thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

Fixed #175

Browse source
This commit is contained in:
Alejandro Saucedo 2021-03-06 11:19:56 +00:00
parent 76bd167485
commit 196c896bd6
3 changed files with 120 additions and 117 deletions
Download patch file Download diff file Expand all files Collapse all files

109
src/Shader.cpp
View file

@ -105,5 +105,114 @@ Shader::compile_source(
resource);
}
const TBuiltInResource Shader::defaultResource = {
/* .MaxLights = */ 0,
/* .MaxClipPlanes = */ 0,
/* .MaxTextureUnits = */ 0,
/* .MaxTextureCoords = */ 0,
/* .MaxVertexAttribs = */ 64,
/* .MaxVertexUniformComponents = */ 4096,
/* .MaxVaryingFloats = */ 64,
/* .MaxVertexTextureImageUnits = */ 0,
/* .MaxCombinedTextureImageUnits = */ 0,
/* .MaxTextureImageUnits = */ 0,
/* .MaxFragmentUniformComponents = */ 0,
/* .MaxDrawBuffers = */ 0,
/* .MaxVertexUniformVectors = */ 128,
/* .MaxVaryingVectors = */ 8,
/* .MaxFragmentUniformVectors = */ 0,
/* .MaxVertexOutputVectors = */ 16,
/* .MaxFragmentInputVectors = */ 0,
/* .MinProgramTexelOffset = */ -8,
/* .MaxProgramTexelOffset = */ 7,
/* .MaxClipDistances = */ 8,
/* .MaxComputeWorkGroupCountX = */ 65535,
/* .MaxComputeWorkGroupCountY = */ 65535,
/* .MaxComputeWorkGroupCountZ = */ 65535,
/* .MaxComputeWorkGroupSizeX = */ 1024,
/* .MaxComputeWorkGroupSizeY = */ 1024,
/* .MaxComputeWorkGroupSizeZ = */ 64,
/* .MaxComputeUniformComponents = */ 1024,
/* .MaxComputeTextureImageUnits = */ 16,
/* .MaxComputeImageUniforms = */ 8,
/* .MaxComputeAtomicCounters = */ 8,
/* .MaxComputeAtomicCounterBuffers = */ 1,
/* .MaxVaryingComponents = */ 60,
/* .MaxVertexOutputComponents = */ 64,
/* .MaxGeometryInputComponents = */ 64,
/* .MaxGeometryOutputComponents = */ 128,
/* .MaxFragmentInputComponents = */ 0,
/* .MaxImageUnits = */ 0,
/* .MaxCombinedImageUnitsAndFragmentOutputs = */ 0,
/* .MaxCombinedShaderOutputResources = */ 8,
/* .MaxImageSamples = */ 0,
/* .MaxVertexImageUniforms = */ 0,
/* .MaxTessControlImageUniforms = */ 0,
/* .MaxTessEvaluationImageUniforms = */ 0,
/* .MaxGeometryImageUniforms = */ 0,
/* .MaxFragmentImageUniforms = */ 0,
/* .MaxCombinedImageUniforms = */ 0,
/* .MaxGeometryTextureImageUnits = */ 0,
/* .MaxGeometryOutputVertices = */ 256,
/* .MaxGeometryTotalOutputComponents = */ 1024,
/* .MaxGeometryUniformComponents = */ 1024,
/* .MaxGeometryVaryingComponents = */ 64,
/* .MaxTessControlInputComponents = */ 128,
/* .MaxTessControlOutputComponents = */ 128,
/* .MaxTessControlTextureImageUnits = */ 0,
/* .MaxTessControlUniformComponents = */ 1024,
/* .MaxTessControlTotalOutputComponents = */ 4096,
/* .MaxTessEvaluationInputComponents = */ 128,
/* .MaxTessEvaluationOutputComponents = */ 128,
/* .MaxTessEvaluationTextureImageUnits = */ 16,
/* .MaxTessEvaluationUniformComponents = */ 1024,
/* .MaxTessPatchComponents = */ 120,
/* .MaxPatchVertices = */ 32,
/* .MaxTessGenLevel = */ 64,
/* .MaxViewports = */ 16,
/* .MaxVertexAtomicCounters = */ 0,
/* .MaxTessControlAtomicCounters = */ 0,
/* .MaxTessEvaluationAtomicCounters = */ 0,
/* .MaxGeometryAtomicCounters = */ 0,
/* .MaxFragmentAtomicCounters = */ 0,
/* .MaxCombinedAtomicCounters = */ 8,
/* .MaxAtomicCounterBindings = */ 1,
/* .MaxVertexAtomicCounterBuffers = */ 0,
/* .MaxTessControlAtomicCounterBuffers = */ 0,
/* .MaxTessEvaluationAtomicCounterBuffers = */ 0,
/* .MaxGeometryAtomicCounterBuffers = */ 0,
/* .MaxFragmentAtomicCounterBuffers = */ 0,
/* .MaxCombinedAtomicCounterBuffers = */ 1,
/* .MaxAtomicCounterBufferSize = */ 16384,
/* .MaxTransformFeedbackBuffers = */ 4,
/* .MaxTransformFeedbackInterleavedComponents = */ 64,
/* .MaxCullDistances = */ 8,
/* .MaxCombinedClipAndCullDistances = */ 8,
/* .MaxSamples = */ 4,
/* .maxMeshOutputVerticesNV = */ 256,
/* .maxMeshOutputPrimitivesNV = */ 512,
/* .maxMeshWorkGroupSizeX_NV = */ 32,
/* .maxMeshWorkGroupSizeY_NV = */ 1,
/* .maxMeshWorkGroupSizeZ_NV = */ 1,
/* .maxTaskWorkGroupSizeX_NV = */ 32,
/* .maxTaskWorkGroupSizeY_NV = */ 1,
/* .maxTaskWorkGroupSizeZ_NV = */ 1,
/* .maxMeshViewCountNV = */ 4,
/* .maxDualSourceDrawBuffersEXT = */ 1,
/* .limits = */
{
/* .nonInductiveForLoops = */ 1,
/* .whileLoops = */ 1,
/* .doWhileLoops = */ 1,
/* .generalUniformIndexing = */ 1,
/* .generalAttributeMatrixVectorIndexing = */ 1,
/* .generalVaryingIndexing = */ 1,
/* .generalSamplerIndexing = */ 1,
/* .generalVariableIndexing = */ 1,
/* .generalConstantMatrixVectorIndexing = */ 1,
}
};
}
#endif // DKOMPUTE_DISABLE_SHADER_UTILS

