0:12 0 (const uint)
0:12 'gl_PrimitiveIDIn' ( in int PrimitiveID)
0:12 Construct uvec4 ( temp 4-component vector of uint)
0:12 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:12 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:12 Constant:
0:12 0 (const uint)
0:12 Constant:
0:12 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Linked geometry stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
invocations = 1
max_vertices = 1
input primitive = points
output primitive = points
0:? Sequence
0:10 Function Definition: main( ( global void)
0:10 Function Parameters:
0:12 Sequence
0:12 move second child to first child ( temp 4-component vector of uint)
0:12 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:12 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:12 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:12 Constant:
0:12 0 (const uint)
0:12 'gl_PrimitiveIDIn' ( in int PrimitiveID)
0:12 Construct uvec4 ( temp 4-component vector of uint)
0:12 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:12 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:12 Constant:
0:12 0 (const uint)
0:12 Constant:
0:12 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Test/baseResults/glsl.450.subgroup.tesc.out 0 → 100644
glsl.450.subgroup.tesc
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
vertices = 1
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in int PrimitiveID)
0:11 Construct uvec4 ( temp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Linked tessellation control stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
vertices = 1
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in int PrimitiveID)
0:11 Construct uvec4 ( temp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Test/baseResults/glsl.450.subgroup.tese.out 0 → 100644
glsl.450.subgroup.tese
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
input primitive = isolines
vertex spacing = none
triangle order = none
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in int PrimitiveID)
0:11 Construct uvec4 ( temp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Linked tessellation evaluation stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
input primitive = isolines
vertex spacing = equal_spacing
triangle order = ccw
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in int PrimitiveID)
0:11 Construct uvec4 ( temp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
Test/baseResults/glsl.450.subgroup.vert.out 0 → 100644
glsl.450.subgroup.vert
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:8 Function Definition: main( ( global void)
0:8 Function Parameters:
0:10 Sequence
0:10 move second child to first child ( temp 4-component vector of uint)
0:10 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:10 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:10 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:10 Constant:
0:10 0 (const uint)
0:10 'gl_VertexID' ( gl_VertexId int VertexId)
0:10 Construct uvec4 ( temp 4-component vector of uint)
0:10 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:10 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:10 Constant:
0:10 0 (const uint)
0:10 Constant:
0:10 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:? 'gl_VertexID' ( gl_VertexId int VertexId)
0:? 'gl_InstanceID' ( gl_InstanceId int InstanceId)
Linked vertex stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:8 Function Definition: main( ( global void)
0:8 Function Parameters:
0:10 Sequence
0:10 move second child to first child ( temp 4-component vector of uint)
0:10 indirect index (layout( column_major std430 offset=0) temp 4-component vector of uint)
0:10 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint)
0:10 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:10 Constant:
0:10 0 (const uint)
0:10 'gl_VertexID' ( gl_VertexId int VertexId)
0:10 Construct uvec4 ( temp 4-component vector of uint)
0:10 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:10 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:10 Constant:
0:10 0 (const uint)
0:10 Constant:
0:10 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of 4-component vector of uint result})
0:? 'gl_VertexID' ( gl_VertexId int VertexId)
0:? 'gl_InstanceID' ( gl_InstanceId int InstanceId)
Test/baseResults/glsl.450.subgroupArithmetic.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupBallot.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupBasic.comp.out 0 → 100644
glsl.450.subgroupBasic.comp
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
local_size = (8, 8, 1)
0:? Sequence
0:12 Function Definition: main( ( global void)
0:12 Function Parameters:
0:14 Sequence
0:14 move second child to first child ( temp int)
0:14 indirect index (layout( column_major shared) temp int)
0:14 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:14 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:14 Constant:
0:14 0 (const int)
0:14 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:14 Constant:
0:14 1 (const int)
0:15 move second child to first child ( temp int)
0:15 indirect index (layout( column_major shared) temp int)
0:15 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:15 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:15 Constant:
0:15 0 (const int)
0:15 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:15 Constant:
0:15 1 (const int)
0:16 move second child to first child ( temp int)
0:16 indirect index (layout( column_major shared) temp int)
0:16 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:16 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:16 Constant:
0:16 0 (const int)
0:16 'gl_NumSubgroups' ( in uint unknown built-in variable)
0:16 Constant:
0:16 1 (const int)
0:17 move second child to first child ( temp int)
0:17 indirect index (layout( column_major shared) temp int)
0:17 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:17 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:17 Constant:
0:17 0 (const int)
0:17 'gl_SubgroupID' ( in uint unknown built-in variable)
0:17 Test condition and select ( temp int)
0:17 Condition
0:17 subgroupElect ( global bool)
0:17 true case
0:17 Constant:
0:17 1 (const int)
0:17 false case
0:17 Constant:
0:17 0 (const int)
0:18 subgroupBarrier ( global void)
0:19 subgroupMemoryBarrier ( global void)
0:20 subgroupMemoryBarrierBuffer ( global void)
0:21 subgroupMemoryBarrierShared ( global void)
0:22 subgroupMemoryBarrierImage ( global void)
0:? Linker Objects
0:? 'gl_WorkGroupSize' ( const 3-component vector of uint WorkGroupSize)
0:? 8 (const uint)
0:? 8 (const uint)
0:? 1 (const uint)
0:? 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
Linked compute stage:
Shader version: 450
Requested GL_KHR_shader_subgroup_basic
local_size = (8, 8, 1)
0:? Sequence
0:12 Function Definition: main( ( global void)
0:12 Function Parameters:
0:14 Sequence
0:14 move second child to first child ( temp int)
0:14 indirect index (layout( column_major shared) temp int)
0:14 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:14 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:14 Constant:
0:14 0 (const int)
0:14 'gl_SubgroupSize' ( in uint unknown built-in variable)
0:14 Constant:
0:14 1 (const int)
0:15 move second child to first child ( temp int)
0:15 indirect index (layout( column_major shared) temp int)
0:15 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:15 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:15 Constant:
0:15 0 (const int)
0:15 'gl_SubgroupInvocationID' ( in uint unknown built-in variable)
0:15 Constant:
0:15 1 (const int)
0:16 move second child to first child ( temp int)
0:16 indirect index (layout( column_major shared) temp int)
0:16 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:16 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:16 Constant:
0:16 0 (const int)
0:16 'gl_NumSubgroups' ( in uint unknown built-in variable)
0:16 Constant:
0:16 1 (const int)
0:17 move second child to first child ( temp int)
0:17 indirect index (layout( column_major shared) temp int)
0:17 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of int)
0:17 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
0:17 Constant:
0:17 0 (const int)
0:17 'gl_SubgroupID' ( in uint unknown built-in variable)
0:17 Test condition and select ( temp int)
0:17 Condition
0:17 subgroupElect ( global bool)
0:17 true case
0:17 Constant:
0:17 1 (const int)
0:17 false case
0:17 Constant:
0:17 0 (const int)
0:18 subgroupBarrier ( global void)
0:19 subgroupMemoryBarrier ( global void)
0:20 subgroupMemoryBarrierBuffer ( global void)
0:21 subgroupMemoryBarrierShared ( global void)
0:22 subgroupMemoryBarrierImage ( global void)
0:? Linker Objects
0:? 'gl_WorkGroupSize' ( const 3-component vector of uint WorkGroupSize)
0:? 8 (const uint)
0:? 8 (const uint)
0:? 1 (const uint)
0:? 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of int a})
Test/baseResults/glsl.450.subgroupClustered.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupPartitioned.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupQuad.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupShuffle.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupShuffleRelative.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.450.subgroupVote.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroup.frag.out 0 → 100644
glsl.es320.subgroup.frag
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:4 Function Definition: main( ( global void)
0:4 Function Parameters:
0:6 Sequence
0:6 move second child to first child ( temp mediump 4-component vector of uint)
0:6 'data' (layout( location=0) out mediump 4-component vector of uint)
0:6 Construct uvec4 ( temp mediump 4-component vector of uint)
0:6 'gl_SubgroupSize' ( flat in mediump uint unknown built-in variable)
0:6 'gl_SubgroupInvocationID' ( flat in mediump uint unknown built-in variable)
0:6 Constant:
0:6 0 (const uint)
0:6 Constant:
0:6 0 (const uint)
0:? Linker Objects
0:? 'data' (layout( location=0) out mediump 4-component vector of uint)
Linked fragment stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:4 Function Definition: main( ( global void)
0:4 Function Parameters:
0:6 Sequence
0:6 move second child to first child ( temp mediump 4-component vector of uint)
0:6 'data' (layout( location=0) out mediump 4-component vector of uint)
0:6 Construct uvec4 ( temp mediump 4-component vector of uint)
0:6 'gl_SubgroupSize' ( flat in mediump uint unknown built-in variable)
0:6 'gl_SubgroupInvocationID' ( flat in mediump uint unknown built-in variable)
0:6 Constant:
0:6 0 (const uint)
0:6 Constant:
0:6 0 (const uint)
0:? Linker Objects
0:? 'data' (layout( location=0) out mediump 4-component vector of uint)
Test/baseResults/glsl.es320.subgroup.geom.out 0 → 100644
glsl.es320.subgroup.geom
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
invocations = -1
max_vertices = 1
input primitive = points
output primitive = points
0:? Sequence
0:10 Function Definition: main( ( global void)
