| /*------------------------------------------------------------------------ |
| * Vulkan Conformance Tests |
| * ------------------------ |
| * |
| * Copyright (c) 2019 The Khronos Group Inc. |
| * Copyright (c) 2019 Valve Corporation. |
| * |
| * Licensed under the Apache License, Version 2.0 (the "License"); |
| * you may not use this file except in compliance with the License. |
| * You may obtain a copy of the License at |
| * |
| * http://www.apache.org/licenses/LICENSE-2.0 |
| * |
| * Unless required by applicable law or agreed to in writing, software |
| * distributed under the License is distributed on an "AS IS" BASIS, |
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| * See the License for the specific language governing permissions and |
| * limitations under the License. |
| * |
| */ /*! |
| * \file |
| * \brief VK_EXT_subgroup_size_control Tests |
| */ /*--------------------------------------------------------------------*/ |
| |
| #include "vktSubgroupsSizeControlTests.hpp" |
| #include "vktSubgroupsTestsUtils.hpp" |
| #include "vktTestCaseUtil.hpp" |
| #include "tcuTestLog.hpp" |
| |
| #include <string> |
| #include <vector> |
| #include <algorithm> |
| |
| using namespace tcu; |
| using namespace std; |
| using namespace vk; |
| using namespace vkt; |
| |
| namespace |
| { |
| |
| enum RequiredSubgroupSizeMode |
| { |
| REQUIRED_SUBGROUP_SIZE_NONE = 0, |
| REQUIRED_SUBGROUP_SIZE_MIN = 1, |
| REQUIRED_SUBGROUP_SIZE_MAX = 2, |
| }; |
| |
| struct CaseDefinition |
| { |
| uint32_t pipelineShaderStageCreateFlags; |
| VkShaderStageFlags shaderStage; |
| bool requiresBallot; |
| uint32_t requiredSubgroupSizeMode; |
| de::SharedPtr<bool> geometryPointSizeSupported; |
| SpirvVersion spirvVersion; |
| |
| bool hasFullSubgroupsFlag(void) const |
| { |
| return ((pipelineShaderStageCreateFlags & VK_PIPELINE_SHADER_STAGE_CREATE_REQUIRE_FULL_SUBGROUPS_BIT_EXT) != |
| 0u); |
| } |
| |
| bool shaderUsesFullSubgroups(void) const |
| { |
| return (hasFullSubgroupsFlag() || (spirvVersion >= SPIRV_VERSION_1_6)); |
| } |
| }; |
| |
| struct TestParams |
| { |
| bool useSpirv16; |
| bool flagsEnabled; |
| string postfix; |
| }; |
| |
| struct internalDataStruct |
| { |
| const Context *context; |
| struct CaseDefinition caseDef; |
| const uint32_t requiredSubgroupSize; |
| const bool |
| isRequiredSubgroupSize; // Indicates if the test uses VkPipelineShaderStageRequiredSubgroupSizeCreateInfo. |
| }; |
| |
| inline bool makeDeBool(bool value) |
| { |
| return (value ? true : false); |
| } |
| |
| UVec3 getLocalSizes(const uint32_t maxWorkGroupSize[3], uint32_t maxWorkGroupInvocations, |
| uint32_t numWorkGroupInvocations) |
| { |
| DE_ASSERT(numWorkGroupInvocations <= maxWorkGroupInvocations); |
| DE_UNREF(maxWorkGroupInvocations); // For release builds. |
| |
| const uint32_t localSizeX = de::gcd(numWorkGroupInvocations, maxWorkGroupSize[0]); |
| const uint32_t localSizeY = de::gcd(std::max(numWorkGroupInvocations / localSizeX, 1u), maxWorkGroupSize[1]); |
| const uint32_t localSizeZ = std::max(numWorkGroupInvocations / (localSizeX * localSizeY), 1u); |
| |
| return UVec3(localSizeX, localSizeY, localSizeZ); |
| } |
| |
| uint32_t getRequiredSubgroupSizeFromMode( |
| Context &context, const CaseDefinition &caseDef, |
| #ifndef CTS_USES_VULKANSC |
| const VkPhysicalDeviceSubgroupSizeControlProperties &subgroupSizeControlProperties) |
| #else |
| const VkPhysicalDeviceSubgroupSizeControlPropertiesEXT &subgroupSizeControlProperties) |
| #endif // CTS_USES_VULKANSC |
| { |
| switch (caseDef.requiredSubgroupSizeMode) |
| { |
| case REQUIRED_SUBGROUP_SIZE_MAX: |
| return subgroupSizeControlProperties.maxSubgroupSize; |
| case REQUIRED_SUBGROUP_SIZE_MIN: |
| return subgroupSizeControlProperties.minSubgroupSize; |
| case REQUIRED_SUBGROUP_SIZE_NONE: |
| return subgroups::getSubgroupSize(context); |
| default: |
| TCU_THROW(NotSupportedError, "Unsupported Subgroup size"); |
| } |
| } |
| |
| static bool checkVertexPipelineStages(const void *internalData, vector<const void *> datas, uint32_t width, uint32_t) |
| { |
| const struct internalDataStruct *checkInternalData = |
| reinterpret_cast<const struct internalDataStruct *>(internalData); |
| const Context *context = checkInternalData->context; |
| const auto &subgroupSizeControlProperties = context->getSubgroupSizeControlProperties(); |
| |
| TestLog &log = context->getTestContext().getLog(); |
| const uint32_t *data = reinterpret_cast<const uint32_t *>(datas[0]); |
| |
| for (uint32_t i = 0; i < width; i++) |
| { |
| if (data[i] > subgroupSizeControlProperties.maxSubgroupSize || |
| data[i] < subgroupSizeControlProperties.minSubgroupSize) |
| { |
| log << TestLog::Message << "gl_SubgroupSize (" << data[i] << ") value is outside limits (" |
| << subgroupSizeControlProperties.minSubgroupSize << ", " |
| << subgroupSizeControlProperties.maxSubgroupSize << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| |
| if (checkInternalData->isRequiredSubgroupSize && data[i] != checkInternalData->requiredSubgroupSize) |
| { |
| log << TestLog::Message << "gl_SubgroupSize (" << data[i] |
| << ") is not equal to the required subgroup size value (" << checkInternalData->requiredSubgroupSize |
| << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| } |
| |
| return true; |
| } |
| |
| static bool checkFragmentPipelineStages(const void *internalData, vector<const void *> datas, uint32_t width, |
| uint32_t height, uint32_t) |
| { |
| const struct internalDataStruct *checkInternalData = |
| reinterpret_cast<const struct internalDataStruct *>(internalData); |
| const Context *context = checkInternalData->context; |
| const auto &subgroupSizeControlProperties = context->getSubgroupSizeControlProperties(); |
| TestLog &log = context->getTestContext().getLog(); |
| const uint32_t *data = reinterpret_cast<const uint32_t *>(datas[0]); |
| |
| for (uint32_t x = 0u; x < width; ++x) |
| { |
| for (uint32_t y = 0u; y < height; ++y) |
| { |
| const uint32_t ndx = (x * height + y); |
| |
| if (data[ndx] > subgroupSizeControlProperties.maxSubgroupSize || |
| data[ndx] < subgroupSizeControlProperties.minSubgroupSize) |
| { |
| log << TestLog::Message << "gl_SubgroupSize (" << data[ndx] << ") value is outside limits (" |
| << subgroupSizeControlProperties.minSubgroupSize << ", " |
| << subgroupSizeControlProperties.maxSubgroupSize << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| |
| if (checkInternalData->isRequiredSubgroupSize && data[ndx] != checkInternalData->requiredSubgroupSize) |