122
src/include/kompute/Shader.hpp
View file

@ -12,124 +12,18 @@
namespace kp {
// The default resource limit for the GLSL compiler, can be overwritten
// Has been adobted by:
// https://github.com/KhronosGroup/glslang/blob/master/StandAlone/ResourceLimits.cpp
const TBuiltInResource defaultResource = {
/* .MaxLights = */ 0,
/* .MaxClipPlanes = */ 0,
/* .MaxTextureUnits = */ 0,
/* .MaxTextureCoords = */ 0,
/* .MaxVertexAttribs = */ 64,
/* .MaxVertexUniformComponents = */ 4096,
/* .MaxVaryingFloats = */ 64,
/* .MaxVertexTextureImageUnits = */ 0,
/* .MaxCombinedTextureImageUnits = */ 0,
/* .MaxTextureImageUnits = */ 0,
/* .MaxFragmentUniformComponents = */ 0,
/* .MaxDrawBuffers = */ 0,
/* .MaxVertexUniformVectors = */ 128,
/* .MaxVaryingVectors = */ 8,
/* .MaxFragmentUniformVectors = */ 0,
/* .MaxVertexOutputVectors = */ 16,
/* .MaxFragmentInputVectors = */ 0,
/* .MinProgramTexelOffset = */ -8,
/* .MaxProgramTexelOffset = */ 7,
/* .MaxClipDistances = */ 8,
/* .MaxComputeWorkGroupCountX = */ 65535,
/* .MaxComputeWorkGroupCountY = */ 65535,
/* .MaxComputeWorkGroupCountZ = */ 65535,
/* .MaxComputeWorkGroupSizeX = */ 1024,
/* .MaxComputeWorkGroupSizeY = */ 1024,
/* .MaxComputeWorkGroupSizeZ = */ 64,
/* .MaxComputeUniformComponents = */ 1024,
/* .MaxComputeTextureImageUnits = */ 16,
/* .MaxComputeImageUniforms = */ 8,
/* .MaxComputeAtomicCounters = */ 8,
/* .MaxComputeAtomicCounterBuffers = */ 1,
/* .MaxVaryingComponents = */ 60,
/* .MaxVertexOutputComponents = */ 64,
/* .MaxGeometryInputComponents = */ 64,
/* .MaxGeometryOutputComponents = */ 128,
/* .MaxFragmentInputComponents = */ 0,
/* .MaxImageUnits = */ 0,
/* .MaxCombinedImageUnitsAndFragmentOutputs = */ 0,
/* .MaxCombinedShaderOutputResources = */ 8,
/* .MaxImageSamples = */ 0,
/* .MaxVertexImageUniforms = */ 0,
/* .MaxTessControlImageUniforms = */ 0,
/* .MaxTessEvaluationImageUniforms = */ 0,
/* .MaxGeometryImageUniforms = */ 0,
/* .MaxFragmentImageUniforms = */ 0,
/* .MaxCombinedImageUniforms = */ 0,
/* .MaxGeometryTextureImageUnits = */ 0,
/* .MaxGeometryOutputVertices = */ 256,
/* .MaxGeometryTotalOutputComponents = */ 1024,
/* .MaxGeometryUniformComponents = */ 1024,
/* .MaxGeometryVaryingComponents = */ 64,
/* .MaxTessControlInputComponents = */ 128,
/* .MaxTessControlOutputComponents = */ 128,
/* .MaxTessControlTextureImageUnits = */ 0,
/* .MaxTessControlUniformComponents = */ 1024,
/* .MaxTessControlTotalOutputComponents = */ 4096,
/* .MaxTessEvaluationInputComponents = */ 128,
/* .MaxTessEvaluationOutputComponents = */ 128,
/* .MaxTessEvaluationTextureImageUnits = */ 16,
/* .MaxTessEvaluationUniformComponents = */ 1024,
/* .MaxTessPatchComponents = */ 120,
/* .MaxPatchVertices = */ 32,