0:10 Function Parameters:
0:12 Sequence
0:12 move second child to first child ( temp highp 4-component vector of uint)
0:12 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:12 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:12 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:12 Constant:
0:12 0 (const uint)
0:12 'gl_PrimitiveIDIn' ( in highp int PrimitiveID)
0:12 Construct uvec4 ( temp highp 4-component vector of uint)
0:12 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:12 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:12 Constant:
0:12 0 (const uint)
0:12 Constant:
0:12 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Linked geometry stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
invocations = 1
max_vertices = 1
input primitive = points
output primitive = points
0:? Sequence
0:10 Function Definition: main( ( global void)
0:10 Function Parameters:
0:12 Sequence
0:12 move second child to first child ( temp highp 4-component vector of uint)
0:12 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:12 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:12 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:12 Constant:
0:12 0 (const uint)
0:12 'gl_PrimitiveIDIn' ( in highp int PrimitiveID)
0:12 Construct uvec4 ( temp highp 4-component vector of uint)
0:12 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:12 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:12 Constant:
0:12 0 (const uint)
0:12 Constant:
0:12 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Test/baseResults/glsl.es320.subgroup.tesc.out 0 → 100644
glsl.es320.subgroup.tesc
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
vertices = 1
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp highp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in highp int PrimitiveID)
0:11 Construct uvec4 ( temp highp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Linked tessellation control stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
vertices = 1
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp highp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in highp int PrimitiveID)
0:11 Construct uvec4 ( temp highp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Test/baseResults/glsl.es320.subgroup.tese.out 0 → 100644
glsl.es320.subgroup.tese
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
input primitive = isolines
vertex spacing = none
triangle order = none
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp highp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in highp int PrimitiveID)
0:11 Construct uvec4 ( temp highp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Linked tessellation evaluation stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
input primitive = isolines
vertex spacing = equal_spacing
triangle order = ccw
0:? Sequence
0:9 Function Definition: main( ( global void)
0:9 Function Parameters:
0:11 Sequence
0:11 move second child to first child ( temp highp 4-component vector of uint)
0:11 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:11 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:11 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:11 Constant:
0:11 0 (const uint)
0:11 'gl_PrimitiveID' ( in highp int PrimitiveID)
0:11 Construct uvec4 ( temp highp 4-component vector of uint)
0:11 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:11 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:11 Constant:
0:11 0 (const uint)
0:11 Constant:
0:11 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
Test/baseResults/glsl.es320.subgroup.vert.out 0 → 100644
glsl.es320.subgroup.vert
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:8 Function Definition: main( ( global void)
0:8 Function Parameters:
0:10 Sequence
0:10 move second child to first child ( temp highp 4-component vector of uint)
0:10 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:10 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:10 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:10 Constant:
0:10 0 (const uint)
0:10 'gl_VertexID' ( gl_VertexId highp int VertexId)
0:10 Construct uvec4 ( temp highp 4-component vector of uint)
0:10 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:10 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:10 Constant:
0:10 0 (const uint)
0:10 Constant:
0:10 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:? 'gl_VertexID' ( gl_VertexId highp int VertexId)
0:? 'gl_InstanceID' ( gl_InstanceId highp int InstanceId)
Linked vertex stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
0:? Sequence
0:8 Function Definition: main( ( global void)
0:8 Function Parameters:
0:10 Sequence
0:10 move second child to first child ( temp highp 4-component vector of uint)
0:10 indirect index (layout( column_major std430 offset=0) temp highp 4-component vector of uint)
0:10 result: direct index for structure (layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint)
0:10 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:10 Constant:
0:10 0 (const uint)
0:10 'gl_VertexID' ( gl_VertexId highp int VertexId)
0:10 Construct uvec4 ( temp highp 4-component vector of uint)
0:10 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:10 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:10 Constant:
0:10 0 (const uint)
0:10 Constant:
0:10 0 (const uint)
0:? Linker Objects
0:? 'anon@0' (layout( set=0 binding=0 column_major std430) buffer block{layout( column_major std430 offset=0) buffer runtime-sized array of highp 4-component vector of uint result})
0:? 'gl_VertexID' ( gl_VertexId highp int VertexId)
0:? 'gl_InstanceID' ( gl_InstanceId highp int InstanceId)
Test/baseResults/glsl.es320.subgroupArithmetic.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupBallot.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupBasic.comp.out 0 → 100644
glsl.es320.subgroupBasic.comp
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
local_size = (8, 8, 1)
0:? Sequence
0:12 Function Definition: main( ( global void)
0:12 Function Parameters:
0:14 Sequence
0:14 move second child to first child ( temp highp int)
0:14 indirect index (layout( column_major shared) temp highp int)
0:14 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:14 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:14 Constant:
0:14 0 (const int)
0:14 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:14 Constant:
0:14 1 (const int)
0:15 move second child to first child ( temp highp int)
0:15 indirect index (layout( column_major shared) temp highp int)
0:15 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:15 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:15 Constant:
0:15 0 (const int)
0:15 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:15 Constant:
0:15 1 (const int)
0:16 move second child to first child ( temp highp int)
0:16 indirect index (layout( column_major shared) temp highp int)
0:16 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:16 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:16 Constant:
0:16 0 (const int)
0:16 'gl_NumSubgroups' ( in highp uint unknown built-in variable)
0:16 Constant:
0:16 1 (const int)
0:17 move second child to first child ( temp highp int)
0:17 indirect index (layout( column_major shared) temp highp int)
0:17 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:17 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:17 Constant:
0:17 0 (const int)
0:17 'gl_SubgroupID' ( in highp uint unknown built-in variable)
0:17 Test condition and select ( temp highp int)
0:17 Condition
0:17 subgroupElect ( global bool)
0:17 true case
0:17 Constant:
0:17 1 (const int)
0:17 false case
0:17 Constant:
0:17 0 (const int)
0:18 subgroupBarrier ( global void)
0:19 subgroupMemoryBarrier ( global void)
0:20 subgroupMemoryBarrierBuffer ( global void)
0:21 subgroupMemoryBarrierShared ( global void)
0:22 subgroupMemoryBarrierImage ( global void)
0:? Linker Objects
0:? 'gl_WorkGroupSize' ( const highp 3-component vector of uint WorkGroupSize)
0:? 8 (const uint)
0:? 8 (const uint)
0:? 1 (const uint)
0:? 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
Linked compute stage:
Shader version: 320
Requested GL_KHR_shader_subgroup_basic
local_size = (8, 8, 1)
0:? Sequence
0:12 Function Definition: main( ( global void)
0:12 Function Parameters:
0:14 Sequence
0:14 move second child to first child ( temp highp int)
0:14 indirect index (layout( column_major shared) temp highp int)
0:14 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:14 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:14 Constant:
0:14 0 (const int)
0:14 'gl_SubgroupSize' ( in mediump uint unknown built-in variable)
0:14 Constant:
0:14 1 (const int)
0:15 move second child to first child ( temp highp int)
0:15 indirect index (layout( column_major shared) temp highp int)
0:15 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:15 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:15 Constant:
0:15 0 (const int)
0:15 'gl_SubgroupInvocationID' ( in mediump uint unknown built-in variable)
0:15 Constant:
0:15 1 (const int)
0:16 move second child to first child ( temp highp int)
0:16 indirect index (layout( column_major shared) temp highp int)
0:16 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:16 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:16 Constant:
0:16 0 (const int)
0:16 'gl_NumSubgroups' ( in highp uint unknown built-in variable)
0:16 Constant:
0:16 1 (const int)
0:17 move second child to first child ( temp highp int)
0:17 indirect index (layout( column_major shared) temp highp int)
0:17 a: direct index for structure (layout( column_major shared) buffer runtime-sized array of highp int)
0:17 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
0:17 Constant:
0:17 0 (const int)
0:17 'gl_SubgroupID' ( in highp uint unknown built-in variable)
0:17 Test condition and select ( temp highp int)
0:17 Condition
0:17 subgroupElect ( global bool)
0:17 true case
0:17 Constant:
0:17 1 (const int)
0:17 false case
0:17 Constant:
0:17 0 (const int)
0:18 subgroupBarrier ( global void)
0:19 subgroupMemoryBarrier ( global void)
0:20 subgroupMemoryBarrierBuffer ( global void)
0:21 subgroupMemoryBarrierShared ( global void)
0:22 subgroupMemoryBarrierImage ( global void)
0:? Linker Objects
0:? 'gl_WorkGroupSize' ( const highp 3-component vector of uint WorkGroupSize)
0:? 8 (const uint)
0:? 8 (const uint)
0:? 1 (const uint)
0:? 'data' (layout( binding=0 column_major shared) buffer block{layout( column_major shared) buffer runtime-sized array of highp int a})
Test/baseResults/glsl.es320.subgroupClustered.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupPartitioned.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupQuad.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupShuffle.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/baseResults/glsl.es320.subgroupShuffleRelative.comp.out 0 → 100644
This source diff could not be displayed because it is too large. You can view the blob instead.