| { |
| log << TestLog::Message << "gl_SubgroupSize (" << data[ndx] |
| << ") is not equal to the required subgroup size value (" << checkInternalData->requiredSubgroupSize |
| << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| } |
| } |
| return true; |
| } |
| |
| static bool checkCompute(const void *internalData, vector<const void *> datas, const uint32_t numWorkgroups[3], |
| const uint32_t localSize[3], uint32_t) |
| { |
| const struct internalDataStruct *checkInternalData = |
| reinterpret_cast<const struct internalDataStruct *>(internalData); |
| const Context *context = checkInternalData->context; |
| const auto &subgroupSizeControlProperties = context->getSubgroupSizeControlProperties(); |
| TestLog &log = context->getTestContext().getLog(); |
| const uint32_t globalSizeX = numWorkgroups[0] * localSize[0]; |
| const uint32_t globalSizeY = numWorkgroups[1] * localSize[1]; |
| const uint32_t globalSizeZ = numWorkgroups[2] * localSize[2]; |
| const uint32_t width = globalSizeX * globalSizeY * globalSizeZ; |
| const uint32_t *data = reinterpret_cast<const uint32_t *>(datas[0]); |
| |
| for (uint32_t i = 0; i < width; i++) |
| { |
| if (data[i] > subgroupSizeControlProperties.maxSubgroupSize || |
| data[i] < subgroupSizeControlProperties.minSubgroupSize) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "gl_SubgroupSize (" << data[i] << ") value is outside limits (" |
| << subgroupSizeControlProperties.minSubgroupSize << ", " |
| << subgroupSizeControlProperties.maxSubgroupSize << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| |
| if (checkInternalData->isRequiredSubgroupSize && data[i] != checkInternalData->requiredSubgroupSize) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "gl_SubgroupSize (" << data[i] << ") is not equal to the required subgroup size value (" |
| << checkInternalData->requiredSubgroupSize << ")" << TestLog::EndMessage; |
| |
| return false; |
| } |
| } |
| |
| return true; |
| } |
| |
| static bool checkComputeRequireFull(const void *internalData, vector<const void *> datas, |
| const uint32_t numWorkgroups[3], const uint32_t localSize[3], uint32_t) |
| { |
| const struct internalDataStruct *checkInternalData = |
| reinterpret_cast<const struct internalDataStruct *>(internalData); |
| const Context *context = checkInternalData->context; |
| const auto &subgroupSizeControlProperties = context->getSubgroupSizeControlProperties(); |
| TestLog &log = context->getTestContext().getLog(); |
| const uint32_t globalSizeX = numWorkgroups[0] * localSize[0]; |
| const uint32_t globalSizeY = numWorkgroups[1] * localSize[1]; |
| const uint32_t globalSizeZ = numWorkgroups[2] * localSize[2]; |
| const uint32_t width = globalSizeX * globalSizeY * globalSizeZ; |
| const UVec4 *data = reinterpret_cast<const UVec4 *>(datas[0]); |
| const uint32_t numSubgroups = |
| (localSize[0] * localSize[1] * localSize[2]) / checkInternalData->requiredSubgroupSize; |
| const bool exactSubgroupSize = |
| (checkInternalData->caseDef.shaderUsesFullSubgroups() && checkInternalData->isRequiredSubgroupSize); |
| |
| for (uint32_t i = 0; i < width; i++) |
| { |
| if (data[i].x() > subgroupSizeControlProperties.maxSubgroupSize || |
| data[i].x() < subgroupSizeControlProperties.minSubgroupSize) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "gl_SubgroupSize value ( " << data[i].x() << ") is outside limits [" |
| << subgroupSizeControlProperties.minSubgroupSize << ", " |
| << subgroupSizeControlProperties.maxSubgroupSize << "]" << TestLog::EndMessage; |
| return false; |
| } |
| |
| if (data[i].x() != data[i].y()) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "gl_SubgroupSize ( " << data[i].x() << ") does not match the active number of subgroup invocations (" |
| << data[i].y() << ")" << TestLog::EndMessage; |
| return false; |
| } |
| |
| if (exactSubgroupSize && data[i].x() != checkInternalData->requiredSubgroupSize) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "expected subgroupSize (" << checkInternalData->requiredSubgroupSize |
| << ") doesn't match gl_SubgroupSize ( " << data[i].x() << ")" << TestLog::EndMessage; |
| return false; |
| } |
| |
| if (exactSubgroupSize && data[i].z() != numSubgroups) |
| { |
| log << TestLog::Message << "[" << localSize[0] << ", " << localSize[1] << ", " << localSize[2] << "] " |
| << "expected number of subgroups dispatched (" << numSubgroups << ") doesn't match gl_NumSubgroups (" |
| << data[i].z() << ")" << TestLog::EndMessage; |
| return false; |
| } |
| } |
| |
| return true; |
| } |
| |
| void initFrameBufferPrograms(SourceCollections &programCollection, CaseDefinition caseDef) |
| { |
| const ShaderBuildOptions buildOptions(programCollection.usedVulkanVersion, caseDef.spirvVersion, 0u); |
| |
| if (VK_SHADER_STAGE_FRAGMENT_BIT != caseDef.shaderStage) |
| subgroups::setFragmentShaderFrameBuffer(programCollection); |
| |
| if (VK_SHADER_STAGE_VERTEX_BIT != caseDef.shaderStage && VK_SHADER_STAGE_FRAGMENT_BIT != caseDef.shaderStage) |
| subgroups::setVertexShaderFrameBuffer(programCollection); |
| |
| string bdyStr = "uint tempResult = gl_SubgroupSize;\n"; |
| |
| if (VK_SHADER_STAGE_VERTEX_BIT == caseDef.shaderStage) |
| { |
| ostringstream vertex; |
| |
| vertex << glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450) << "\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "layout(location = 0) in highp vec4 in_position;\n" |
| << "layout(location = 0) out float out_color;\n" |
| << "\n" |
| << "void main (void)\n" |
| << "{\n" |
| << bdyStr << " out_color = float(tempResult);\n" |
| << " gl_Position = in_position;\n" |
| << " gl_PointSize = 1.0f;\n" |
| << "}\n"; |
| |
| programCollection.glslSources.add("vert") << glu::VertexSource(vertex.str()) << buildOptions; |
| } |
| else if (VK_SHADER_STAGE_GEOMETRY_BIT == caseDef.shaderStage) |
| { |
| ostringstream geometry; |
| |
| geometry << glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450) << "\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "layout(points) in;\n" |
| << "layout(points, max_vertices = 1) out;\n" |
| << "layout(location = 0) out float out_color;\n" |
| << "void main (void)\n" |
| << "{\n" |
| << bdyStr << " out_color = float(tempResult);\n" |
| << " gl_Position = gl_in[0].gl_Position;\n" |
| << " gl_PointSize = 1.0f;" |
| << " EmitVertex();\n" |
| << " EndPrimitive();\n" |
| << "}\n"; |
| |
| programCollection.glslSources.add("geometry") << glu::GeometrySource(geometry.str()) << buildOptions; |