/* .MaxTessGenLevel = */ 64,
/* .MaxViewports = */ 16,
/* .MaxVertexAtomicCounters = */ 0,
/* .MaxTessControlAtomicCounters = */ 0,
/* .MaxTessEvaluationAtomicCounters = */ 0,
/* .MaxGeometryAtomicCounters = */ 0,
/* .MaxFragmentAtomicCounters = */ 0,
/* .MaxCombinedAtomicCounters = */ 8,
/* .MaxAtomicCounterBindings = */ 1,
/* .MaxVertexAtomicCounterBuffers = */ 0,
/* .MaxTessControlAtomicCounterBuffers = */ 0,
/* .MaxTessEvaluationAtomicCounterBuffers = */ 0,
/* .MaxGeometryAtomicCounterBuffers = */ 0,
/* .MaxFragmentAtomicCounterBuffers = */ 0,
/* .MaxCombinedAtomicCounterBuffers = */ 1,
/* .MaxAtomicCounterBufferSize = */ 16384,
/* .MaxTransformFeedbackBuffers = */ 4,
/* .MaxTransformFeedbackInterleavedComponents = */ 64,
/* .MaxCullDistances = */ 8,
/* .MaxCombinedClipAndCullDistances = */ 8,
/* .MaxSamples = */ 4,
/* .maxMeshOutputVerticesNV = */ 256,
/* .maxMeshOutputPrimitivesNV = */ 512,
/* .maxMeshWorkGroupSizeX_NV = */ 32,
/* .maxMeshWorkGroupSizeY_NV = */ 1,
/* .maxMeshWorkGroupSizeZ_NV = */ 1,
/* .maxTaskWorkGroupSizeX_NV = */ 32,
/* .maxTaskWorkGroupSizeY_NV = */ 1,
/* .maxTaskWorkGroupSizeZ_NV = */ 1,
/* .maxMeshViewCountNV = */ 4,
/* .maxDualSourceDrawBuffersEXT = */ 1,
/* .limits = */
{
/* .nonInductiveForLoops = */ 1,
/* .whileLoops = */ 1,
/* .doWhileLoops = */ 1,
/* .generalUniformIndexing = */ 1,
/* .generalAttributeMatrixVectorIndexing = */ 1,
/* .generalVaryingIndexing = */ 1,
/* .generalSamplerIndexing = */ 1,
/* .generalVariableIndexing = */ 1,
/* .generalConstantMatrixVectorIndexing = */ 1,
}
};
/**
Shader utily class with functions to compile and process glsl files.
*/
class Shader
{
public:
// The default resource limit for the GLSL compiler, can be overwritten
// Has been adopted by:
// https://github.com/KhronosGroup/glslang/blob/master/StandAlone/ResourceLimits.cpp
const static TBuiltInResource defaultResource;
/**
* Compile multiple sources with optional filenames. Currently this function
* uses the glslang C++ interface which is not thread safe so this funciton
@ -150,7 +44,7 @@ class Shader
const std::vector<std::string>& files = {},
const std::string& entryPoint = "main",
std::vector<std::pair<std::string, std::string>> definitions = {},
const TBuiltInResource& resources = defaultResource);
const TBuiltInResource& resources = Shader::defaultResource);
/**
* Compile a single glslang source from string value. Currently this
@ -170,7 +64,7 @@ class Shader
const std::string& source,
const std::string& entryPoint = "main",
std::vector<std::pair<std::string, std::string>> definitions = {},
const TBuiltInResource& resources = defaultResource);
const TBuiltInResource& resources = Shader::defaultResource);
};
}

