Add test shader for subgroup.

Update SPIRV-Tools/glslang commits.
Use vulkan1.1 environment for testing.
Found new "errors" in SPIRV-Tools, so disable validation on those shaders
for now.

checkout_glslang_spirv_tools.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
-GLSLANG_REV=7cec64fc42eba4587d2f3d73c03cb68ceb18dca4
-SPIRV_TOOLS_REV=fe9121f72144f1e1657cd21a55a1fec07c940e56
+GLSLANG_REV=461ea09943e0e88ea854ab9e3b42d17d728af2ad
+SPIRV_TOOLS_REV=53bc1623ecd3cc304d0d6feed8385e70c7ab30d3
 
 if [ -d external/glslang ]; then
 	echo "Updating glslang to revision $GLSLANG_REV."

reference/opt/shaders/vulkan/comp/subgroups.nocompat.invalid.vk.comp.vk

@@ -0,0 +1,110 @@
+#version 450
+#extension GL_KHR_shader_subgroup_basic : require
+#extension GL_KHR_shader_subgroup_ballot : require
+#extension GL_KHR_shader_subgroup_shuffle : require
+#extension GL_KHR_shader_subgroup_shuffle_relative : require
+#extension GL_KHR_shader_subgroup_vote : require
+#extension GL_KHR_shader_subgroup_arithmetic : require
+#extension GL_KHR_shader_subgroup_clustered : require
+#extension GL_KHR_shader_subgroup_quad : require
+layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
+
+layout(set = 0, binding = 0, std430) buffer SSBO
+{
+    float FragColor;
+} _9;
+
+void main()
+{
+    _9.FragColor = float(gl_NumSubgroups);
+    _9.FragColor = float(gl_SubgroupID);
+    _9.FragColor = float(gl_SubgroupSize);
+    _9.FragColor = float(gl_SubgroupInvocationID);
+    subgroupMemoryBarrier();
+    subgroupBarrier();
+    subgroupMemoryBarrier();
+    subgroupMemoryBarrierBuffer();
+    subgroupMemoryBarrierShared();
+    subgroupMemoryBarrierImage();
+    bool elected = subgroupElect();
+    _9.FragColor = vec4(gl_SubgroupEqMask).x;
+    _9.FragColor = vec4(gl_SubgroupGeMask).x;
+    _9.FragColor = vec4(gl_SubgroupGtMask).x;
+    _9.FragColor = vec4(gl_SubgroupLeMask).x;
+    _9.FragColor = vec4(gl_SubgroupLtMask).x;
+    vec4 broadcasted = subgroupBroadcast(vec4(10.0), 8u);
+    vec3 first = subgroupBroadcastFirst(vec3(20.0));
+    uvec4 ballot_value = subgroupBallot(true);
+    bool inverse_ballot_value = subgroupInverseBallot(ballot_value);
+    bool bit_extracted = subgroupBallotBitExtract(uvec4(10u), 8u);
+    uint bit_count = subgroupBallotBitCount(ballot_value);
+    uint inclusive_bit_count = subgroupBallotInclusiveBitCount(ballot_value);
+    uint exclusive_bit_count = subgroupBallotExclusiveBitCount(ballot_value);
+    uint lsb = subgroupBallotFindLSB(ballot_value);
+    uint msb = subgroupBallotFindMSB(ballot_value);
+    uint shuffled = subgroupShuffle(10u, 8u);
+    uint shuffled_xor = subgroupShuffleXor(30u, 8u);
+    uint shuffled_up = subgroupShuffleUp(20u, 4u);