Test/glsl.450.subgroup.frag 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout(location = 0) out uvec4 data;
void main (void)
{
data = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.450.subgroup.geom 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout(points) in;
layout(points, max_vertices = 1) out;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveIDIn] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.450.subgroup.tesc 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout(vertices=1) out;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.450.subgroup.tese 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout(isolines) in;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.450.subgroup.vert 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_VertexID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.450.subgroupBallot.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_ballot: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
uvec4 relMask = gl_SubgroupEqMask +
gl_SubgroupGeMask +
gl_SubgroupGtMask +
gl_SubgroupLeMask +
gl_SubgroupLtMask;
uvec4 result = subgroupBallot(true);
data[invocation].u4.x = subgroupBallotBitCount(result);
data[invocation].u4.y = subgroupBallotBitExtract(result, 0) ? 1 : 0;
data[invocation].u4.z = subgroupBallotInclusiveBitCount(result) + subgroupBallotExclusiveBitCount(result);
data[invocation].u4.w = subgroupBallotFindLSB(result) + subgroupBallotFindMSB(result);
if ((relMask == result) && subgroupInverseBallot(data[0].u4))
{
data[invocation].f4.x = subgroupBroadcast(data[0].f4.x, 3);
data[invocation].f4.xy = subgroupBroadcast(data[1].f4.xy, 3);
data[invocation].f4.xyz = subgroupBroadcast(data[2].f4.xyz, 3);
data[invocation].f4 = subgroupBroadcast(data[3].f4, 3);
data[invocation].i4.x = subgroupBroadcast(data[0].i4.x, 2);
data[invocation].i4.xy = subgroupBroadcast(data[1].i4.xy, 2);
data[invocation].i4.xyz = subgroupBroadcast(data[2].i4.xyz, 2);
data[invocation].i4 = subgroupBroadcast(data[3].i4, 2);
data[invocation].u4.x = subgroupBroadcast(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupBroadcast(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupBroadcast(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupBroadcast(data[3].u4, 1);
data[invocation].d4.x = subgroupBroadcast(data[0].d4.x, 0);
data[invocation].d4.xy = subgroupBroadcast(data[1].d4.xy, 0);
data[invocation].d4.xyz = subgroupBroadcast(data[2].d4.xyz, 0);
data[invocation].d4 = subgroupBroadcast(data[3].d4, 0);
data[invocation].i4.x = int(subgroupBroadcast(data[0].i4.x < 0, 1));
data[invocation].i4.xy = ivec2(subgroupBroadcast(lessThan(data[1].i4.xy, ivec2(0)), 1));
data[invocation].i4.xyz = ivec3(subgroupBroadcast(lessThan(data[1].i4.xyz, ivec3(0)), 1));
data[invocation].i4 = ivec4(subgroupBroadcast(lessThan(data[1].i4, ivec4(0)), 1));
}
else
{
data[invocation].f4.x = subgroupBroadcastFirst(data[0].f4.x);
data[invocation].f4.xy = subgroupBroadcastFirst(data[1].f4.xy);
data[invocation].f4.xyz = subgroupBroadcastFirst(data[2].f4.xyz);
data[invocation].f4 = subgroupBroadcastFirst(data[3].f4);
data[invocation].i4.x = subgroupBroadcastFirst(data[0].i4.x);
data[invocation].i4.xy = subgroupBroadcastFirst(data[1].i4.xy);
data[invocation].i4.xyz = subgroupBroadcastFirst(data[2].i4.xyz);
data[invocation].i4 = subgroupBroadcastFirst(data[3].i4);
data[invocation].u4.x = subgroupBroadcastFirst(data[0].u4.x);
data[invocation].u4.xy = subgroupBroadcastFirst(data[1].u4.xy);
data[invocation].u4.xyz = subgroupBroadcastFirst(data[2].u4.xyz);
data[invocation].u4 = subgroupBroadcastFirst(data[3].u4);
data[invocation].d4.x = subgroupBroadcastFirst(data[0].d4.x);
data[invocation].d4.xy = subgroupBroadcastFirst(data[1].d4.xy);
data[invocation].d4.xyz = subgroupBroadcastFirst(data[2].d4.xyz);
data[invocation].d4 = subgroupBroadcastFirst(data[3].d4);
data[invocation].i4.x = int(subgroupBroadcastFirst(data[0].i4.x < 0));
data[invocation].i4.xy = ivec2(subgroupBroadcastFirst(lessThan(data[1].i4.xy, ivec2(0))));
data[invocation].i4.xyz = ivec3(subgroupBroadcastFirst(lessThan(data[1].i4.xyz, ivec3(0))));
data[invocation].i4 = ivec4(subgroupBroadcastFirst(lessThan(data[1].i4, ivec4(0))));
}
}
Test/glsl.450.subgroupBallotNeg.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_ballot: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
uvec4 relMask = gl_SubgroupEqMask +
gl_SubgroupGeMask +
gl_SubgroupGtMask +
gl_SubgroupLeMask +
gl_SubgroupLtMask;
uvec4 result = subgroupBallot(true);
data[invocation].u4.x = subgroupBallotBitCount(result);
data[invocation].u4.y = subgroupBallotBitExtract(result, 0) ? 1 : 0;
data[invocation].u4.z = subgroupBallotInclusiveBitCount(result) + subgroupBallotExclusiveBitCount(result);
data[invocation].u4.w = subgroupBallotFindLSB(result) + subgroupBallotFindMSB(result);
data[invocation].f4.x = subgroupBroadcast(data[0].f4.x, invocation); // ERROR: not constant
}
Test/glsl.450.subgroupBasic.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_basic: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffer
{
int a[];
} data;
void main()
{
data.a[gl_SubgroupSize] = 1;
data.a[gl_SubgroupInvocationID] = 1;
data.a[gl_NumSubgroups] = 1;
data.a[gl_SubgroupID] = (subgroupElect()) ? 1 : 0;
subgroupBarrier();
subgroupMemoryBarrier();
subgroupMemoryBarrierBuffer();
subgroupMemoryBarrierShared();
subgroupMemoryBarrierImage();
}
Test/glsl.450.subgroupClustered.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_clustered: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
data[invocation].f4.x = subgroupClusteredAdd(data[0].f4.x, 1);
data[invocation].f4.xy = subgroupClusteredAdd(data[1].f4.xy, 1);
data[invocation].f4.xyz = subgroupClusteredAdd(data[2].f4.xyz, 1);
data[invocation].f4 = subgroupClusteredAdd(data[3].f4, 1);
data[invocation].i4.x = subgroupClusteredAdd(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredAdd(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredAdd(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredAdd(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredAdd(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredAdd(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredAdd(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredAdd(data[3].u4, 1);
data[invocation].d4.x = subgroupClusteredAdd(data[0].d4.x, 1);
data[invocation].d4.xy = subgroupClusteredAdd(data[1].d4.xy, 1);
data[invocation].d4.xyz = subgroupClusteredAdd(data[2].d4.xyz, 1);
data[invocation].d4 = subgroupClusteredAdd(data[3].d4, 1);
data[invocation].f4.x = subgroupClusteredMul(data[0].f4.x, 1);
data[invocation].f4.xy = subgroupClusteredMul(data[1].f4.xy, 1);
data[invocation].f4.xyz = subgroupClusteredMul(data[2].f4.xyz, 1);
data[invocation].f4 = subgroupClusteredMul(data[3].f4, 1);
data[invocation].i4.x = subgroupClusteredMul(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredMul(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredMul(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredMul(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredMul(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredMul(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredMul(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredMul(data[3].u4, 1);
data[invocation].d4.x = subgroupClusteredMul(data[0].d4.x, 1);
data[invocation].d4.xy = subgroupClusteredMul(data[1].d4.xy, 1);
data[invocation].d4.xyz = subgroupClusteredMul(data[2].d4.xyz, 1);
data[invocation].d4 = subgroupClusteredMul(data[3].d4, 1);
data[invocation].f4.x = subgroupClusteredMin(data[0].f4.x, 1);
data[invocation].f4.xy = subgroupClusteredMin(data[1].f4.xy, 1);
data[invocation].f4.xyz = subgroupClusteredMin(data[2].f4.xyz, 1);
data[invocation].f4 = subgroupClusteredMin(data[3].f4, 1);
data[invocation].i4.x = subgroupClusteredMin(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredMin(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredMin(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredMin(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredMin(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredMin(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredMin(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredMin(data[3].u4, 1);
data[invocation].d4.x = subgroupClusteredMin(data[0].d4.x, 1);
data[invocation].d4.xy = subgroupClusteredMin(data[1].d4.xy, 1);
data[invocation].d4.xyz = subgroupClusteredMin(data[2].d4.xyz, 1);
data[invocation].d4 = subgroupClusteredMin(data[3].d4, 1);
data[invocation].f4.x = subgroupClusteredMax(data[0].f4.x, 1);