| } |
| else if (VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT == caseDef.shaderStage) |
| { |
| ostringstream controlSource; |
| |
| controlSource << glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450) << "\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "layout(vertices = 2) out;\n" |
| << "layout(location = 0) out float out_color[];\n" |
| << "\n" |
| << "void main (void)\n" |
| << "{\n" |
| << " if (gl_InvocationID == 0)\n" |
| << " {\n" |
| << " gl_TessLevelOuter[0] = 1.0f;\n" |
| << " gl_TessLevelOuter[1] = 1.0f;\n" |
| << " }\n" |
| << bdyStr << " out_color[gl_InvocationID ] = float(tempResult);\n" |
| << " gl_out[gl_InvocationID].gl_Position = gl_in[gl_InvocationID].gl_Position;\n" |
| << "}\n"; |
| |
| programCollection.glslSources.add("tesc") |
| << glu::TessellationControlSource(controlSource.str()) << buildOptions; |
| subgroups::setTesEvalShaderFrameBuffer(programCollection); |
| } |
| else if (VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT == caseDef.shaderStage) |
| { |
| ostringstream evaluationSource; |
| evaluationSource << glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450) << "\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "layout(isolines, equal_spacing, ccw ) in;\n" |
| << "layout(location = 0) out float out_color;\n" |
| << "void main (void)\n" |
| << "{\n" |
| << bdyStr << " out_color = float(tempResult);\n" |
| << " gl_Position = mix(gl_in[0].gl_Position, gl_in[1].gl_Position, gl_TessCoord.x);\n" |
| << "}\n"; |
| |
| subgroups::setTesCtrlShaderFrameBuffer(programCollection); |
| programCollection.glslSources.add("tese") |
| << glu::TessellationEvaluationSource(evaluationSource.str()) << buildOptions; |
| } |
| else if (VK_SHADER_STAGE_FRAGMENT_BIT == caseDef.shaderStage) |
| { |
| const string vertex = string(glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450)) + |
| "\n" |
| "void main (void)\n" |
| "{\n" |
| " vec2 uv = vec2(float(gl_VertexIndex & 1), float((gl_VertexIndex >> 1) & 1));\n" |
| " gl_Position = vec4(uv * 4.0f -2.0f, 0.0f, 1.0f);\n" |
| " gl_PointSize = 1.0f;\n" |
| "}\n"; |
| programCollection.glslSources.add("vert") << glu::VertexSource(vertex) << buildOptions; |
| |
| ostringstream fragmentSource; |
| |
| fragmentSource << glu::getGLSLVersionDeclaration(glu::GLSL_VERSION_450) << "\n" |
| << "precision highp int;\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "layout(location = 0) out uint out_color;\n" |
| << "void main()\n" |
| << "{\n" |
| << bdyStr << " out_color = tempResult;\n" |
| << "}\n"; |
| |
| programCollection.glslSources.add("fragment") << glu::FragmentSource(fragmentSource.str()) << buildOptions; |
| } |
| else |
| { |
| DE_FATAL("Unsupported shader stage"); |
| } |
| } |
| |
| string getExtHeader(const CaseDefinition &) |
| { |
| return "#extension GL_KHR_shader_subgroup_basic: enable\n"; |
| } |
| |
| vector<string> getPerStageHeadDeclarations(const CaseDefinition &caseDef) |
| { |
| const uint32_t stageCount = subgroups::getStagesCount(caseDef.shaderStage); |
| const bool fragment = (caseDef.shaderStage & VK_SHADER_STAGE_FRAGMENT_BIT) != 0; |
| vector<string> result(stageCount, string()); |
| |
| if (fragment) |
| result.reserve(result.size() + 1); |
| |
| for (size_t i = 0; i < result.size(); ++i) |
| { |
| result[i] = "layout(set = 0, binding = " + de::toString(i) + |
| ", std430) buffer Buffer1\n" |
| "{\n" |
| " uint result[];\n" |
| "};\n"; |
| } |
| |
| if (fragment) |
| { |
| const string fragPart = "layout(location = 0) out uint result;\n"; |
| |
| result.push_back(fragPart); |
| } |
| |
| return result; |
| } |
| |
| string getTestSource(const CaseDefinition &) |
| { |
| return " uint tempResult = gl_SubgroupSize;\n" |
| " tempRes = tempResult;\n"; |
| } |
| |
| void initPrograms(SourceCollections &programCollection, CaseDefinition caseDef) |
| { |
| ShaderBuildOptions buildOptions(programCollection.usedVulkanVersion, caseDef.spirvVersion, 0u, |
| (caseDef.spirvVersion == vk::SPIRV_VERSION_1_4)); |
| const string extHeader = getExtHeader(caseDef); |
| const string testSrc = getTestSource(caseDef); |
| const vector<string> headDeclarations = getPerStageHeadDeclarations(caseDef); |
| |
| subgroups::initStdPrograms(programCollection, buildOptions, caseDef.shaderStage, VK_FORMAT_R32_UINT, |
| *caseDef.geometryPointSizeSupported, extHeader, testSrc, "", headDeclarations); |
| } |
| |
| void initProgramsRequireFull(SourceCollections &programCollection, CaseDefinition caseDef) |
| { |
| if (VK_SHADER_STAGE_COMPUTE_BIT != caseDef.shaderStage) |
| DE_FATAL("Unsupported shader stage"); |
| |
| ostringstream src; |
| |
| src << "#version 450\n" |
| << "#extension GL_KHR_shader_subgroup_basic: enable\n" |
| << "#extension GL_KHR_shader_subgroup_ballot: enable\n" |
| << "layout (local_size_x_id = 0, local_size_y_id = 1, " |
| "local_size_z_id = 2) in;\n" |
| << "layout(set = 0, binding = 0, std430) buffer Buffer1\n" |
| << "{\n" |
| << " uvec4 result[];\n" |
| << "};\n" |
| << "\n" |
| << "void main (void)\n" |
| << "{\n" |
| << " uvec3 globalSize = gl_NumWorkGroups * gl_WorkGroupSize;\n" |
| << " highp uint offset = globalSize.x * ((globalSize.y * " |
| "gl_GlobalInvocationID.z) + gl_GlobalInvocationID.y) + " |
| "gl_GlobalInvocationID.x;\n" |
| << " result[offset].x = gl_SubgroupSize;\n" // save the subgroup size value |
| << " uint numActive = subgroupBallotBitCount(subgroupBallot(true));\n" |
| << " result[offset].y = numActive;\n" // save the number of active subgroup invocations |
| << " result[offset].z = gl_NumSubgroups;\n" // save the number of subgroups dispatched. |
| << "}\n"; |
| |
| programCollection.glslSources.add("comp") |
| << glu::ComputeSource(src.str()) |
| << ShaderBuildOptions(programCollection.usedVulkanVersion, caseDef.spirvVersion, 0u); |
| } |
| |
| void supportedCheck(Context &context) |
| { |
| if (!subgroups::isSubgroupSupported(context)) |
| TCU_THROW(NotSupportedError, "Subgroup operations are not supported"); |
| |
| context.requireDeviceFunctionality("VK_EXT_subgroup_size_control"); |
| } |
| |
| void supportedCheckFeatures(Context &context, CaseDefinition caseDef) |
| { |
| supportedCheck(context); |
| |
| if (!subgroups::areSubgroupOperationsSupportedForStage(context, caseDef.shaderStage)) |
| { |
| TCU_THROW(NotSupportedError, "Shader stage is required to support subgroup operations!"); |
| } |
| |
| if (caseDef.shaderStage == VK_SHADER_STAGE_ALL_GRAPHICS) |
| { |
| const VkPhysicalDeviceFeatures &features = context.getDeviceFeatures(); |