+    uint shuffled_down = subgroupShuffleDown(20u, 4u);
+    bool has_all = subgroupAll(true);
+    bool has_any = subgroupAny(true);
+    bool has_equal = subgroupAllEqual(true);
+    vec4 added = subgroupAdd(vec4(20.0));
+    ivec4 iadded = subgroupAdd(ivec4(20));
+    vec4 multiplied = subgroupMul(vec4(20.0));
+    ivec4 imultiplied = subgroupMul(ivec4(20));
+    vec4 lo = subgroupMin(vec4(20.0));
+    vec4 hi = subgroupMax(vec4(20.0));
+    ivec4 slo = subgroupMin(ivec4(20));
+    ivec4 shi = subgroupMax(ivec4(20));
+    uvec4 ulo = subgroupMin(uvec4(20u));
+    uvec4 uhi = subgroupMax(uvec4(20u));
+    uvec4 anded = subgroupAnd(ballot_value);
+    uvec4 ored = subgroupOr(ballot_value);
+    uvec4 xored = subgroupXor(ballot_value);
+    added = subgroupInclusiveAdd(added);
+    iadded = subgroupInclusiveAdd(iadded);
+    multiplied = subgroupInclusiveMul(multiplied);
+    imultiplied = subgroupInclusiveMul(imultiplied);
+    lo = subgroupInclusiveMin(lo);
+    hi = subgroupInclusiveMax(hi);
+    slo = subgroupInclusiveMin(slo);
+    shi = subgroupInclusiveMax(shi);
+    ulo = subgroupInclusiveMin(ulo);
+    uhi = subgroupInclusiveMax(uhi);
+    anded = subgroupInclusiveAnd(anded);
+    ored = subgroupInclusiveOr(ored);
+    xored = subgroupInclusiveXor(ored);
+    added = subgroupExclusiveAdd(lo);
+    added = subgroupExclusiveAdd(multiplied);
+    multiplied = subgroupExclusiveMul(multiplied);
+    iadded = subgroupExclusiveAdd(imultiplied);
+    imultiplied = subgroupExclusiveMul(imultiplied);
+    lo = subgroupExclusiveMin(lo);
+    hi = subgroupExclusiveMax(hi);
+    ulo = subgroupExclusiveMin(ulo);
+    uhi = subgroupExclusiveMax(uhi);
+    slo = subgroupExclusiveMin(slo);
+    shi = subgroupExclusiveMax(shi);
+    anded = subgroupExclusiveAnd(anded);
+    ored = subgroupExclusiveOr(ored);
+    xored = subgroupExclusiveXor(ored);
+    added = subgroupClusteredAdd(added, 4u);
+    multiplied = subgroupClusteredMul(multiplied, 4u);
+    iadded = subgroupClusteredAdd(iadded, 4u);
+    imultiplied = subgroupClusteredMul(imultiplied, 4u);
+    lo = subgroupClusteredMin(lo, 4u);
+    hi = subgroupClusteredMax(hi, 4u);
+    ulo = subgroupClusteredMin(ulo, 4u);
+    uhi = subgroupClusteredMax(uhi, 4u);
+    slo = subgroupClusteredMin(slo, 4u);
+    shi = subgroupClusteredMax(shi, 4u);
+    anded = subgroupClusteredAnd(anded, 4u);
+    ored = subgroupClusteredOr(ored, 4u);
+    xored = subgroupClusteredXor(xored, 4u);
+    vec4 swap_horiz = subgroupQuadSwapHorizontal(vec4(20.0));
+    vec4 swap_vertical = subgroupQuadSwapVertical(vec4(20.0));
+    vec4 swap_diagonal = subgroupQuadSwapDiagonal(vec4(20.0));
+    vec4 quad_broadcast = subgroupQuadBroadcast(vec4(20.0), 3u);
+}
+