data[invocation].f4.xy = subgroupClusteredMax(data[1].f4.xy, 1);
data[invocation].f4.xyz = subgroupClusteredMax(data[2].f4.xyz, 1);
data[invocation].f4 = subgroupClusteredMax(data[3].f4, 1);
data[invocation].i4.x = subgroupClusteredMax(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredMax(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredMax(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredMax(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredMax(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredMax(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredMax(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredMax(data[3].u4, 1);
data[invocation].d4.x = subgroupClusteredMax(data[0].d4.x, 1);
data[invocation].d4.xy = subgroupClusteredMax(data[1].d4.xy, 1);
data[invocation].d4.xyz = subgroupClusteredMax(data[2].d4.xyz, 1);
data[invocation].d4 = subgroupClusteredMax(data[3].d4, 1);
data[invocation].i4.x = subgroupClusteredAnd(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredAnd(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredAnd(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredAnd(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredAnd(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredAnd(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredAnd(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredAnd(data[3].u4, 1);
data[invocation].i4.x = int(subgroupClusteredAnd(data[0].i4.x < 0, 1));
data[invocation].i4.xy = ivec2(subgroupClusteredAnd(lessThan(data[1].i4.xy, ivec2(0)), 1));
data[invocation].i4.xyz = ivec3(subgroupClusteredAnd(lessThan(data[1].i4.xyz, ivec3(0)), 1));
data[invocation].i4 = ivec4(subgroupClusteredAnd(lessThan(data[1].i4, ivec4(0)), 1));
data[invocation].i4.x = subgroupClusteredOr(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredOr(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredOr(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredOr(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredOr(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredOr(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredOr(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredOr(data[3].u4, 1);
data[invocation].i4.x = int(subgroupClusteredOr(data[0].i4.x < 0, 1));
data[invocation].i4.xy = ivec2(subgroupClusteredOr(lessThan(data[1].i4.xy, ivec2(0)), 1));
data[invocation].i4.xyz = ivec3(subgroupClusteredOr(lessThan(data[1].i4.xyz, ivec3(0)), 1));
data[invocation].i4 = ivec4(subgroupClusteredOr(lessThan(data[1].i4, ivec4(0)), 1));
data[invocation].i4.x = subgroupClusteredXor(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupClusteredXor(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupClusteredXor(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupClusteredXor(data[3].i4, 1);
data[invocation].u4.x = subgroupClusteredXor(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupClusteredXor(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupClusteredXor(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupClusteredXor(data[3].u4, 1);
data[invocation].i4.x = int(subgroupClusteredXor(data[0].i4.x < 0, 1));
data[invocation].i4.xy = ivec2(subgroupClusteredXor(lessThan(data[1].i4.xy, ivec2(0)), 1));
data[invocation].i4.xyz = ivec3(subgroupClusteredXor(lessThan(data[1].i4.xyz, ivec3(0)), 1));
data[invocation].i4 = ivec4(subgroupClusteredXor(lessThan(data[1].i4, ivec4(0)), 1));
}
Test/glsl.450.subgroupClusteredNeg.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_clustered: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
int a = 1;
const int aConst = 1;
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
data[invocation].f4.xy = subgroupClusteredAdd(data[1].f4.xy, 0); // ERROR, less than 1
data[invocation].f4.x = subgroupClusteredMul(data[0].f4.x, 3); // ERROR, not a power of 2
data[invocation].i4.xy = subgroupClusteredMin(data[1].i4.xy, 8);
data[invocation].i4.xyz = subgroupClusteredMin(data[2].i4.xyz, 6); // ERROR, not a power of 2
data[invocation].f4.x = subgroupClusteredMax(data[0].f4.x, -1); // ERROR, less than 1
data[invocation].i4 = subgroupClusteredAnd(data[3].i4, -3); // ERROR, less than 1
data[invocation].i4.x = subgroupClusteredOr(data[0].i4.x, a); // ERROR, not constant
data[invocation].i4.xy = subgroupClusteredOr(data[1].i4.xy, aConst);
data[invocation].i4.x = subgroupClusteredXor(data[0].i4.x, 1 + a); // ERROR, not constant
data[invocation].i4.xy = subgroupClusteredXor(data[1].i4.xy, aConst + a); // ERROR, not constant
data[invocation].i4.xyz = subgroupClusteredXor(data[2].i4.xyz, 1 + aConst);
}
Test/glsl.450.subgroupQuad.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_quad: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
data[invocation].f4.x = subgroupQuadBroadcast(data[0].f4.x, 1);
data[invocation].f4.xy = subgroupQuadBroadcast(data[1].f4.xy, 1);
data[invocation].f4.xyz = subgroupQuadBroadcast(data[2].f4.xyz, 1);
data[invocation].f4 = subgroupQuadBroadcast(data[3].f4, 1);
data[invocation].i4.x = subgroupQuadBroadcast(data[0].i4.x, 1);
data[invocation].i4.xy = subgroupQuadBroadcast(data[1].i4.xy, 1);
data[invocation].i4.xyz = subgroupQuadBroadcast(data[2].i4.xyz, 1);
data[invocation].i4 = subgroupQuadBroadcast(data[3].i4, 1);
data[invocation].u4.x = subgroupQuadBroadcast(data[0].u4.x, 1);
data[invocation].u4.xy = subgroupQuadBroadcast(data[1].u4.xy, 1);
data[invocation].u4.xyz = subgroupQuadBroadcast(data[2].u4.xyz, 1);
data[invocation].u4 = subgroupQuadBroadcast(data[3].u4, 1);
data[invocation].d4.x = subgroupQuadBroadcast(data[0].d4.x, 1);
data[invocation].d4.xy = subgroupQuadBroadcast(data[1].d4.xy, 1);
data[invocation].d4.xyz = subgroupQuadBroadcast(data[2].d4.xyz, 1);
data[invocation].d4 = subgroupQuadBroadcast(data[3].d4, 1);
data[invocation].i4.x = int(subgroupQuadBroadcast(data[0].i4.x < 0, 1));
data[invocation].i4.xy = ivec2(subgroupQuadBroadcast(lessThan(data[1].i4.xy, ivec2(0)), 1));
data[invocation].i4.xyz = ivec3(subgroupQuadBroadcast(lessThan(data[1].i4.xyz, ivec3(0)), 1));
data[invocation].i4 = ivec4(subgroupQuadBroadcast(lessThan(data[1].i4, ivec4(0)), 1));
data[invocation].f4.x = subgroupQuadSwapHorizontal(data[0].f4.x);
data[invocation].f4.xy = subgroupQuadSwapHorizontal(data[1].f4.xy);
data[invocation].f4.xyz = subgroupQuadSwapHorizontal(data[2].f4.xyz);
data[invocation].f4 = subgroupQuadSwapHorizontal(data[3].f4);
data[invocation].i4.x = subgroupQuadSwapHorizontal(data[0].i4.x);
data[invocation].i4.xy = subgroupQuadSwapHorizontal(data[1].i4.xy);
data[invocation].i4.xyz = subgroupQuadSwapHorizontal(data[2].i4.xyz);
data[invocation].i4 = subgroupQuadSwapHorizontal(data[3].i4);
data[invocation].u4.x = subgroupQuadSwapHorizontal(data[0].u4.x);
data[invocation].u4.xy = subgroupQuadSwapHorizontal(data[1].u4.xy);
data[invocation].u4.xyz = subgroupQuadSwapHorizontal(data[2].u4.xyz);
data[invocation].u4 = subgroupQuadSwapHorizontal(data[3].u4);
data[invocation].d4.x = subgroupQuadSwapHorizontal(data[0].d4.x);
data[invocation].d4.xy = subgroupQuadSwapHorizontal(data[1].d4.xy);
data[invocation].d4.xyz = subgroupQuadSwapHorizontal(data[2].d4.xyz);
data[invocation].d4 = subgroupQuadSwapHorizontal(data[3].d4);
data[invocation].i4.x = int(subgroupQuadSwapHorizontal(data[0].i4.x < 0));
data[invocation].i4.xy = ivec2(subgroupQuadSwapHorizontal(lessThan(data[1].i4.xy, ivec2(0))));
data[invocation].i4.xyz = ivec3(subgroupQuadSwapHorizontal(lessThan(data[1].i4.xyz, ivec3(0))));
data[invocation].i4 = ivec4(subgroupQuadSwapHorizontal(lessThan(data[1].i4, ivec4(0))));
data[invocation].f4.x = subgroupQuadSwapVertical(data[0].f4.x);
data[invocation].f4.xy = subgroupQuadSwapVertical(data[1].f4.xy);
data[invocation].f4.xyz = subgroupQuadSwapVertical(data[2].f4.xyz);
data[invocation].f4 = subgroupQuadSwapVertical(data[3].f4);
data[invocation].i4.x = subgroupQuadSwapVertical(data[0].i4.x);
data[invocation].i4.xy = subgroupQuadSwapVertical(data[1].i4.xy);
data[invocation].i4.xyz = subgroupQuadSwapVertical(data[2].i4.xyz);
data[invocation].i4 = subgroupQuadSwapVertical(data[3].i4);
data[invocation].u4.x = subgroupQuadSwapVertical(data[0].u4.x);
data[invocation].u4.xy = subgroupQuadSwapVertical(data[1].u4.xy);
data[invocation].u4.xyz = subgroupQuadSwapVertical(data[2].u4.xyz);
data[invocation].u4 = subgroupQuadSwapVertical(data[3].u4);
data[invocation].d4.x = subgroupQuadSwapVertical(data[0].d4.x);
data[invocation].d4.xy = subgroupQuadSwapVertical(data[1].d4.xy);
data[invocation].d4.xyz = subgroupQuadSwapVertical(data[2].d4.xyz);
data[invocation].d4 = subgroupQuadSwapVertical(data[3].d4);