| |
| if (!features.tessellationShader || !features.geometryShader) |
| TCU_THROW(NotSupportedError, "Device does not support tessellation or geometry shaders"); |
| } |
| |
| if (caseDef.requiresBallot && |
| !subgroups::isSubgroupFeatureSupportedForDevice(context, VK_SUBGROUP_FEATURE_BALLOT_BIT)) |
| { |
| TCU_THROW(NotSupportedError, "Device does not support subgroup ballot operations"); |
| } |
| |
| if (caseDef.requiredSubgroupSizeMode != REQUIRED_SUBGROUP_SIZE_NONE || |
| caseDef.pipelineShaderStageCreateFlags == VK_PIPELINE_SHADER_STAGE_CREATE_ALLOW_VARYING_SUBGROUP_SIZE_BIT_EXT) |
| { |
| if (context.getSubgroupSizeControlFeatures().subgroupSizeControl == false) |
| TCU_THROW(NotSupportedError, "Device does not support varying subgroup sizes nor required subgroup size"); |
| |
| if (caseDef.requiredSubgroupSizeMode != REQUIRED_SUBGROUP_SIZE_NONE) |
| { |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| if ((subgroupSizeControlProperties.requiredSubgroupSizeStages & caseDef.shaderStage) != caseDef.shaderStage) |
| TCU_THROW(NotSupportedError, |
| "Device does not support setting required subgroup size for the stages selected"); |
| } |
| } |
| |
| if (caseDef.hasFullSubgroupsFlag()) |
| { |
| if (context.getSubgroupSizeControlFeatures().computeFullSubgroups == false) |
| TCU_THROW(NotSupportedError, "Device does not support full subgroups in compute shaders"); |
| } |
| |
| *caseDef.geometryPointSizeSupported = subgroups::isTessellationAndGeometryPointSizeSupported(context); |
| |
| #ifndef CTS_USES_VULKANSC |
| if (isAllRayTracingStages(caseDef.shaderStage)) |
| { |
| context.requireDeviceFunctionality("VK_KHR_ray_tracing_pipeline"); |
| } |
| else if (isAllMeshShadingStages(caseDef.shaderStage)) |
| { |
| context.requireDeviceCoreFeature(DEVICE_CORE_FEATURE_VERTEX_PIPELINE_STORES_AND_ATOMICS); |
| context.requireDeviceFunctionality("VK_EXT_mesh_shader"); |
| |
| if ((caseDef.shaderStage & VK_SHADER_STAGE_TASK_BIT_EXT) != 0u) |
| { |
| const auto &features = context.getMeshShaderFeaturesEXT(); |
| if (!features.taskShader) |
| TCU_THROW(NotSupportedError, "Task shaders not supported"); |
| } |
| } |
| #endif // CTS_USES_VULKANSC |
| |
| if (caseDef.spirvVersion > vk::getMaxSpirvVersionForVulkan(context.getUsedApiVersion())) |
| TCU_THROW(NotSupportedError, "Shader requires SPIR-V version higher than available"); |
| } |
| |
| void supportedCheckFeaturesShader(Context &context, CaseDefinition caseDef) |
| { |
| supportedCheckFeatures(context, caseDef); |
| |
| subgroups::supportedCheckShader(context, caseDef.shaderStage); |
| } |
| |
| TestStatus noSSBOtest(Context &context, const CaseDefinition caseDef) |
| { |
| const VkFormat format = VK_FORMAT_R32_UINT; |
| const uint32_t &flags = caseDef.pipelineShaderStageCreateFlags; |
| const struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| 0u, |
| false, |
| }; |
| |
| switch (caseDef.shaderStage) |
| { |
| case VK_SHADER_STAGE_VERTEX_BIT: |
| return subgroups::makeVertexFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkVertexPipelineStages, flags, 0u); |
| case VK_SHADER_STAGE_GEOMETRY_BIT: |
| return subgroups::makeGeometryFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkVertexPipelineStages, flags, 0u); |
| case VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT: |
| return subgroups::makeTessellationEvaluationFrameBufferTestRequiredSubgroupSize( |
| context, format, nullptr, 0, &internalData, checkVertexPipelineStages, caseDef.shaderStage, flags, 0u); |
| case VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT: |
| return subgroups::makeTessellationEvaluationFrameBufferTestRequiredSubgroupSize( |
| context, format, nullptr, 0, &internalData, checkVertexPipelineStages, caseDef.shaderStage, flags, 0u); |
| case VK_SHADER_STAGE_FRAGMENT_BIT: |
| return subgroups::makeFragmentFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkFragmentPipelineStages, flags, 0u); |
| default: |
| TCU_THROW(InternalError, "Unhandled shader stage"); |
| } |
| } |
| |
| TestStatus test(Context &context, const CaseDefinition caseDef) |
| { |
| if (isAllComputeStages(caseDef.shaderStage)) |
| { |
| const uint32_t numWorkgroups[3] = {1, 1, 1}; |
| const uint32_t subgroupSize = subgroups::getSubgroupSize(context); |
| const auto &physicalDeviceProperties = context.getDeviceProperties(); |
| const auto &maxWorkGroupSize = physicalDeviceProperties.limits.maxComputeWorkGroupSize; |
| const auto &maxInvocations = physicalDeviceProperties.limits.maxComputeWorkGroupInvocations; |
| // Calculate the local workgroup sizes to exercise the maximum supported by the driver |
| const UVec3 localSize = getLocalSizes(maxWorkGroupSize, maxInvocations, maxInvocations); |
| const uint32_t localSizesToTestCount = 16; |
| const uint32_t localSizesToTest[localSizesToTestCount][3] = { |
| {1, 1, 1}, |
| {32, 4, 1}, |
| {32, 1, 4}, |
| {1, 32, 4}, |
| {1, 4, 32}, |
| {4, 1, 32}, |
| {4, 32, 1}, |
| {subgroupSize, 1, 1}, |
| {1, subgroupSize, 1}, |
| {1, 1, subgroupSize}, |
| {3, 5, 7}, |
| {128, 1, 1}, |
| {1, 128, 1}, |
| {1, 1, 64}, |
| {localSize.x(), localSize.y(), localSize.z()}, |
| {1, 1, 1} // Isn't used, just here to make double buffering checks easier |
| }; |
| const struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| subgroupSize, |
| false, |
| }; |
| |
| return subgroups::makeComputeTestRequiredSubgroupSize( |
| context, VK_FORMAT_R32_UINT, nullptr, 0, &internalData, checkCompute, |
| caseDef.pipelineShaderStageCreateFlags, numWorkgroups, makeDeBool(internalData.isRequiredSubgroupSize), |
| subgroupSize, localSizesToTest, localSizesToTestCount); |
| } |
| #ifndef CTS_USES_VULKANSC |
| else if (isAllMeshShadingStages(caseDef.shaderStage)) |
| { |
| const bool isMesh = ((caseDef.shaderStage & VK_SHADER_STAGE_MESH_BIT_EXT) != 0u); |
| const bool isTask = ((caseDef.shaderStage & VK_SHADER_STAGE_TASK_BIT_EXT) != 0u); |
| |
| DE_ASSERT(isMesh != isTask); |
| DE_UNREF(isTask); // For release builds. |
| |
| const uint32_t numWorkgroups[3] = {1, 1, 1}; |
| const uint32_t subgroupSize = subgroups::getSubgroupSize(context); |
| const auto &meshProperties = context.getMeshShaderPropertiesEXT(); |
| const auto &maxWorkGroupSize = |
| (isMesh ? meshProperties.maxMeshWorkGroupSize : meshProperties.maxTaskWorkGroupSize); |
| const auto &maxInvocations = |
| (isMesh ? meshProperties.maxMeshWorkGroupInvocations : meshProperties.maxTaskWorkGroupInvocations); |
| // Calculate the local workgroup sizes to exercise the maximum supported by the driver |
