| 1 |
#version 450 |
| 2 |
#extension GL_KHR_shader_subgroup_basic : require |
| 3 |
#extension GL_KHR_shader_subgroup_ballot : require |
| 4 |
#extension GL_KHR_shader_subgroup_vote : require |
| 5 |
#extension GL_KHR_shader_subgroup_shuffle : require |
| 6 |
#extension GL_KHR_shader_subgroup_shuffle_relative : require |
| 7 |
#extension GL_KHR_shader_subgroup_arithmetic : require |
| 8 |
#extension GL_KHR_shader_subgroup_clustered : require |
| 9 |
#extension GL_KHR_shader_subgroup_quad : require |
| 10 |
layout(local_size_x = 1) in; |
| 11 |
|
| 12 |
layout(std430, binding = 0) buffer SSBO |
| 13 |
{ |
| 14 |
float FragColor; |
| 15 |
}; |
| 16 |
|
| 17 |
void main() |
| 18 |
{ |
| 19 |
// basic |
| 20 |
//FragColor = float(gl_NumSubgroups); |
| 21 |
//FragColor = float(gl_SubgroupID); |
| 22 |
FragColor = float(gl_SubgroupSize); |
| 23 |
FragColor = float(gl_SubgroupInvocationID); |
| 24 |
subgroupBarrier(); |
| 25 |
subgroupMemoryBarrier(); |
| 26 |
subgroupMemoryBarrierBuffer(); |
| 27 |
subgroupMemoryBarrierShared(); |
| 28 |
subgroupMemoryBarrierImage(); |
| 29 |
bool elected = subgroupElect(); |
| 30 |
|
| 31 |
// ballot |
| 32 |
FragColor = float(gl_SubgroupEqMask); |
| 33 |
FragColor = float(gl_SubgroupGeMask); |
| 34 |
FragColor = float(gl_SubgroupGtMask); |
| 35 |
FragColor = float(gl_SubgroupLeMask); |
| 36 |
FragColor = float(gl_SubgroupLtMask); |
| 37 |
vec4 broadcasted = subgroupBroadcast(vec4(10.0), 8u); |
| 38 |
vec3 first = subgroupBroadcastFirst(vec3(20.0)); |
| 39 |
uvec4 ballot_value = subgroupBallot(true); |
| 40 |
//bool inverse_ballot_value = subgroupInverseBallot(ballot_value); |
| 41 |
//bool bit_extracted = subgroupBallotBitExtract(uvec4(10u), 8u); |
| 42 |
uint bit_count = subgroupBallotBitCount(ballot_value); |
| 43 |
//uint inclusive_bit_count = subgroupBallotInclusiveBitCount(ballot_value); |
| 44 |
//uint exclusive_bit_count = subgroupBallotExclusiveBitCount(ballot_value); |
| 45 |
//uint lsb = subgroupBallotFindLSB(ballot_value); |
| 46 |
//uint msb = subgroupBallotFindMSB(ballot_value); |
| 47 |
|
| 48 |
// shuffle |
| 49 |
//uint shuffled = subgroupShuffle(10u, 8u); |
| 50 |
//uint shuffled_xor = subgroupShuffleXor(30u, 8u); |
| 51 |
|
| 52 |
// shuffle relative |
| 53 |
//uint shuffled_up = subgroupShuffleUp(20u, 4u); |
| 54 |
//uint shuffled_down = subgroupShuffleDown(20u, 4u); |
| 55 |
|
| 56 |
// vote |
| 57 |
bool has_all = subgroupAll(true); |
| 58 |
bool has_any = subgroupAny(true); |
| 59 |
bool has_equal = subgroupAllEqual(true); |
| 60 |
|
| 61 |
// arithmetic |
| 62 |
vec4 added = subgroupAdd(vec4(20.0)); |
| 63 |
ivec4 iadded = subgroupAdd(ivec4(20)); |
| 64 |
vec4 multiplied = subgroupMul(vec4(20.0)); |
| 65 |
ivec4 imultiplied = subgroupMul(ivec4(20)); |
| 66 |