reference/shaders/vulkan/comp/subgroups.nocompat.invalid.vk.comp.vk

@@ -0,0 +1,110 @@
+#version 450
+#extension GL_KHR_shader_subgroup_basic : require
+#extension GL_KHR_shader_subgroup_ballot : require
+#extension GL_KHR_shader_subgroup_shuffle : require
+#extension GL_KHR_shader_subgroup_shuffle_relative : require
+#extension GL_KHR_shader_subgroup_vote : require
+#extension GL_KHR_shader_subgroup_arithmetic : require
+#extension GL_KHR_shader_subgroup_clustered : require
+#extension GL_KHR_shader_subgroup_quad : require
+layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
+
+layout(set = 0, binding = 0, std430) buffer SSBO
+{
+    float FragColor;
+} _9;
+
+void main()
+{
+    _9.FragColor = float(gl_NumSubgroups);
+    _9.FragColor = float(gl_SubgroupID);
+    _9.FragColor = float(gl_SubgroupSize);
+    _9.FragColor = float(gl_SubgroupInvocationID);
+    subgroupMemoryBarrier();
+    subgroupBarrier();
+    subgroupMemoryBarrier();
+    subgroupMemoryBarrierBuffer();
+    subgroupMemoryBarrierShared();
+    subgroupMemoryBarrierImage();
+    bool elected = subgroupElect();
+    _9.FragColor = vec4(gl_SubgroupEqMask).x;
+    _9.FragColor = vec4(gl_SubgroupGeMask).x;
+    _9.FragColor = vec4(gl_SubgroupGtMask).x;
+    _9.FragColor = vec4(gl_SubgroupLeMask).x;
+    _9.FragColor = vec4(gl_SubgroupLtMask).x;
+    vec4 broadcasted = subgroupBroadcast(vec4(10.0), 8u);
+    vec3 first = subgroupBroadcastFirst(vec3(20.0));
+    uvec4 ballot_value = subgroupBallot(true);
+    bool inverse_ballot_value = subgroupInverseBallot(ballot_value);
+    bool bit_extracted = subgroupBallotBitExtract(uvec4(10u), 8u);
+    uint bit_count = subgroupBallotBitCount(ballot_value);
+    uint inclusive_bit_count = subgroupBallotInclusiveBitCount(ballot_value);
+    uint exclusive_bit_count = subgroupBallotExclusiveBitCount(ballot_value);
+    uint lsb = subgroupBallotFindLSB(ballot_value);
+    uint msb = subgroupBallotFindMSB(ballot_value);
+    uint shuffled = subgroupShuffle(10u, 8u);
+    uint shuffled_xor = subgroupShuffleXor(30u, 8u);
+    uint shuffled_up = subgroupShuffleUp(20u, 4u);
+    uint shuffled_down = subgroupShuffleDown(20u, 4u);
+    bool has_all = subgroupAll(true);
+    bool has_any = subgroupAny(true);
+    bool has_equal = subgroupAllEqual(true);
+    vec4 added = subgroupAdd(vec4(20.0));
+    ivec4 iadded = subgroupAdd(ivec4(20));
+    vec4 multiplied = subgroupMul(vec4(20.0));
+    ivec4 imultiplied = subgroupMul(ivec4(20));
+    vec4 lo = subgroupMin(vec4(20.0));
+    vec4 hi = subgroupMax(vec4(20.0));
+    ivec4 slo = subgroupMin(ivec4(20));
+    ivec4 shi = subgroupMax(ivec4(20));
+    uvec4 ulo = subgroupMin(uvec4(20u));
+    uvec4 uhi = subgroupMax(uvec4(20u));
+    uvec4 anded = subgroupAnd(ballot_value);
+    uvec4 ored = subgroupOr(ballot_value);
+    uvec4 xored = subgroupXor(ballot_value);
+    added = subgroupInclusiveAdd(added);
+    iadded = subgroupInclusiveAdd(iadded);
+    multiplied = subgroupInclusiveMul(multiplied);
+    imultiplied = subgroupInclusiveMul(imultiplied);
+    lo = subgroupInclusiveMin(lo);
+    hi = subgroupInclusiveMax(hi);
+    slo = subgroupInclusiveMin(slo);
+    shi = subgroupInclusiveMax(shi);
+    ulo = subgroupInclusiveMin(ulo);
+    uhi = subgroupInclusiveMax(uhi);
+    anded = subgroupInclusiveAnd(anded);
+    ored = subgroupInclusiveOr(ored);
+    xored = subgroupInclusiveXor(ored);
+    added = subgroupExclusiveAdd(lo);
+    added = subgroupExclusiveAdd(multiplied);
+    multiplied = subgroupExclusiveMul(multiplied);
+    iadded = subgroupExclusiveAdd(imultiplied);
+    imultiplied = subgroupExclusiveMul(imultiplied);
+    lo = subgroupExclusiveMin(lo);
+    hi = subgroupExclusiveMax(hi);
+    ulo = subgroupExclusiveMin(ulo);
+    uhi = subgroupExclusiveMax(uhi);
+    slo = subgroupExclusiveMin(slo);
+    shi = subgroupExclusiveMax(shi);
+    anded = subgroupExclusiveAnd(anded);
+    ored = subgroupExclusiveOr(ored);
+    xored = subgroupExclusiveXor(ored);
+    added = subgroupClusteredAdd(added, 4u);
+    multiplied = subgroupClusteredMul(multiplied, 4u);
+    iadded = subgroupClusteredAdd(iadded, 4u);
+    imultiplied = subgroupClusteredMul(imultiplied, 4u);
+    lo = subgroupClusteredMin(lo, 4u);
+    hi = subgroupClusteredMax(hi, 4u);
+    ulo = subgroupClusteredMin(ulo, 4u);
+    uhi = subgroupClusteredMax(uhi, 4u);
+    slo = subgroupClusteredMin(slo, 4u);
+    shi = subgroupClusteredMax(shi, 4u);
+    anded = subgroupClusteredAnd(anded, 4u);
+    ored = subgroupClusteredOr(ored, 4u);
+    xored = subgroupClusteredXor(xored, 4u);
+    vec4 swap_horiz = subgroupQuadSwapHorizontal(vec4(20.0));
+    vec4 swap_vertical = subgroupQuadSwapVertical(vec4(20.0));
+    vec4 swap_diagonal = subgroupQuadSwapDiagonal(vec4(20.0));
+    vec4 quad_broadcast = subgroupQuadBroadcast(vec4(20.0), 3u);
+}
+