data[invocation].i4.x = int(subgroupQuadSwapVertical(data[0].i4.x < 0));
data[invocation].i4.xy = ivec2(subgroupQuadSwapVertical(lessThan(data[1].i4.xy, ivec2(0))));
data[invocation].i4.xyz = ivec3(subgroupQuadSwapVertical(lessThan(data[1].i4.xyz, ivec3(0))));
data[invocation].i4 = ivec4(subgroupQuadSwapVertical(lessThan(data[1].i4, ivec4(0))));
data[invocation].f4.x = subgroupQuadSwapDiagonal(data[0].f4.x);
data[invocation].f4.xy = subgroupQuadSwapDiagonal(data[1].f4.xy);
data[invocation].f4.xyz = subgroupQuadSwapDiagonal(data[2].f4.xyz);
data[invocation].f4 = subgroupQuadSwapDiagonal(data[3].f4);
data[invocation].i4.x = subgroupQuadSwapDiagonal(data[0].i4.x);
data[invocation].i4.xy = subgroupQuadSwapDiagonal(data[1].i4.xy);
data[invocation].i4.xyz = subgroupQuadSwapDiagonal(data[2].i4.xyz);
data[invocation].i4 = subgroupQuadSwapDiagonal(data[3].i4);
data[invocation].u4.x = subgroupQuadSwapDiagonal(data[0].u4.x);
data[invocation].u4.xy = subgroupQuadSwapDiagonal(data[1].u4.xy);
data[invocation].u4.xyz = subgroupQuadSwapDiagonal(data[2].u4.xyz);
data[invocation].u4 = subgroupQuadSwapDiagonal(data[3].u4);
data[invocation].d4.x = subgroupQuadSwapDiagonal(data[0].d4.x);
data[invocation].d4.xy = subgroupQuadSwapDiagonal(data[1].d4.xy);
data[invocation].d4.xyz = subgroupQuadSwapDiagonal(data[2].d4.xyz);
data[invocation].d4 = subgroupQuadSwapDiagonal(data[3].d4);
data[invocation].i4.x = int(subgroupQuadSwapDiagonal(data[0].i4.x < 0));
data[invocation].i4.xy = ivec2(subgroupQuadSwapDiagonal(lessThan(data[1].i4.xy, ivec2(0))));
data[invocation].i4.xyz = ivec3(subgroupQuadSwapDiagonal(lessThan(data[1].i4.xyz, ivec3(0))));
data[invocation].i4 = ivec4(subgroupQuadSwapDiagonal(lessThan(data[1].i4, ivec4(0))));
}
Test/glsl.450.subgroupShuffle.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_shuffle: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
data[invocation].f4.x = subgroupShuffle(data[0].f4.x, invocation);
data[invocation].f4.xy = subgroupShuffle(data[1].f4.xy, invocation);
data[invocation].f4.xyz = subgroupShuffle(data[2].f4.xyz, invocation);
data[invocation].f4 = subgroupShuffle(data[3].f4, invocation);
data[invocation].i4.x = subgroupShuffle(data[0].i4.x, invocation);
data[invocation].i4.xy = subgroupShuffle(data[1].i4.xy, invocation);
data[invocation].i4.xyz = subgroupShuffle(data[2].i4.xyz, invocation);
data[invocation].i4 = subgroupShuffle(data[3].i4, invocation);
data[invocation].u4.x = subgroupShuffle(data[0].u4.x, invocation);
data[invocation].u4.xy = subgroupShuffle(data[1].u4.xy, invocation);
data[invocation].u4.xyz = subgroupShuffle(data[2].u4.xyz, invocation);
data[invocation].u4 = subgroupShuffle(data[3].u4, invocation);
data[invocation].d4.x = subgroupShuffle(data[0].d4.x, invocation);
data[invocation].d4.xy = subgroupShuffle(data[1].d4.xy, invocation);
data[invocation].d4.xyz = subgroupShuffle(data[2].d4.xyz, invocation);
data[invocation].d4 = subgroupShuffle(data[3].d4, invocation);
data[invocation].i4.x = int(subgroupShuffle(data[0].i4.x < 0, invocation));
data[invocation].i4.xy = ivec2(subgroupShuffle(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[invocation].i4.xyz = ivec3(subgroupShuffle(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[invocation].i4 = ivec4(subgroupShuffle(lessThan(data[1].i4, ivec4(0)), invocation));
data[invocation].f4.x = subgroupShuffleXor(data[0].f4.x, invocation);
data[invocation].f4.xy = subgroupShuffleXor(data[1].f4.xy, invocation);
data[invocation].f4.xyz = subgroupShuffleXor(data[2].f4.xyz, invocation);
data[invocation].f4 = subgroupShuffleXor(data[3].f4, invocation);
data[invocation].i4.x = subgroupShuffleXor(data[0].i4.x, invocation);
data[invocation].i4.xy = subgroupShuffleXor(data[1].i4.xy, invocation);
data[invocation].i4.xyz = subgroupShuffleXor(data[2].i4.xyz, invocation);
data[invocation].i4 = subgroupShuffleXor(data[3].i4, invocation);
data[invocation].u4.x = subgroupShuffleXor(data[0].u4.x, invocation);
data[invocation].u4.xy = subgroupShuffleXor(data[1].u4.xy, invocation);
data[invocation].u4.xyz = subgroupShuffleXor(data[2].u4.xyz, invocation);
data[invocation].u4 = subgroupShuffleXor(data[3].u4, invocation);
data[invocation].d4.x = subgroupShuffleXor(data[0].d4.x, invocation);
data[invocation].d4.xy = subgroupShuffleXor(data[1].d4.xy, invocation);
data[invocation].d4.xyz = subgroupShuffleXor(data[2].d4.xyz, invocation);
data[invocation].d4 = subgroupShuffleXor(data[3].d4, invocation);
data[invocation].i4.x = int(subgroupShuffleXor(data[0].i4.x < 0, invocation));
data[invocation].i4.xy = ivec2(subgroupShuffleXor(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[invocation].i4.xyz = ivec3(subgroupShuffleXor(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[invocation].i4 = ivec4(subgroupShuffleXor(lessThan(data[1].i4, ivec4(0)), invocation));
}
Test/glsl.450.subgroupShuffleRelative.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_shuffle_relative: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
data[invocation].f4.x = subgroupShuffleUp(data[0].f4.x, invocation);
data[invocation].f4.xy = subgroupShuffleUp(data[1].f4.xy, invocation);
data[invocation].f4.xyz = subgroupShuffleUp(data[2].f4.xyz, invocation);
data[invocation].f4 = subgroupShuffleUp(data[3].f4, invocation);
data[invocation].i4.x = subgroupShuffleUp(data[0].i4.x, invocation);
data[invocation].i4.xy = subgroupShuffleUp(data[1].i4.xy, invocation);
data[invocation].i4.xyz = subgroupShuffleUp(data[2].i4.xyz, invocation);
data[invocation].i4 = subgroupShuffleUp(data[3].i4, invocation);
data[invocation].u4.x = subgroupShuffleUp(data[0].u4.x, invocation);
data[invocation].u4.xy = subgroupShuffleUp(data[1].u4.xy, invocation);
data[invocation].u4.xyz = subgroupShuffleUp(data[2].u4.xyz, invocation);
data[invocation].u4 = subgroupShuffleUp(data[3].u4, invocation);
data[invocation].d4.x = subgroupShuffleUp(data[0].d4.x, invocation);
data[invocation].d4.xy = subgroupShuffleUp(data[1].d4.xy, invocation);
data[invocation].d4.xyz = subgroupShuffleUp(data[2].d4.xyz, invocation);
data[invocation].d4 = subgroupShuffleUp(data[3].d4, invocation);
data[invocation].i4.x = int(subgroupShuffleUp(data[0].i4.x < 0, invocation));
data[invocation].i4.xy = ivec2(subgroupShuffleUp(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[invocation].i4.xyz = ivec3(subgroupShuffleUp(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[invocation].i4 = ivec4(subgroupShuffleUp(lessThan(data[1].i4, ivec4(0)), invocation));
data[invocation].f4.x = subgroupShuffleDown(data[0].f4.x, invocation);
data[invocation].f4.xy = subgroupShuffleDown(data[1].f4.xy, invocation);
data[invocation].f4.xyz = subgroupShuffleDown(data[2].f4.xyz, invocation);
data[invocation].f4 = subgroupShuffleDown(data[3].f4, invocation);
data[invocation].i4.x = subgroupShuffleDown(data[0].i4.x, invocation);
data[invocation].i4.xy = subgroupShuffleDown(data[1].i4.xy, invocation);
data[invocation].i4.xyz = subgroupShuffleDown(data[2].i4.xyz, invocation);
data[invocation].i4 = subgroupShuffleDown(data[3].i4, invocation);
data[invocation].u4.x = subgroupShuffleDown(data[0].u4.x, invocation);
data[invocation].u4.xy = subgroupShuffleDown(data[1].u4.xy, invocation);
data[invocation].u4.xyz = subgroupShuffleDown(data[2].u4.xyz, invocation);
data[invocation].u4 = subgroupShuffleDown(data[3].u4, invocation);
data[invocation].d4.x = subgroupShuffleDown(data[0].d4.x, invocation);
data[invocation].d4.xy = subgroupShuffleDown(data[1].d4.xy, invocation);
data[invocation].d4.xyz = subgroupShuffleDown(data[2].d4.xyz, invocation);
data[invocation].d4 = subgroupShuffleDown(data[3].d4, invocation);
data[invocation].i4.x = int(subgroupShuffleDown(data[0].i4.x < 0, invocation));
data[invocation].i4.xy = ivec2(subgroupShuffleDown(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[invocation].i4.xyz = ivec3(subgroupShuffleDown(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[invocation].i4 = ivec4(subgroupShuffleDown(lessThan(data[1].i4, ivec4(0)), invocation));
}
Test/glsl.450.subgroupVote.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_vote: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
dvec4 d4;
int r;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4;
if (subgroupAll(data[invocation].r < 0))
{
data[invocation].r = int(subgroupAllEqual(data[0].f4.x));
data[invocation].r = int(subgroupAllEqual(data[1].f4.xy));
data[invocation].r = int(subgroupAllEqual(data[2].f4.xyz));
data[invocation].r = int(subgroupAllEqual(data[3].f4));
data[invocation].r = int(subgroupAllEqual(data[0].i4.x));