| const UVec3 localSize = getLocalSizes(maxWorkGroupSize, maxInvocations, maxInvocations); |
| const uint32_t localSizesToTestCount = 16; |
| const uint32_t localSizesToTest[localSizesToTestCount][3] = { |
| {1, 1, 1}, |
| {32, 4, 1}, |
| {32, 1, 4}, |
| {1, 32, 4}, |
| {1, 4, 32}, |
| {4, 1, 32}, |
| {4, 32, 1}, |
| {subgroupSize, 1, 1}, |
| {1, subgroupSize, 1}, |
| {1, 1, subgroupSize}, |
| {3, 5, 7}, |
| {128, 1, 1}, |
| {1, 128, 1}, |
| {1, 1, 64}, |
| {localSize.x(), localSize.y(), localSize.z()}, |
| {1, 1, 1} // Isn't used, just here to make double buffering checks easier |
| }; |
| const struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| subgroupSize, |
| false, |
| }; |
| |
| return subgroups::makeMeshTestRequiredSubgroupSize( |
| context, VK_FORMAT_R32_UINT, nullptr, 0, &internalData, checkCompute, |
| caseDef.pipelineShaderStageCreateFlags, numWorkgroups, makeDeBool(internalData.isRequiredSubgroupSize), |
| subgroupSize, localSizesToTest, localSizesToTestCount); |
| } |
| #endif // CTS_USES_VULKANSC |
| else if (isAllGraphicsStages(caseDef.shaderStage)) |
| { |
| const VkShaderStageFlags stages = subgroups::getPossibleGraphicsSubgroupStages(context, caseDef.shaderStage); |
| struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| 0u, |
| false, |
| }; |
| |
| return subgroups::allStagesRequiredSubgroupSize( |
| context, VK_FORMAT_R32_UINT, nullptr, 0, &internalData, checkVertexPipelineStages, stages, |
| caseDef.pipelineShaderStageCreateFlags, caseDef.pipelineShaderStageCreateFlags, |
| caseDef.pipelineShaderStageCreateFlags, caseDef.pipelineShaderStageCreateFlags, |
| caseDef.pipelineShaderStageCreateFlags, nullptr); |
| } |
| #ifndef CTS_USES_VULKANSC |
| else if (isAllRayTracingStages(caseDef.shaderStage)) |
| { |
| const VkShaderStageFlags stages = subgroups::getPossibleRayTracingSubgroupStages(context, caseDef.shaderStage); |
| const vector<uint32_t> flags(6, caseDef.pipelineShaderStageCreateFlags); |
| const struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| 0u, |
| false, |
| }; |
| |
| return subgroups::allRayTracingStagesRequiredSubgroupSize(context, VK_FORMAT_R32_UINT, nullptr, 0, |
| &internalData, checkVertexPipelineStages, stages, |
| flags.data(), nullptr); |
| } |
| #endif // CTS_USES_VULKANSC |
| else |
| TCU_THROW(InternalError, "Unknown stage or invalid stage set"); |
| } |
| |
| TestStatus testRequireFullSubgroups(Context &context, const CaseDefinition caseDef) |
| { |
| DE_ASSERT(VK_SHADER_STAGE_COMPUTE_BIT == caseDef.shaderStage); |
| DE_ASSERT(caseDef.requiredSubgroupSizeMode == REQUIRED_SUBGROUP_SIZE_NONE); |
| |
| const uint32_t numWorkgroups[3] = {1, 1, 1}; |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| const VkPhysicalDeviceProperties &physicalDeviceProperties = context.getDeviceProperties(); |
| // Calculate the local workgroup sizes to exercise the maximum supported by the driver |
| const auto &maxWorkGroupSize = physicalDeviceProperties.limits.maxComputeWorkGroupSize; |
| const auto &maxInvocations = physicalDeviceProperties.limits.maxComputeWorkGroupInvocations; |
| const UVec3 localSize = getLocalSizes(maxWorkGroupSize, maxInvocations, maxInvocations); |
| const uint32_t subgroupSize = subgroups::getSubgroupSize(context); |
| // For full subgroups and allow varying subgroup size, localsize X must be a multiple of maxSubgroupSize. |
| // We set local size X for this test to the maximum, regardless if allow varying subgroup size is enabled or not. |
| const uint32_t localSizesToTestCount = 7; |
| const uint32_t localSizesToTest[localSizesToTestCount][3] = { |
| {subgroupSizeControlProperties.maxSubgroupSize, 1, 1}, |
| {subgroupSizeControlProperties.maxSubgroupSize, 4, 1}, |
| {subgroupSizeControlProperties.maxSubgroupSize, 1, 4}, |
| {subgroupSizeControlProperties.maxSubgroupSize * 2, 1, 2}, |
| {subgroupSizeControlProperties.maxSubgroupSize * 4, 1, 1}, |
| {localSize.x(), localSize.y(), localSize.z()}, |
| {1, 1, 1} // Isn't used, just here to make double buffering checks easier |
| }; |
| const struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| subgroupSize, |
| false, |
| }; |
| |
| DE_ASSERT(caseDef.requiredSubgroupSizeMode == REQUIRED_SUBGROUP_SIZE_NONE); |
| |
| return subgroups::makeComputeTestRequiredSubgroupSize( |
| context, VK_FORMAT_R32G32B32A32_UINT, nullptr, 0, &internalData, checkComputeRequireFull, |
| caseDef.pipelineShaderStageCreateFlags, numWorkgroups, makeDeBool(internalData.isRequiredSubgroupSize), |
| subgroupSize, localSizesToTest, localSizesToTestCount); |
| } |
| |
| TestStatus testRequireSubgroupSize(Context &context, const CaseDefinition caseDef) |
| { |
| if (isAllComputeStages(caseDef.shaderStage)) |
| { |
| const uint32_t numWorkgroups[3] = {1, 1, 1}; |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| const VkPhysicalDeviceProperties &physicalDeviceProperties = context.getDeviceProperties(); |
| const uint32_t requiredSubgroupSize = |
| getRequiredSubgroupSizeFromMode(context, caseDef, subgroupSizeControlProperties); |
| const uint64_t maxSubgroupLimitSize = |
| (uint64_t)requiredSubgroupSize * subgroupSizeControlProperties.maxComputeWorkgroupSubgroups; |
| const uint32_t maxTotalLocalSize = (uint32_t)min<uint64_t>( |
| maxSubgroupLimitSize, physicalDeviceProperties.limits.maxComputeWorkGroupInvocations); |
| const auto &maxWorkGroupSize = physicalDeviceProperties.limits.maxComputeWorkGroupSize; |
| const auto &maxInvocations = physicalDeviceProperties.limits.maxComputeWorkGroupInvocations; |
| const UVec3 localSize = getLocalSizes(maxWorkGroupSize, maxInvocations, maxTotalLocalSize); |
| const bool shaderUsesFullSubgroups = caseDef.shaderUsesFullSubgroups(); |
| const uint32_t localSizesToTest[5][3] = { |
| {localSize.x(), localSize.y(), localSize.z()}, |
| {requiredSubgroupSize, 1, 1}, |
| {1, requiredSubgroupSize, 1}, |
| {1, 1, requiredSubgroupSize}, |
| {1, 1, 1} // Isn't used, just here to make double buffering checks easier |
| }; |
| |
| // If the shader uses full subgroups, use only the first two entries so the local size in X is a multiple of the requested |
| // subgroup size, as required by the spec. |
| uint32_t localSizesToTestCount = 5; |
| if (shaderUsesFullSubgroups) |
| localSizesToTestCount = 3; |
| |
| const internalDataStruct internalData = { |
| &context, // const Context* context; |
| caseDef, // struct CaseDefinition caseDef; |
| requiredSubgroupSize, // uint32_t requiredSubgroupSize; |