test_shaders.py

@@ -97,7 +97,7 @@ def cross_compile_msl(shader, spirv, opt):
     if spirv:
         subprocess.check_call(['spirv-as', '-o', spirv_path, shader])
     else:
-        subprocess.check_call(['glslangValidator', '-V', '-o', spirv_path, shader])
+        subprocess.check_call(['glslangValidator', '--target-env', 'vulkan1.1', '-V', '-o', spirv_path, shader])
 
     if opt:
         subprocess.check_call(['spirv-opt', '-O', '-o', spirv_path, spirv_path])
@@ -110,7 +110,10 @@ def cross_compile_msl(shader, spirv, opt):
         msl_args.append('20000')
 
     subprocess.check_call(msl_args)
-    subprocess.check_call(['spirv-val', spirv_path])
+
+    if not shader_is_invalid_spirv(msl_path):
+        subprocess.check_call(['spirv-val', '--target-env', 'vulkan1.1', spirv_path])
+
     return (spirv_path, msl_path)
 
 def shader_model_hlsl(shader):
@@ -138,7 +141,7 @@ def shader_to_win_path(shader):
     return shader
 
 def validate_shader_hlsl(shader):
-    subprocess.check_call(['glslangValidator', '-e', 'main', '-D', '-V', shader])
+    subprocess.check_call(['glslangValidator', '-e', 'main', '-D', '--target-env', 'vulkan1.1', '-V', shader])
     is_no_fxc = '.nofxc.' in shader
     if (not force_no_external_validation) and (not is_no_fxc):
         try:
@@ -168,7 +171,7 @@ def cross_compile_hlsl(shader, spirv, opt):
     if spirv:
         subprocess.check_call(['spirv-as', '-o', spirv_path, shader])
     else:
-        subprocess.check_call(['glslangValidator', '-V', '-o', spirv_path, shader])
+        subprocess.check_call(['glslangValidator', '--target-env', 'vulkan1.1', '-V', '-o', spirv_path, shader])
 