data[invocation].r = int(subgroupAllEqual(data[1].i4.xy));
data[invocation].r = int(subgroupAllEqual(data[2].i4.xyz));
data[invocation].r = int(subgroupAllEqual(data[3].i4));
data[invocation].r = int(subgroupAllEqual(data[0].u4.x));
data[invocation].r = int(subgroupAllEqual(data[1].u4.xy));
data[invocation].r = int(subgroupAllEqual(data[2].u4.xyz));
data[invocation].r = int(subgroupAllEqual(data[3].u4));
}
else if (subgroupAny(data[invocation].r < 0))
{
data[invocation].r = int(subgroupAllEqual(data[0].d4.x));
data[invocation].r = int(subgroupAllEqual(data[1].d4.xy));
data[invocation].r = int(subgroupAllEqual(data[2].d4.xyz));
data[invocation].r = int(subgroupAllEqual(data[3].d4));
data[invocation].r = int(int(subgroupAllEqual(data[0].i4.x < 0)));
data[invocation].r = int(ivec2(subgroupAllEqual(lessThan(data[1].i4.xy, ivec2(0)))));
data[invocation].r = int(ivec3(subgroupAllEqual(lessThan(data[1].i4.xyz, ivec3(0)))));
data[invocation].r = int(ivec4(subgroupAllEqual(lessThan(data[1].i4, ivec4(0)))));
}
}
Test/glsl.es320.subgroup.frag 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout(location = 0) out uvec4 data;
void main (void)
{
data = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.es320.subgroup.geom 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout(points) in;
layout(points, max_vertices = 1) out;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveIDIn] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.es320.subgroup.tesc 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout(vertices=1) out;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.es320.subgroup.tese 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout(isolines) in;
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_PrimitiveID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.es320.subgroup.vert 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout(set = 0, binding = 0, std430) buffer Output
{
uvec4 result[];
};
void main (void)
{
result[gl_VertexID] = uvec4(gl_SubgroupSize, gl_SubgroupInvocationID, 0, 0);
}
Test/glsl.es320.subgroupBallot.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_ballot: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
uvec4 relMask = gl_SubgroupEqMask +
gl_SubgroupGeMask +
gl_SubgroupGtMask +
gl_SubgroupLeMask +
gl_SubgroupLtMask;
uvec4 result = subgroupBallot(true);
data[0].u4.x = subgroupBallotBitCount(result);
data[0].u4.y = subgroupBallotBitExtract(result, 0u) ? 1u : 0u;
data[0].u4.z = subgroupBallotInclusiveBitCount(result) + subgroupBallotExclusiveBitCount(result);
data[0].u4.w = subgroupBallotFindLSB(result) + subgroupBallotFindMSB(result);
if ((relMask == result) && subgroupInverseBallot(data[0].u4))
{
data[1].f4.x = subgroupBroadcast(data[0].f4.x, 3u);
data[1].f4.xy = subgroupBroadcast(data[1].f4.xy, 3u);
data[1].f4.xyz = subgroupBroadcast(data[2].f4.xyz, 3u);
data[1].f4 = subgroupBroadcast(data[3].f4, 3u);
data[2].i4.x = subgroupBroadcast(data[0].i4.x, 2u);
data[2].i4.xy = subgroupBroadcast(data[1].i4.xy, 2u);
data[2].i4.xyz = subgroupBroadcast(data[2].i4.xyz, 2u);
data[2].i4 = subgroupBroadcast(data[3].i4, 2u);
data[3].u4.x = subgroupBroadcast(data[0].u4.x, 1u);
data[3].u4.xy = subgroupBroadcast(data[1].u4.xy, 1u);
data[3].u4.xyz = subgroupBroadcast(data[2].u4.xyz, 1u);
data[3].u4 = subgroupBroadcast(data[3].u4, 1u);
data[0].i4.x = int(subgroupBroadcast(data[0].i4.x < 0, 1u));
data[0].i4.xy = ivec2(subgroupBroadcast(lessThan(data[1].i4.xy, ivec2(0)), 1u));
data[0].i4.xyz = ivec3(subgroupBroadcast(lessThan(data[1].i4.xyz, ivec3(0)), 1u));
data[0].i4 = ivec4(subgroupBroadcast(lessThan(data[1].i4, ivec4(0)), 1u));
}
else
{
data[1].f4.x = subgroupBroadcastFirst(data[0].f4.x);
data[1].f4.xy = subgroupBroadcastFirst(data[1].f4.xy);
data[1].f4.xyz = subgroupBroadcastFirst(data[2].f4.xyz);
data[1].f4 = subgroupBroadcastFirst(data[3].f4);
data[2].i4.x = subgroupBroadcastFirst(data[0].i4.x);
data[2].i4.xy = subgroupBroadcastFirst(data[1].i4.xy);
data[2].i4.xyz = subgroupBroadcastFirst(data[2].i4.xyz);
data[2].i4 = subgroupBroadcastFirst(data[3].i4);
data[3].u4.x = subgroupBroadcastFirst(data[0].u4.x);
data[3].u4.xy = subgroupBroadcastFirst(data[1].u4.xy);
data[3].u4.xyz = subgroupBroadcastFirst(data[2].u4.xyz);
data[3].u4 = subgroupBroadcastFirst(data[3].u4);
data[0].i4.x = int(subgroupBroadcastFirst(data[0].i4.x < 0));
data[0].i4.xy = ivec2(subgroupBroadcastFirst(lessThan(data[1].i4.xy, ivec2(0))));
data[0].i4.xyz = ivec3(subgroupBroadcastFirst(lessThan(data[1].i4.xyz, ivec3(0))));
data[0].i4 = ivec4(subgroupBroadcastFirst(lessThan(data[1].i4, ivec4(0))));
}
}
Test/glsl.es320.subgroupBallotNeg.comp 0 → 100644
#version 450
#extension GL_KHR_shader_subgroup_ballot: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
uvec4 relMask = gl_SubgroupEqMask +
gl_SubgroupGeMask +
gl_SubgroupGtMask +
gl_SubgroupLeMask +
gl_SubgroupLtMask;
uvec4 result = subgroupBallot(true);
data[0].u4.x = subgroupBallotBitCount(result);
data[0].u4.y = subgroupBallotBitExtract(result, 0) ? 1u : 0u;
data[0].u4.z = subgroupBallotInclusiveBitCount(result) + subgroupBallotExclusiveBitCount(result);
data[0].u4.w = subgroupBallotFindLSB(result) + subgroupBallotFindMSB(result);
data[1].f4.x = subgroupBroadcast(data[0].f4.x, invocation); // ERROR: not constant
}
Test/glsl.es320.subgroupBasic.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_basic: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffer
{
int a[];
} data;
void main()
{
data.a[gl_SubgroupSize] = 1;
data.a[gl_SubgroupInvocationID] = 1;
data.a[gl_NumSubgroups] = 1;
data.a[gl_SubgroupID] = (subgroupElect()) ? 1 : 0;
subgroupBarrier();
subgroupMemoryBarrier();
subgroupMemoryBarrierBuffer();
subgroupMemoryBarrierShared();
subgroupMemoryBarrierImage();
}
Test/glsl.es320.subgroupClustered.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_clustered: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
data[0].f4.x = subgroupClusteredAdd(data[0].f4.x, 1u);
data[0].f4.xy = subgroupClusteredAdd(data[1].f4.xy, 1u);
data[0].f4.xyz = subgroupClusteredAdd(data[2].f4.xyz, 1u);
data[0].f4 = subgroupClusteredAdd(data[3].f4, 1u);
data[1].i4.x = subgroupClusteredAdd(data[0].i4.x, 1u);
data[1].i4.xy = subgroupClusteredAdd(data[1].i4.xy, 1u);
data[1].i4.xyz = subgroupClusteredAdd(data[2].i4.xyz, 1u);
data[1].i4 = subgroupClusteredAdd(data[3].i4, 1u);
data[2].u4.x = subgroupClusteredAdd(data[0].u4.x, 1u);
data[2].u4.xy = subgroupClusteredAdd(data[1].u4.xy, 1u);
data[2].u4.xyz = subgroupClusteredAdd(data[2].u4.xyz, 1u);
data[2].u4 = subgroupClusteredAdd(data[3].u4, 1u);
data[3].f4.x = subgroupClusteredMul(data[0].f4.x, 1u);
data[3].f4.xy = subgroupClusteredMul(data[1].f4.xy, 1u);
data[3].f4.xyz = subgroupClusteredMul(data[2].f4.xyz, 1u);
data[3].f4 = subgroupClusteredMul(data[3].f4, 1u);
data[0].i4.x = subgroupClusteredMul(data[0].i4.x, 1u);
data[0].i4.xy = subgroupClusteredMul(data[1].i4.xy, 1u);
data[0].i4.xyz = subgroupClusteredMul(data[2].i4.xyz, 1u);
data[0].i4 = subgroupClusteredMul(data[3].i4, 1u);
data[1].u4.x = subgroupClusteredMul(data[0].u4.x, 1u);
data[1].u4.xy = subgroupClusteredMul(data[1].u4.xy, 1u);
data[1].u4.xyz = subgroupClusteredMul(data[2].u4.xyz, 1u);
data[1].u4 = subgroupClusteredMul(data[3].u4, 1u);
data[2].f4.x = subgroupClusteredMin(data[0].f4.x, 1u);
data[2].f4.xy = subgroupClusteredMin(data[1].f4.xy, 1u);
data[2].f4.xyz = subgroupClusteredMin(data[2].f4.xyz, 1u);
data[2].f4 = subgroupClusteredMin(data[3].f4, 1u);
data[3].i4.x = subgroupClusteredMin(data[0].i4.x, 1u);
data[3].i4.xy = subgroupClusteredMin(data[1].i4.xy, 1u);
data[3].i4.xyz = subgroupClusteredMin(data[2].i4.xyz, 1u);
data[3].i4 = subgroupClusteredMin(data[3].i4, 1u);
data[0].u4.x = subgroupClusteredMin(data[0].u4.x, 1u);
data[0].u4.xy = subgroupClusteredMin(data[1].u4.xy, 1u);
data[0].u4.xyz = subgroupClusteredMin(data[2].u4.xyz, 1u);
data[0].u4 = subgroupClusteredMin(data[3].u4, 1u);
data[1].f4.x = subgroupClusteredMax(data[0].f4.x, 1u);
data[1].f4.xy = subgroupClusteredMax(data[1].f4.xy, 1u);
data[1].f4.xyz = subgroupClusteredMax(data[2].f4.xyz, 1u);
data[1].f4 = subgroupClusteredMax(data[3].f4, 1u);
data[2].i4.x = subgroupClusteredMax(data[0].i4.x, 1u);
data[2].i4.xy = subgroupClusteredMax(data[1].i4.xy, 1u);
data[2].i4.xyz = subgroupClusteredMax(data[2].i4.xyz, 1u);
data[2].i4 = subgroupClusteredMax(data[3].i4, 1u);
data[3].u4.x = subgroupClusteredMax(data[0].u4.x, 1u);
data[3].u4.xy = subgroupClusteredMax(data[1].u4.xy, 1u);
data[3].u4.xyz = subgroupClusteredMax(data[2].u4.xyz, 1u);
data[3].u4 = subgroupClusteredMax(data[3].u4, 1u);
data[0].i4.x = subgroupClusteredAnd(data[0].i4.x, 1u);