vec4 lo = subgroupMin(vec4(20.0)); |
| 67 |
vec4 hi = subgroupMax(vec4(20.0)); |
| 68 |
ivec4 slo = subgroupMin(ivec4(20)); |
| 69 |
ivec4 shi = subgroupMax(ivec4(20)); |
| 70 |
uvec4 ulo = subgroupMin(uvec4(20)); |
| 71 |
uvec4 uhi = subgroupMax(uvec4(20)); |
| 72 |
uvec4 anded = subgroupAnd(ballot_value); |
| 73 |
uvec4 ored = subgroupOr(ballot_value); |
| 74 |
uvec4 xored = subgroupXor(ballot_value); |
| 75 |
|
| 76 |
added = subgroupInclusiveAdd(added); |
| 77 |
iadded = subgroupInclusiveAdd(iadded); |
| 78 |
multiplied = subgroupInclusiveMul(multiplied); |
| 79 |
imultiplied = subgroupInclusiveMul(imultiplied); |
| 80 |
#if 0 |
| 81 |
lo = subgroupInclusiveMin(lo); |
| 82 |
hi = subgroupInclusiveMax(hi); |
| 83 |
slo = subgroupInclusiveMin(slo); |
| 84 |
shi = subgroupInclusiveMax(shi); |
| 85 |
ulo = subgroupInclusiveMin(ulo); |
| 86 |
uhi = subgroupInclusiveMax(uhi); |
| 87 |
anded = subgroupInclusiveAnd(anded); |
| 88 |
ored = subgroupInclusiveOr(ored); |
| 89 |
xored = subgroupInclusiveXor(ored); |
| 90 |
added = subgroupExclusiveAdd(lo); |
| 91 |
#endif |
| 92 |
|
| 93 |
added = subgroupExclusiveAdd(multiplied); |
| 94 |
multiplied = subgroupExclusiveMul(multiplied); |
| 95 |
iadded = subgroupExclusiveAdd(imultiplied); |
| 96 |
imultiplied = subgroupExclusiveMul(imultiplied); |
| 97 |
#if 0 |
| 98 |
lo = subgroupExclusiveMin(lo); |
| 99 |
hi = subgroupExclusiveMax(hi); |
| 100 |
ulo = subgroupExclusiveMin(ulo); |
| 101 |
uhi = subgroupExclusiveMax(uhi); |
| 102 |
slo = subgroupExclusiveMin(slo); |
| 103 |
shi = subgroupExclusiveMax(shi); |
| 104 |
anded = subgroupExclusiveAnd(anded); |
| 105 |
ored = subgroupExclusiveOr(ored); |
| 106 |
xored = subgroupExclusiveXor(ored); |
| 107 |
#endif |
| 108 |
|
| 109 |
#if 0 |
| 110 |
// clustered |
| 111 |
added = subgroupClusteredAdd(added, 4u); |
| 112 |
multiplied = subgroupClusteredMul(multiplied, 4u); |
| 113 |
iadded = subgroupClusteredAdd(iadded, 4u); |
| 114 |
imultiplied = subgroupClusteredMul(imultiplied, 4u); |
| 115 |
lo = subgroupClusteredMin(lo, 4u); |
| 116 |
hi = subgroupClusteredMax(hi, 4u); |
| 117 |
ulo = subgroupClusteredMin(ulo, 4u); |
| 118 |
uhi = subgroupClusteredMax(uhi, 4u); |
| 119 |
slo = subgroupClusteredMin(slo, 4u); |
| 120 |
shi = subgroupClusteredMax(shi, 4u); |
| 121 |
anded = subgroupClusteredAnd(anded, 4u); |
| 122 |
ored = subgroupClusteredOr(ored, 4u); |
| 123 |
xored = subgroupClusteredXor(xored, 4u); |
| 124 |
#endif |
| 125 |
|
| 126 |
// quad |
| 127 |
vec4 swap_horiz = subgroupQuadSwapHorizontal(vec4(20.0)); |
| 128 |
vec4 swap_vertical = subgroupQuadSwapVertical(vec4(20.0)); |
| 129 |
vec4 swap_diagonal = subgroupQuadSwapDiagonal(vec4(20.0)); |
| 130 |
vec4 quad_broadcast = subgroupQuadBroadcast(vec4(20.0), 3u); |
| 131 |
} |