6
test/TestShaderResources.cpp
View file

@ -32,7 +32,7 @@ void compileShaderWithGivenResources(const std::string shaderString, const TBuil
TEST(TestShaderResources, TestNoMaxLight)
{
TBuiltInResource noMaxLightResources = kp::defaultResource;
TBuiltInResource noMaxLightResources = kp::Shader::defaultResource;
noMaxLightResources.maxLights=0;
EXPECT_NO_THROW(compileShaderWithGivenResources(shaderString, noMaxLightResources));
@ -41,7 +41,7 @@ TEST(TestShaderResources, TestNoMaxLight)
TEST(TestShaderResources, TestSmallComputeWorkGroupSizeX)
{
TBuiltInResource smallComputeWorkGroupSizeXResources = kp::defaultResource;
TBuiltInResource smallComputeWorkGroupSizeXResources = kp::Shader::defaultResource;
smallComputeWorkGroupSizeXResources.maxComputeWorkGroupSizeX=0;
ASSERT_THROW(compileShaderWithGivenResources(shaderString, smallComputeWorkGroupSizeXResources), std::runtime_error);
@ -50,7 +50,7 @@ TEST(TestShaderResources, TestSmallComputeWorkGroupSizeX)
TEST(TestShaderResources, TestNoWhileLoopLimit)
{
TBuiltInResource noWhileLoopLimitResources = kp::defaultResource;
TBuiltInResource noWhileLoopLimitResources = kp::Shader::defaultResource;
noWhileLoopLimitResources.limits.whileLoops=0;
ASSERT_THROW(compileShaderWithGivenResources(shaderString, noWhileLoopLimitResources), std::runtime_error);