     if opt:
         subprocess.check_call(['spirv-opt', '-O', '-o', spirv_path, spirv_path])
@@ -177,7 +180,9 @@ def cross_compile_hlsl(shader, spirv, opt):
 
     sm = shader_to_sm(shader)
     subprocess.check_call([spirv_cross_path, '--entry', 'main', '--output', hlsl_path, spirv_path, '--hlsl-enable-compat', '--hlsl', '--shader-model', sm])
-    subprocess.check_call(['spirv-val', spirv_path])
+
+    if not shader_is_invalid_spirv(hlsl_path):
+        subprocess.check_call(['spirv-val', '--target-env', 'vulkan1.1', spirv_path])
 
     validate_shader_hlsl(hlsl_path)
     
@@ -185,7 +190,7 @@ def cross_compile_hlsl(shader, spirv, opt):
 
 def validate_shader(shader, vulkan):
     if vulkan:
-        subprocess.check_call(['glslangValidator', '-V', shader])
+        subprocess.check_call(['glslangValidator', '--target-env', 'vulkan1.1', '-V', shader])
     else:
         subprocess.check_call(['glslangValidator', shader])
 
@@ -202,13 +207,13 @@ def cross_compile(shader, vulkan, spirv, invalid_spirv, eliminate, is_legacy, fl
     if spirv:
         subprocess.check_call(['spirv-as', '-o', spirv_path, shader])
     else:
-        subprocess.check_call(['glslangValidator', '-V', '-o', spirv_path, shader])
+        subprocess.check_call(['glslangValidator', '--target-env', 'vulkan1.1', '-V', '-o', spirv_path, shader])
 
     if opt and (not invalid_spirv):
         subprocess.check_call(['spirv-opt', '-O', '-o', spirv_path, spirv_path])
 
     if not invalid_spirv:
-        subprocess.check_call(['spirv-val', spirv_path])
+        subprocess.check_call(['spirv-val', '--target-env', 'vulkan1.1', spirv_path])
 
     extra_args = []
     if eliminate:
@@ -223,11 +228,14 @@ def cross_compile(shader, vulkan, spirv, invalid_spirv, eliminate, is_legacy, fl
         extra_args += ['--flatten-multidimensional-arrays']
 
     spirv_cross_path = './spirv-cross'
-    subprocess.check_call([spirv_cross_path, '--entry', 'main', '--output', glsl_path, spirv_path] + extra_args)
 
     # A shader might not be possible to make valid GLSL from, skip validation for this case.
     if (not ('nocompat' in glsl_path)) and (not spirv):
+        subprocess.check_call([spirv_cross_path, '--entry', 'main', '--output', glsl_path, spirv_path] + extra_args)
         validate_shader(glsl_path, False)
+    else:
+        os.remove(glsl_path)
+        glsl_path = None
 
     if vulkan or spirv:
         subprocess.check_call([spirv_cross_path, '--entry', 'main', '--vulkan-semantics', '--output', vulkan_glsl_path, spirv_path] + extra_args)
@@ -343,7 +351,8 @@ def test_shader(stats, shader, update, keep, opt):
     if stats and (not vulkan) and (not is_spirv) and (not desktop):
         cross_stats = get_shader_stats(glsl)
 
-    regression_check(shader, glsl, update, keep, opt)
+    if glsl:
+        regression_check(shader, glsl, update, keep, opt)
     if vulkan_glsl:
         regression_check((shader[0], shader[1] + '.vk'), vulkan_glsl, update, keep, opt)
     os.remove(spirv)