data[0].i4.xy = subgroupClusteredAnd(data[1].i4.xy, 1u);
data[0].i4.xyz = subgroupClusteredAnd(data[2].i4.xyz, 1u);
data[0].i4 = subgroupClusteredAnd(data[3].i4, 1u);
data[1].u4.x = subgroupClusteredAnd(data[0].u4.x, 1u);
data[1].u4.xy = subgroupClusteredAnd(data[1].u4.xy, 1u);
data[1].u4.xyz = subgroupClusteredAnd(data[2].u4.xyz, 1u);
data[1].u4 = subgroupClusteredAnd(data[3].u4, 1u);
data[2].i4.x = int(subgroupClusteredAnd(data[0].i4.x < 0, 1u));
data[2].i4.xy = ivec2(subgroupClusteredAnd(lessThan(data[1].i4.xy, ivec2(0)), 1u));
data[2].i4.xyz = ivec3(subgroupClusteredAnd(lessThan(data[1].i4.xyz, ivec3(0)), 1u));
data[2].i4 = ivec4(subgroupClusteredAnd(lessThan(data[1].i4, ivec4(0)), 1u));
data[3].i4.x = subgroupClusteredOr(data[0].i4.x, 1u);
data[3].i4.xy = subgroupClusteredOr(data[1].i4.xy, 1u);
data[3].i4.xyz = subgroupClusteredOr(data[2].i4.xyz, 1u);
data[3].i4 = subgroupClusteredOr(data[3].i4, 1u);
data[0].u4.x = subgroupClusteredOr(data[0].u4.x, 1u);
data[0].u4.xy = subgroupClusteredOr(data[1].u4.xy, 1u);
data[0].u4.xyz = subgroupClusteredOr(data[2].u4.xyz, 1u);
data[0].u4 = subgroupClusteredOr(data[3].u4, 1u);
data[1].i4.x = int(subgroupClusteredOr(data[0].i4.x < 0, 1u));
data[1].i4.xy = ivec2(subgroupClusteredOr(lessThan(data[1].i4.xy, ivec2(0)), 1u));
data[1].i4.xyz = ivec3(subgroupClusteredOr(lessThan(data[1].i4.xyz, ivec3(0)), 1u));
data[1].i4 = ivec4(subgroupClusteredOr(lessThan(data[1].i4, ivec4(0)), 1u));
data[2].i4.x = subgroupClusteredXor(data[0].i4.x, 1u);
data[2].i4.xy = subgroupClusteredXor(data[1].i4.xy, 1u);
data[2].i4.xyz = subgroupClusteredXor(data[2].i4.xyz, 1u);
data[2].i4 = subgroupClusteredXor(data[3].i4, 1u);
data[3].u4.x = subgroupClusteredXor(data[0].u4.x, 1u);
data[3].u4.xy = subgroupClusteredXor(data[1].u4.xy, 1u);
data[3].u4.xyz = subgroupClusteredXor(data[2].u4.xyz, 1u);
data[3].u4 = subgroupClusteredXor(data[3].u4, 1u);
data[0].i4.x = int(subgroupClusteredXor(data[0].i4.x < 0, 1u));
data[0].i4.xy = ivec2(subgroupClusteredXor(lessThan(data[1].i4.xy, ivec2(0)), 1u));
data[0].i4.xyz = ivec3(subgroupClusteredXor(lessThan(data[1].i4.xyz, ivec3(0)), 1u));
data[0].i4 = ivec4(subgroupClusteredXor(lessThan(data[1].i4, ivec4(0)), 1u));
}
Test/glsl.es320.subgroupClusteredNeg.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_clustered: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
int a = 1;
const int aConst = 1;
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
data[0].f4.xy = subgroupClusteredAdd(data[1].f4.xy, 0u); // ERROR, less than 1
data[0].f4.x = subgroupClusteredMul(data[0].f4.x, 3u); // ERROR, not a power of 2
data[1].i4.xy = subgroupClusteredMin(data[1].i4.xy, 8u);
data[1].i4.xyz = subgroupClusteredMin(data[2].i4.xyz, 6u); // ERROR, not a power of 2
data[3].i4.x = subgroupClusteredOr(data[0].i4.x, uint(a)); // ERROR, not constant
data[3].i4.xy = subgroupClusteredOr(data[1].i4.xy, uint(aConst));
data[0].i4.x = subgroupClusteredXor(data[0].i4.x, uint(1 + a)); // ERROR, not constant
data[0].i4.xy = subgroupClusteredXor(data[1].i4.xy, uint(aConst + a)); // ERROR, not constant
data[0].i4.xyz = subgroupClusteredXor(data[2].i4.xyz, uint(1 + aConst));
}
Test/glsl.es320.subgroupQuad.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_quad: enable
layout (local_size_x = 8) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
data[0].f4.x = subgroupQuadBroadcast(data[0].f4.x, 1u);
data[0].f4.xy = subgroupQuadBroadcast(data[1].f4.xy, 1u);
data[0].f4.xyz = subgroupQuadBroadcast(data[2].f4.xyz, 1u);
data[0].f4 = subgroupQuadBroadcast(data[3].f4, 1u);
data[0].i4.x = subgroupQuadBroadcast(data[0].i4.x, 1u);
data[0].i4.xy = subgroupQuadBroadcast(data[1].i4.xy, 1u);
data[0].i4.xyz = subgroupQuadBroadcast(data[2].i4.xyz, 1u);
data[0].i4 = subgroupQuadBroadcast(data[3].i4, 1u);
data[0].u4.x = subgroupQuadBroadcast(data[0].u4.x, 1u);
data[0].u4.xy = subgroupQuadBroadcast(data[1].u4.xy, 1u);
data[0].u4.xyz = subgroupQuadBroadcast(data[2].u4.xyz, 1u);
data[0].u4 = subgroupQuadBroadcast(data[3].u4, 1u);
data[1].i4.x = int(subgroupQuadBroadcast(data[0].i4.x < 0, 1u));
data[1].i4.xy = ivec2(subgroupQuadBroadcast(lessThan(data[1].i4.xy, ivec2(0)), 1u));
data[1].i4.xyz = ivec3(subgroupQuadBroadcast(lessThan(data[1].i4.xyz, ivec3(0)), 1u));
data[1].i4 = ivec4(subgroupQuadBroadcast(lessThan(data[1].i4, ivec4(0)), 1u));
data[1].f4.x = subgroupQuadSwapHorizontal(data[0].f4.x);
data[1].f4.xy = subgroupQuadSwapHorizontal(data[1].f4.xy);
data[1].f4.xyz = subgroupQuadSwapHorizontal(data[2].f4.xyz);
data[1].f4 = subgroupQuadSwapHorizontal(data[3].f4);
data[1].i4.x = subgroupQuadSwapHorizontal(data[0].i4.x);
data[1].i4.xy = subgroupQuadSwapHorizontal(data[1].i4.xy);
data[1].i4.xyz = subgroupQuadSwapHorizontal(data[2].i4.xyz);
data[1].i4 = subgroupQuadSwapHorizontal(data[3].i4);
data[1].u4.x = subgroupQuadSwapHorizontal(data[0].u4.x);
data[1].u4.xy = subgroupQuadSwapHorizontal(data[1].u4.xy);
data[1].u4.xyz = subgroupQuadSwapHorizontal(data[2].u4.xyz);
data[1].u4 = subgroupQuadSwapHorizontal(data[3].u4);
data[2].i4.x = int(subgroupQuadSwapHorizontal(data[0].i4.x < 0));
data[2].i4.xy = ivec2(subgroupQuadSwapHorizontal(lessThan(data[1].i4.xy, ivec2(0))));
data[2].i4.xyz = ivec3(subgroupQuadSwapHorizontal(lessThan(data[1].i4.xyz, ivec3(0))));
data[2].i4 = ivec4(subgroupQuadSwapHorizontal(lessThan(data[1].i4, ivec4(0))));
data[2].f4.x = subgroupQuadSwapVertical(data[0].f4.x);
data[2].f4.xy = subgroupQuadSwapVertical(data[1].f4.xy);
data[2].f4.xyz = subgroupQuadSwapVertical(data[2].f4.xyz);
data[2].f4 = subgroupQuadSwapVertical(data[3].f4);
data[2].i4.x = subgroupQuadSwapVertical(data[0].i4.x);
data[2].i4.xy = subgroupQuadSwapVertical(data[1].i4.xy);
data[2].i4.xyz = subgroupQuadSwapVertical(data[2].i4.xyz);
data[2].i4 = subgroupQuadSwapVertical(data[3].i4);
data[2].u4.x = subgroupQuadSwapVertical(data[0].u4.x);
data[2].u4.xy = subgroupQuadSwapVertical(data[1].u4.xy);
data[2].u4.xyz = subgroupQuadSwapVertical(data[2].u4.xyz);
data[2].u4 = subgroupQuadSwapVertical(data[3].u4);
data[3].i4.x = int(subgroupQuadSwapVertical(data[0].i4.x < 0));
data[3].i4.xy = ivec2(subgroupQuadSwapVertical(lessThan(data[1].i4.xy, ivec2(0))));
data[3].i4.xyz = ivec3(subgroupQuadSwapVertical(lessThan(data[1].i4.xyz, ivec3(0))));
data[3].i4 = ivec4(subgroupQuadSwapVertical(lessThan(data[1].i4, ivec4(0))));
data[3].f4.x = subgroupQuadSwapDiagonal(data[0].f4.x);
data[3].f4.xy = subgroupQuadSwapDiagonal(data[1].f4.xy);
data[3].f4.xyz = subgroupQuadSwapDiagonal(data[2].f4.xyz);
data[3].f4 = subgroupQuadSwapDiagonal(data[3].f4);
data[3].i4.x = subgroupQuadSwapDiagonal(data[0].i4.x);
data[3].i4.xy = subgroupQuadSwapDiagonal(data[1].i4.xy);
data[3].i4.xyz = subgroupQuadSwapDiagonal(data[2].i4.xyz);
data[3].i4 = subgroupQuadSwapDiagonal(data[3].i4);
data[3].u4.x = subgroupQuadSwapDiagonal(data[0].u4.x);
data[3].u4.xy = subgroupQuadSwapDiagonal(data[1].u4.xy);
data[3].u4.xyz = subgroupQuadSwapDiagonal(data[2].u4.xyz);
data[3].u4 = subgroupQuadSwapDiagonal(data[3].u4);
data[3].i4.x = int(subgroupQuadSwapDiagonal(data[0].i4.x < 0));
data[3].i4.xy = ivec2(subgroupQuadSwapDiagonal(lessThan(data[1].i4.xy, ivec2(0))));
data[3].i4.xyz = ivec3(subgroupQuadSwapDiagonal(lessThan(data[1].i4.xyz, ivec3(0))));
data[3].i4 = ivec4(subgroupQuadSwapDiagonal(lessThan(data[1].i4, ivec4(0))));
}
Test/glsl.es320.subgroupShuffle.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_shuffle: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
data[0].f4.x = subgroupShuffle(data[0].f4.x, invocation);
data[0].f4.xy = subgroupShuffle(data[1].f4.xy, invocation);
data[0].f4.xyz = subgroupShuffle(data[2].f4.xyz, invocation);
data[0].f4 = subgroupShuffle(data[3].f4, invocation);
data[0].i4.x = subgroupShuffle(data[0].i4.x, invocation);
data[0].i4.xy = subgroupShuffle(data[1].i4.xy, invocation);
data[0].i4.xyz = subgroupShuffle(data[2].i4.xyz, invocation);
data[0].i4 = subgroupShuffle(data[3].i4, invocation);
data[1].u4.x = subgroupShuffle(data[0].u4.x, invocation);
data[1].u4.xy = subgroupShuffle(data[1].u4.xy, invocation);
data[1].u4.xyz = subgroupShuffle(data[2].u4.xyz, invocation);
data[1].u4 = subgroupShuffle(data[3].u4, invocation);
data[1].i4.x = int(subgroupShuffle(data[0].i4.x < 0, invocation));
data[1].i4.xy = ivec2(subgroupShuffle(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[1].i4.xyz = ivec3(subgroupShuffle(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[1].i4 = ivec4(subgroupShuffle(lessThan(data[1].i4, ivec4(0)), invocation));
data[2].f4.x = subgroupShuffleXor(data[0].f4.x, invocation);