| true, // bool isRequiredSubgroupSize; |
| }; |
| |
| // Depending on the flag and SPIR-V version we need to run one verification function or another. |
| const auto checkFunction = (shaderUsesFullSubgroups ? checkComputeRequireFull : checkCompute); |
| |
| return subgroups::makeComputeTestRequiredSubgroupSize( |
| context, VK_FORMAT_R32G32B32A32_UINT, nullptr, 0, &internalData, checkFunction, |
| caseDef.pipelineShaderStageCreateFlags, numWorkgroups, makeDeBool(internalData.isRequiredSubgroupSize), |
| requiredSubgroupSize, localSizesToTest, localSizesToTestCount); |
| } |
| #ifndef CTS_USES_VULKANSC |
| else if (isAllMeshShadingStages(caseDef.shaderStage)) |
| { |
| const auto isMesh = ((caseDef.shaderStage & VK_SHADER_STAGE_MESH_BIT_EXT) != 0u); |
| const auto isTask = ((caseDef.shaderStage & VK_SHADER_STAGE_TASK_BIT_EXT) != 0u); |
| |
| DE_ASSERT(isMesh != isTask); |
| DE_UNREF(isTask); // For release builds. |
| |
| const uint32_t numWorkgroups[3] = {1, 1, 1}; |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| const auto &meshProperties = context.getMeshShaderPropertiesEXT(); |
| const uint32_t requiredSubgroupSize = |
| getRequiredSubgroupSizeFromMode(context, caseDef, subgroupSizeControlProperties); |
| const auto &maxWorkGroupSize = |
| (isMesh ? meshProperties.maxMeshWorkGroupSize : meshProperties.maxTaskWorkGroupSize); |
| const auto &maxInvocations = |
| (isMesh ? meshProperties.maxMeshWorkGroupInvocations : meshProperties.maxTaskWorkGroupInvocations); |
| const UVec3 localSize = getLocalSizes(maxWorkGroupSize, maxInvocations, maxInvocations); |
| const bool shaderUsesFullSubgroups = caseDef.shaderUsesFullSubgroups(); |
| const uint32_t localSizesToTest[5][3] = { |
| {requiredSubgroupSize, 1, 1}, |
| {1, requiredSubgroupSize, 1}, |
| {1, 1, requiredSubgroupSize}, |
| {localSize.x(), localSize.y(), localSize.z()}, |
| {1, 1, 1} // Isn't used, just here to make double buffering checks easier |
| }; |
| |
| // If the shader uses full subgroups, use only the first two entries so the local size in X is a multiple of the requested |
| // subgroup size, as required by the spec. |
| uint32_t localSizesToTestCount = 5; |
| if (shaderUsesFullSubgroups) |
| localSizesToTestCount = 3; |
| |
| const internalDataStruct internalData = { |
| &context, // const Context* context; |
| caseDef, // struct CaseDefinition caseDef; |
| requiredSubgroupSize, // uint32_t requiredSubgroupSize; |
| true, // bool isRequiredSubgroupSize; |
| }; |
| |
| // Depending on the flag and SPIR-V version we need to run one verification function or another. |
| const auto checkFunction = (shaderUsesFullSubgroups ? checkComputeRequireFull : checkCompute); |
| |
| return subgroups::makeMeshTestRequiredSubgroupSize( |
| context, VK_FORMAT_R32G32B32A32_UINT, nullptr, 0, &internalData, checkFunction, |
| caseDef.pipelineShaderStageCreateFlags, numWorkgroups, makeDeBool(internalData.isRequiredSubgroupSize), |
| requiredSubgroupSize, localSizesToTest, localSizesToTestCount); |
| } |
| #endif // CTS_USES_VULKANSC |
| else if (isAllGraphicsStages(caseDef.shaderStage)) |
| { |
| const VkShaderStageFlags stages = subgroups::getPossibleGraphicsSubgroupStages(context, caseDef.shaderStage); |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| const uint32_t requiredSubgroupSize = |
| getRequiredSubgroupSizeFromMode(context, caseDef, subgroupSizeControlProperties); |
| const uint32_t requiredSubgroupSizes[5] = {requiredSubgroupSize, requiredSubgroupSize, requiredSubgroupSize, |
| requiredSubgroupSize, requiredSubgroupSize}; |
| const internalDataStruct internalData = { |
| &context, // const Context* context; |
| caseDef, // struct CaseDefinition caseDef; |
| requiredSubgroupSize, // uint32_t requiredSubgroupSize; |
| true, // bool isRequiredSubgroupSize; |
| }; |
| |
| return subgroups::allStagesRequiredSubgroupSize( |
| context, VK_FORMAT_R32_UINT, nullptr, 0, &internalData, checkVertexPipelineStages, stages, |
| caseDef.pipelineShaderStageCreateFlags, caseDef.pipelineShaderStageCreateFlags, |
| caseDef.pipelineShaderStageCreateFlags, caseDef.pipelineShaderStageCreateFlags, |
| caseDef.pipelineShaderStageCreateFlags, requiredSubgroupSizes); |
| } |
| #ifndef CTS_USES_VULKANSC |
| else if (isAllRayTracingStages(caseDef.shaderStage)) |
| { |
| const VkShaderStageFlags stages = subgroups::getPossibleRayTracingSubgroupStages(context, caseDef.shaderStage); |
| const VkPhysicalDeviceSubgroupSizeControlProperties &subgroupSizeControlProperties = |
| context.getSubgroupSizeControlProperties(); |
| const uint32_t requiredSubgroupSize = |
| getRequiredSubgroupSizeFromMode(context, caseDef, subgroupSizeControlProperties); |
| const vector<uint32_t> flags(6, caseDef.pipelineShaderStageCreateFlags); |
| const vector<uint32_t> requiredSubgroupSizes(6, requiredSubgroupSize); |
| const struct internalDataStruct internalData = { |
| &context, // const Context* context; |
| caseDef, // struct CaseDefinition caseDef; |
| requiredSubgroupSize, // uint32_t requiredSubgroupSize; |
| true, // bool isRequiredSubgroupSize; |
| }; |
| |
| return subgroups::allRayTracingStagesRequiredSubgroupSize(context, VK_FORMAT_R32_UINT, nullptr, 0, |
| &internalData, checkVertexPipelineStages, stages, |
| flags.data(), requiredSubgroupSizes.data()); |
| } |
| #endif // CTS_USES_VULKANSC |
| else |
| TCU_THROW(InternalError, "Unknown stage or invalid stage set"); |
| } |
| |
| TestStatus noSSBOtestRequireSubgroupSize(Context &context, const CaseDefinition caseDef) |
| { |
| const auto &subgroupSizeControlProperties = context.getSubgroupSizeControlProperties(); |
| const uint32_t requiredSubgroupSize = |
| getRequiredSubgroupSizeFromMode(context, caseDef, subgroupSizeControlProperties); |
| const VkFormat format = VK_FORMAT_R32_UINT; |
| const uint32_t &flags = caseDef.pipelineShaderStageCreateFlags; |
| const uint32_t &size = requiredSubgroupSize; |
| struct internalDataStruct internalData = { |
| &context, |
| caseDef, |
| requiredSubgroupSize, |
| true, |
| }; |
| |
| switch (caseDef.shaderStage) |
| { |
| case VK_SHADER_STAGE_VERTEX_BIT: |
| return subgroups::makeVertexFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkVertexPipelineStages, flags, size); |
| case VK_SHADER_STAGE_GEOMETRY_BIT: |
| return subgroups::makeGeometryFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkVertexPipelineStages, flags, size); |
| case VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT: |
| return subgroups::makeTessellationEvaluationFrameBufferTestRequiredSubgroupSize( |
| context, format, nullptr, 0, &internalData, checkVertexPipelineStages, caseDef.shaderStage, flags, size); |
| case VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT: |
| return subgroups::makeTessellationEvaluationFrameBufferTestRequiredSubgroupSize( |
| context, format, nullptr, 0, &internalData, checkVertexPipelineStages, caseDef.shaderStage, flags, size); |
| case VK_SHADER_STAGE_FRAGMENT_BIT: |
| return subgroups::makeFragmentFrameBufferTestRequiredSubgroupSize(context, format, nullptr, 0, &internalData, |
| checkFragmentPipelineStages, flags, size); |
| default: |
| TCU_THROW(InternalError, "Unhandled shader stage"); |
| } |
| } |
| |
| TestStatus testSanitySubgroupSizeProperties(Context &context) |
| { |
| #ifndef CTS_USES_VULKANSC |
| VkPhysicalDeviceSubgroupSizeControlProperties subgroupSizeControlProperties; |
| subgroupSizeControlProperties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_SIZE_CONTROL_PROPERTIES; |
| #else |
| VkPhysicalDeviceSubgroupSizeControlPropertiesEXT subgroupSizeControlProperties; |
| subgroupSizeControlProperties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_SIZE_CONTROL_PROPERTIES_EXT; |
| #endif // CTS_USES_VULKANSC |
| |
| subgroupSizeControlProperties.pNext = nullptr; |
| |
| VkPhysicalDeviceSubgroupProperties subgroupProperties; |
| subgroupProperties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_PROPERTIES; |
| subgroupProperties.pNext = &subgroupSizeControlProperties; |
| |
| VkPhysicalDeviceProperties2 properties; |
| properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2; |
| properties.pNext = &subgroupProperties; |
| |
| context.getInstanceInterface().getPhysicalDeviceProperties2(context.getPhysicalDevice(), &properties); |
| |
| if (subgroupProperties.subgroupSize > subgroupSizeControlProperties.maxSubgroupSize || |
| subgroupProperties.subgroupSize < subgroupSizeControlProperties.minSubgroupSize) |
| { |
| ostringstream error; |
| error << "subgroupSize (" << subgroupProperties.subgroupSize << ") is not between maxSubgroupSize ("; |
| error << subgroupSizeControlProperties.maxSubgroupSize << ") and minSubgroupSize ("; |
| error << subgroupSizeControlProperties.minSubgroupSize << ")"; |
| |
| return TestStatus::fail(error.str().c_str()); |
| } |
| |
| return TestStatus::pass("OK"); |
| } |
| } // namespace |
| |
| namespace vkt |
| { |
| namespace subgroups |
| { |
| TestCaseGroup *createSubgroupsSizeControlTests(TestContext &testCtx) |
| { |
| de::MovePtr<TestCaseGroup> group(new TestCaseGroup(testCtx, "size_control")); |
| de::MovePtr<TestCaseGroup> framebufferGroup(new TestCaseGroup(testCtx, "framebuffer")); |
| de::MovePtr<TestCaseGroup> computeGroup(new TestCaseGroup(testCtx, "compute")); |
| de::MovePtr<TestCaseGroup> graphicsGroup(new TestCaseGroup(testCtx, "graphics")); |
| #ifndef CTS_USES_VULKANSC |
| de::MovePtr<TestCaseGroup> raytracingGroup(new TestCaseGroup(testCtx, "ray_tracing")); |
| de::MovePtr<TestCaseGroup> meshGroup(new TestCaseGroup(testCtx, "mesh")); |
| #endif // CTS_USES_VULKANSC |
| de::MovePtr<TestCaseGroup> genericGroup(new TestCaseGroup(testCtx, "generic")); |
| const VkShaderStageFlags fbStages[] = { |
| VK_SHADER_STAGE_VERTEX_BIT, |
| VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT, |
| VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT, |
| VK_SHADER_STAGE_GEOMETRY_BIT, |
| VK_SHADER_STAGE_FRAGMENT_BIT, |
| }; |
| #ifndef CTS_USES_VULKANSC |
| const VkShaderStageFlags meshStages[] = { |
| VK_SHADER_STAGE_MESH_BIT_EXT, |
| VK_SHADER_STAGE_TASK_BIT_EXT, |
| }; |
| #endif // CTS_USES_VULKANSC |
| |
| // Test sanity of the subgroup size properties. |
| { |
| addFunctionCase(genericGroup.get(), "subgroup_size_properties", supportedCheck, |
| testSanitySubgroupSizeProperties); |
| } |
| |
| const TestParams testParams[] = {{false, true, ""}, {true, false, "_spirv16"}, {true, true, "_flags_spirv16"}}; |
| |
| for (const auto ¶ms : testParams) |
| { |
| // Allow varying subgroup cases. |
| const uint32_t flagsVary = VK_PIPELINE_SHADER_STAGE_CREATE_ALLOW_VARYING_SUBGROUP_SIZE_BIT_EXT; |
| const CaseDefinition caseDefVary = {params.flagsEnabled ? flagsVary : 0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| false, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| |
| addFunctionCaseWithPrograms(computeGroup.get(), "allow_varying_subgroup_size" + params.postfix, |
| supportedCheckFeatures, initPrograms, test, caseDefVary); |
| addFunctionCaseWithPrograms(graphicsGroup.get(), "allow_varying_subgroup_size" + params.postfix, |
| supportedCheckFeaturesShader, initPrograms, test, caseDefVary); |
| |
| for (int stageIndex = 0; stageIndex < DE_LENGTH_OF_ARRAY(fbStages); ++stageIndex) |
| { |
| const CaseDefinition caseDefStage = {params.flagsEnabled ? flagsVary : 0u, |
| fbStages[stageIndex], |
| false, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| |
| string name = |
| getShaderStageName(caseDefStage.shaderStage) + "_allow_varying_subgroup_size" + params.postfix; |
| addFunctionCaseWithPrograms(framebufferGroup.get(), name, supportedCheckFeaturesShader, |
| initFrameBufferPrograms, noSSBOtest, caseDefStage); |
| } |
| |
| #ifndef CTS_USES_VULKANSC |
| for (const auto &stage : meshStages) |
| { |
| const CaseDefinition caseDefMesh = {(params.flagsEnabled ? flagsVary : 0u), |
| stage, |
| false, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| (params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_4)}; |
| const std::string name = getShaderStageName(stage) + "_allow_varying_subgroup_size" + params.postfix; |
| addFunctionCaseWithPrograms(meshGroup.get(), name, supportedCheckFeatures, initPrograms, test, caseDefMesh); |
| } |
| #endif // CTS_USES_VULKANSC |
| |
| // Require full subgroups together with allow varying subgroup (only compute shaders). |
| const uint32_t flagsFullVary = VK_PIPELINE_SHADER_STAGE_CREATE_REQUIRE_FULL_SUBGROUPS_BIT_EXT | |
| VK_PIPELINE_SHADER_STAGE_CREATE_ALLOW_VARYING_SUBGROUP_SIZE_BIT_EXT; |
| const CaseDefinition caseDefFullVary = {params.flagsEnabled ? flagsFullVary : 0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| true, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms( |
| computeGroup.get(), "require_full_subgroups_allow_varying_subgroup_size" + params.postfix, |
| supportedCheckFeatures, initProgramsRequireFull, testRequireFullSubgroups, caseDefFullVary); |
| |
| // Require full subgroups cases (only compute shaders). |
| const uint32_t flagsFull = VK_PIPELINE_SHADER_STAGE_CREATE_REQUIRE_FULL_SUBGROUPS_BIT_EXT; |