data[2].f4.xy = subgroupShuffleXor(data[1].f4.xy, invocation);
data[2].f4.xyz = subgroupShuffleXor(data[2].f4.xyz, invocation);
data[2].f4 = subgroupShuffleXor(data[3].f4, invocation);
data[2].i4.x = subgroupShuffleXor(data[0].i4.x, invocation);
data[2].i4.xy = subgroupShuffleXor(data[1].i4.xy, invocation);
data[2].i4.xyz = subgroupShuffleXor(data[2].i4.xyz, invocation);
data[2].i4 = subgroupShuffleXor(data[3].i4, invocation);
data[3].u4.x = subgroupShuffleXor(data[0].u4.x, invocation);
data[3].u4.xy = subgroupShuffleXor(data[1].u4.xy, invocation);
data[3].u4.xyz = subgroupShuffleXor(data[2].u4.xyz, invocation);
data[3].u4 = subgroupShuffleXor(data[3].u4, invocation);
data[3].i4.x = int(subgroupShuffleXor(data[0].i4.x < 0, invocation));
data[3].i4.xy = ivec2(subgroupShuffleXor(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[3].i4.xyz = ivec3(subgroupShuffleXor(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[3].i4 = ivec4(subgroupShuffleXor(lessThan(data[1].i4, ivec4(0)), invocation));
}
Test/glsl.es320.subgroupShuffleRelative.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_shuffle_relative: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
data[0].f4.x = subgroupShuffleUp(data[0].f4.x, invocation);
data[0].f4.xy = subgroupShuffleUp(data[1].f4.xy, invocation);
data[0].f4.xyz = subgroupShuffleUp(data[2].f4.xyz, invocation);
data[0].f4 = subgroupShuffleUp(data[3].f4, invocation);
data[0].i4.x = subgroupShuffleUp(data[0].i4.x, invocation);
data[0].i4.xy = subgroupShuffleUp(data[1].i4.xy, invocation);
data[0].i4.xyz = subgroupShuffleUp(data[2].i4.xyz, invocation);
data[0].i4 = subgroupShuffleUp(data[3].i4, invocation);
data[1].u4.x = subgroupShuffleUp(data[0].u4.x, invocation);
data[1].u4.xy = subgroupShuffleUp(data[1].u4.xy, invocation);
data[1].u4.xyz = subgroupShuffleUp(data[2].u4.xyz, invocation);
data[1].u4 = subgroupShuffleUp(data[3].u4, invocation);
data[1].i4.x = int(subgroupShuffleUp(data[0].i4.x < 0, invocation));
data[1].i4.xy = ivec2(subgroupShuffleUp(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[1].i4.xyz = ivec3(subgroupShuffleUp(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[1].i4 = ivec4(subgroupShuffleUp(lessThan(data[1].i4, ivec4(0)), invocation));
data[2].f4.x = subgroupShuffleDown(data[0].f4.x, invocation);
data[2].f4.xy = subgroupShuffleDown(data[1].f4.xy, invocation);
data[2].f4.xyz = subgroupShuffleDown(data[2].f4.xyz, invocation);
data[2].f4 = subgroupShuffleDown(data[3].f4, invocation);
data[2].i4.x = subgroupShuffleDown(data[0].i4.x, invocation);
data[2].i4.xy = subgroupShuffleDown(data[1].i4.xy, invocation);
data[2].i4.xyz = subgroupShuffleDown(data[2].i4.xyz, invocation);
data[2].i4 = subgroupShuffleDown(data[3].i4, invocation);
data[3].u4.x = subgroupShuffleDown(data[0].u4.x, invocation);
data[3].u4.xy = subgroupShuffleDown(data[1].u4.xy, invocation);
data[3].u4.xyz = subgroupShuffleDown(data[2].u4.xyz, invocation);
data[3].u4 = subgroupShuffleDown(data[3].u4, invocation);
data[3].i4.x = int(subgroupShuffleDown(data[0].i4.x < 0, invocation));
data[3].i4.xy = ivec2(subgroupShuffleDown(lessThan(data[1].i4.xy, ivec2(0)), invocation));
data[3].i4.xyz = ivec3(subgroupShuffleDown(lessThan(data[1].i4.xyz, ivec3(0)), invocation));
data[3].i4 = ivec4(subgroupShuffleDown(lessThan(data[1].i4, ivec4(0)), invocation));
}
Test/glsl.es320.subgroupVote.comp 0 → 100644
#version 320 es
#extension GL_KHR_shader_subgroup_vote: enable
layout (local_size_x = 8, local_size_y = 8, local_size_z = 1) in;
layout(binding = 0) buffer Buffers
{
vec4 f4;
ivec4 i4;
uvec4 u4;
int r;
} data[4];
void main()
{
uint invocation = (gl_SubgroupInvocationID + gl_SubgroupSize) % 4u;
if (subgroupAll(data[0].r < 0))
{
data[0].r = int(subgroupAllEqual(data[0].f4.x));
data[0].r = int(subgroupAllEqual(data[1].f4.xy));
data[0].r = int(subgroupAllEqual(data[2].f4.xyz));
data[0].r = int(subgroupAllEqual(data[3].f4));
data[0].r = int(subgroupAllEqual(data[0].i4.x));
data[0].r = int(subgroupAllEqual(data[1].i4.xy));
data[0].r = int(subgroupAllEqual(data[2].i4.xyz));
data[0].r = int(subgroupAllEqual(data[3].i4));
data[0].r = int(subgroupAllEqual(data[0].u4.x));
data[0].r = int(subgroupAllEqual(data[1].u4.xy));
data[0].r = int(subgroupAllEqual(data[2].u4.xyz));
data[0].r = int(subgroupAllEqual(data[3].u4));
}
else if (subgroupAny(data[1].r < 0))
{
data[1].r = int(int(subgroupAllEqual(data[0].i4.x < 0)));
data[1].r = int(ivec2(subgroupAllEqual(lessThan(data[1].i4.xy, ivec2(0)))));
data[1].r = int(ivec3(subgroupAllEqual(lessThan(data[1].i4.xyz, ivec3(0)))));
data[1].r = int(ivec4(subgroupAllEqual(lessThan(data[1].i4, ivec4(0)))));
}
}
glslang/MachineIndependent/Initialize.cpp
......@@ -1873,7 +1873,8 @@ void TBuiltIns::initialize(int version, EProfile profile, const SpvVersion& spvV
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
commonBuiltins.append(
"void subgroupBarrier();"
"void subgroupMemoryBarrier();"
......@@ -6126,7 +6127,8 @@ void TBuiltIns::initialize(int version, EProfile profile, const SpvVersion& spvV
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
const char* ballotDecls =
"in mediump uint gl_SubgroupSize;"
"in mediump uint gl_SubgroupInvocationID;"
......@@ -8160,7 +8162,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_SubgroupSize", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupInvocationID", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupEqMask", 1, &E_GL_KHR_shader_subgroup_ballot);
......@@ -8479,7 +8482,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_SubgroupSize", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupInvocationID", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupEqMask", 1, &E_GL_KHR_shader_subgroup_ballot);
......@@ -8663,7 +8667,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_SubgroupSize", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupInvocationID", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupEqMask", 1, &E_GL_KHR_shader_subgroup_ballot);
......@@ -8690,7 +8695,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_NumSubgroups", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupID", 1, &E_GL_KHR_shader_subgroup_basic);
......@@ -8859,7 +8865,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_NumSubgroups", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupID", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupSize", 1, &E_GL_KHR_shader_subgroup_basic);
......@@ -8948,7 +8955,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.setVariableExtensions("gl_NumSubgroups", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupID", 1, &E_GL_KHR_shader_subgroup_basic);
symbolTable.setVariableExtensions("gl_SubgroupSize", 1, &E_GL_KHR_shader_subgroup_basic);
......@@ -9365,7 +9373,8 @@ void TBuiltIns::identifyBuiltIns(int version, EProfile profile, const SpvVersion
}
// GL_KHR_shader_subgroup
if (spvVersion.vulkan > 0) {
if ((profile == EEsProfile && version >= 310) ||
(profile != EEsProfile && version >= 140)) {
symbolTable.relateToOperator("subgroupBarrier", EOpSubgroupBarrier);
symbolTable.relateToOperator("subgroupMemoryBarrier", EOpSubgroupMemoryBarrier);
symbolTable.relateToOperator("subgroupMemoryBarrierBuffer", EOpSubgroupMemoryBarrierBuffer);
......
gtests/AST.FromFile.cpp
......@@ -236,7 +236,39 @@ INSTANTIATE_TEST_CASE_P(
"precise_struct_block.vert",
"maxClipDistances.vert",
"findFunction.frag",
"constantUnaryConversion.comp"
"constantUnaryConversion.comp",
"glsl.450.subgroup.frag",
"glsl.450.subgroup.geom",
"glsl.450.subgroup.tesc",
"glsl.450.subgroup.tese",
"glsl.450.subgroup.vert",
"glsl.450.subgroupArithmetic.comp",
"glsl.450.subgroupBasic.comp",
"glsl.450.subgroupBallot.comp",
"glsl.450.subgroupBallotNeg.comp",
"glsl.450.subgroupClustered.comp",
"glsl.450.subgroupClusteredNeg.comp",
"glsl.450.subgroupPartitioned.comp",
"glsl.450.subgroupShuffle.comp",
"glsl.450.subgroupShuffleRelative.comp",
"glsl.450.subgroupQuad.comp",
"glsl.450.subgroupVote.comp",
"glsl.es320.subgroup.frag",
"glsl.es320.subgroup.geom",
"glsl.es320.subgroup.tesc",
"glsl.es320.subgroup.tese",
"glsl.es320.subgroup.vert",
"glsl.es320.subgroupArithmetic.comp",
"glsl.es320.subgroupBasic.comp",
"glsl.es320.subgroupBallot.comp",
"glsl.es320.subgroupBallotNeg.comp",
"glsl.es320.subgroupClustered.comp",
"glsl.es320.subgroupClusteredNeg.comp",
"glsl.es320.subgroupPartitioned.comp",
"glsl.es320.subgroupShuffle.comp",
"glsl.es320.subgroupShuffleRelative.comp",
"glsl.es320.subgroupQuad.comp",
"glsl.es320.subgroupVote.comp",
})),
FileNameAsCustomTestSuffix
);
......
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