| const CaseDefinition caseDefFull = {params.flagsEnabled ? flagsFull : 0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| true, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms(computeGroup.get(), "require_full_subgroups" + params.postfix, |
| supportedCheckFeatures, initProgramsRequireFull, testRequireFullSubgroups, |
| caseDefFull); |
| |
| // Tests to check setting a required subgroup size value, together with require full subgroups (only compute shaders). |
| const CaseDefinition caseDefMaxFull = {params.flagsEnabled ? flagsFull : 0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| true, |
| REQUIRED_SUBGROUP_SIZE_MAX, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms( |
| computeGroup.get(), "required_subgroup_size_max_require_full_subgroups" + params.postfix, |
| supportedCheckFeatures, initProgramsRequireFull, testRequireSubgroupSize, caseDefMaxFull); |
| |
| const CaseDefinition caseDefMinFull = {params.flagsEnabled ? flagsFull : 0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| true, |
| REQUIRED_SUBGROUP_SIZE_MIN, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms( |
| computeGroup.get(), "required_subgroup_size_min_require_full_subgroups" + params.postfix, |
| supportedCheckFeatures, initProgramsRequireFull, testRequireSubgroupSize, caseDefMinFull); |
| |
| // Ray tracing cases with allow varying subgroup. |
| #ifndef CTS_USES_VULKANSC |
| const uint32_t flagsRayTracing = VK_PIPELINE_SHADER_STAGE_CREATE_ALLOW_VARYING_SUBGROUP_SIZE_BIT_EXT; |
| const CaseDefinition caseDefAllRaytracing = {params.flagsEnabled ? flagsRayTracing : 0u, |
| SHADER_STAGE_ALL_RAY_TRACING, |
| false, |
| REQUIRED_SUBGROUP_SIZE_NONE, |
| de::SharedPtr<bool>(new bool), |
| params.useSpirv16 ? SPIRV_VERSION_1_6 : SPIRV_VERSION_1_4}; |
| addFunctionCaseWithPrograms(raytracingGroup.get(), "allow_varying_subgroup_size" + params.postfix, |
| supportedCheckFeaturesShader, initPrograms, test, caseDefAllRaytracing); |
| #endif // CTS_USES_VULKANSC |
| } |
| |
| // Tests to check setting a required subgroup size value. |
| { |
| const CaseDefinition caseDefAllGraphicsMax = {0u, |
| VK_SHADER_STAGE_ALL_GRAPHICS, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MAX, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms(graphicsGroup.get(), "required_subgroup_size_max", supportedCheckFeaturesShader, |
| initPrograms, testRequireSubgroupSize, caseDefAllGraphicsMax); |
| const CaseDefinition caseDefComputeMax = {0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MAX, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms(computeGroup.get(), "required_subgroup_size_max", supportedCheckFeatures, |
| initPrograms, testRequireSubgroupSize, caseDefComputeMax); |
| #ifndef CTS_USES_VULKANSC |
| const CaseDefinition caseDefAllRaytracingMax = {0u, |
| SHADER_STAGE_ALL_RAY_TRACING, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MAX, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_4}; |
| addFunctionCaseWithPrograms(raytracingGroup.get(), "required_subgroup_size_max", supportedCheckFeaturesShader, |
| initPrograms, testRequireSubgroupSize, caseDefAllRaytracingMax); |
| #endif // CTS_USES_VULKANSC |
| |
| const CaseDefinition caseDefAllGraphicsMin = {0u, |
| VK_SHADER_STAGE_ALL_GRAPHICS, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MIN, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms(graphicsGroup.get(), "required_subgroup_size_min", supportedCheckFeaturesShader, |
| initPrograms, testRequireSubgroupSize, caseDefAllGraphicsMin); |
| const CaseDefinition caseDefComputeMin = {0u, |
| VK_SHADER_STAGE_COMPUTE_BIT, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MIN, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms(computeGroup.get(), "required_subgroup_size_min", supportedCheckFeatures, |
| initPrograms, testRequireSubgroupSize, caseDefComputeMin); |
| #ifndef CTS_USES_VULKANSC |
| const CaseDefinition caseDefAllRaytracingMin = {0u, |
| SHADER_STAGE_ALL_RAY_TRACING, |
| false, |
| REQUIRED_SUBGROUP_SIZE_MIN, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_4}; |
| addFunctionCaseWithPrograms(raytracingGroup.get(), "required_subgroup_size_min", supportedCheckFeaturesShader, |
| initPrograms, testRequireSubgroupSize, caseDefAllRaytracingMin); |
| #endif // CTS_USES_VULKANSC |
| for (int stageIndex = 0; stageIndex < DE_LENGTH_OF_ARRAY(fbStages); ++stageIndex) |
| { |
| const CaseDefinition caseDefStageMax = {0u, |
| fbStages[stageIndex], |
| false, |
| REQUIRED_SUBGROUP_SIZE_MAX, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms( |
| framebufferGroup.get(), getShaderStageName(caseDefStageMax.shaderStage) + "_required_subgroup_size_max", |
| supportedCheckFeaturesShader, initFrameBufferPrograms, noSSBOtestRequireSubgroupSize, caseDefStageMax); |
| const CaseDefinition caseDefStageMin = {0u, |
| fbStages[stageIndex], |
| false, |
| REQUIRED_SUBGROUP_SIZE_MIN, |
| de::SharedPtr<bool>(new bool), |
| SPIRV_VERSION_1_3}; |
| addFunctionCaseWithPrograms( |
| framebufferGroup.get(), getShaderStageName(caseDefStageMin.shaderStage) + "_required_subgroup_size_min", |
| supportedCheckFeaturesShader, initFrameBufferPrograms, noSSBOtestRequireSubgroupSize, caseDefStageMin); |
| } |
| |
| #ifndef CTS_USES_VULKANSC |
| for (const auto &stage : meshStages) |
| { |
| const auto stageName = getShaderStageName(stage); |
| |
| const CaseDefinition caseDefMeshMax = { |
| 0u, stage, false, REQUIRED_SUBGROUP_SIZE_MAX, de::SharedPtr<bool>(new bool), SPIRV_VERSION_1_4}; |
| addFunctionCaseWithPrograms(meshGroup.get(), "required_subgroup_size_max_" + stageName, |
| supportedCheckFeatures, initPrograms, testRequireSubgroupSize, caseDefMeshMax); |
| const CaseDefinition caseDefMeshMin = { |
| 0u, stage, false, REQUIRED_SUBGROUP_SIZE_MIN, de::SharedPtr<bool>(new bool), SPIRV_VERSION_1_4}; |
| addFunctionCaseWithPrograms(meshGroup.get(), "required_subgroup_size_min_" + stageName, |
| supportedCheckFeatures, initPrograms, testRequireSubgroupSize, caseDefMeshMin); |
| } |
| #endif // CTS_USES_VULKANSC |
| } |
| |
| group->addChild(genericGroup.release()); |
| group->addChild(graphicsGroup.release()); |
| group->addChild(computeGroup.release()); |
| group->addChild(framebufferGroup.release()); |
| #ifndef CTS_USES_VULKANSC |
| group->addChild(raytracingGroup.release()); |
| group->addChild(meshGroup.release()); |
| #endif // CTS_USES_VULKANSC |
| |
| return group.release(); |
| } |
| |
| } // namespace subgroups |
| } // namespace vkt |
