The VkScript format is a clone of the format used by VkRunner as described in [1].
The # symbol can be used to start a comment which extends to the end of the line.
The \ can be used at the end of a line to signal a continuation, the new line will be skipped and parsing will treat the following line as a continuation of the current line.
Any command below which accepts a binding will accept either a single integer value which will have a descriptor set of 0 and a binding of the value give or a string can be provided of the form set integer:binding integer in which case the descriptor set value will be set and the binding value will be binding.
The format is broken down into five main sections:
requireshadersindicesvertex datatestThe require section lists all of the requirements for the testing environment. There are four types of information that can be encoded in the require section.
The feature list contains a list of features that are required in order for the test to execute. If a feature is missing an error will be reported and the test will fail. The features are listed below in the Available Require Features section.
The framebuffer and depthstencil commands allow setting the format for the given buffer. The valid values are listed below in the Image Formats section.
The fbsize command allows setting the width and height of the framebuffer.
The fence_timeout option allows setting an integer number of milliseconds for any fence timeouts.
The last option is extensions. Any string which isn't listed above is assumed to be an extension. The extensions must be of the format [a-zA-Z0-9_]+. If the device extension is not available we will report it is not available and the test will continue.
[require] independentBlend VK_KHR_storage_buffer_storage_class
The shader section allows you to specify the content of the shaders under test. This can be done as GLSL, SPIRV-ASM or SPIRV-Hex depending on how the shader is formatted. There is also a special passthrough vertex shader which can be used which just passes the vec4 input location 0 through to the gl_Position. The shader format is specified in the header after the word shader. The default is GLSL, SPIRV-ASM is specified as spirv and SPIRV-Hex as spirv hex.
The shaders accepted are:
computefragmentgeometrytessellation controltessellation evaulationvertex[fragment shader]
#version 430
layout(location = 0) out vec4 color_out;
void main() {
color_out = vec4(1, 2, 3, 4);
}
Other example shader header lines are:
[fragment shader spirv hex] -- a hex encoded SPIRV binary fragment shader[tessellation evaluation shader spirv] -- a spirv-asm tessellation evaluation shader[vertex shader passthrough]The vertex data section provides vertex attributes and data for draw array commands. The data is formated with a header row followed by data rows.
The headers can be provided in one of two forms. The first, attribute_location/format where attribute_location is the location of the attribute to be bound. The format is one of the Image Formats listed below. The second, attribute_location/gl_type/glsl_type. The gl_type is one of the types listed in the GL Types section below. The glsl_type is one listed in the GLSL Types section below.
[vertex data] 0/R32G32B32_SFLOAT 1/R8G8B8_UNORM -1 -1 0.25 255 0 0 # ending comment # Another Row 0.25 -1 0.25 255 0 255
The indices section contains the list of indices to use along with the provided vertex data. The indices are used if the indexed option is provided to the draw arrays command. The indices themselves are a list of integer indexes to use.
[indices] # comment line 1 2 3 4 5 6 # another comment 7 8 9 10 11 12
The test section contains a list of commands which can be executed to perform the actual testing. The commands range from setting up pipeline parameters, executing compute shaders and probing buffers to verify results.
draw rect [ortho] [patch] _x_ _y_ _width_ _height_The draw rect command draws a rectangle at the given coordinates. The vertices are uploaded at location 0 as a vec3. The ortho modifier scales the coordinates to be in the range of [-1,1] instead of [0,window size]. The patch modifier sets the draw call to use a given topology. Accepted possible topology value are listed in Topologies. The patch size will be set to 4.
draw arrays [indexed] [instanced] _topology_ _first_vertex_ _vertex_count_ [instance_count]The draw arrays command uses data from the vertex data section when executing the draw command. The topology is from the Topologies list. If the indexed modifier is provided then the indices section data will be used as well.
compute _x_ _y_ _z_Executes the compute shader with the given x, y, and z parameters.
_stage_ entrypoint _name_Sets the stage shader to use the entry point of name for subsequent executions.
probe all (rgb|rgba) _r_ _g_ _b_ [_a_]Probes the entire window to verify all pixels are of color r,g,b and optionally a. If rgba is specified then the a parameter is required. If rgb is specified then the a parameter is dis-allowed.
[relative] probe [rect] (rgb|rgba) (_x_, _y_[, _width_, _height_]) (_r_, _g_, _b_[, _a_])Probes a portion of the window to verify the pixes are of color r,g,b and optionally a. If rgba is specifed then the a parameter is required. If rgb is specified then the a parameter is dis-allowed. If rect is specified then width and height are required. If rect is not specified then width and height are dis-allowed and a value of 1 will be used for each parameter. If the relative parameter is provided the coordinates are normalized to be in the range [0.0, 1.0].
probe ssbo _type_ _binding_ _offset_ _comparison_ _values_+Probes the value in the storage buffer at binding and offset within that binding. The type is the data type to be probed as seen in the Data Types section below.
The comparison operators are:
== (equal)!= (not equal)< (less than)<= (less or equal)>= (greater or equal)~= (fuzzy equal, for floating point comparisons using tolerances)The values provided must be a non-zero multiple of the type.
uniform _type_ _offset _values_+Sets the push constants at offset. The type is from the Data Types section below. The values must be a non-zero multiple of the requested type.
When setting push constant data each call to uniform must use the same type or the script will be rejected as invalid.
uniform ubo _binding_ _type_ _offset_ _values_+Sets the values in the uniform buffer at binding and offset. The type is from the Data Types section below. The values must be a non-zero multiple of the requested type.
When setting data into a single binding, each call to uniform ubo must use the same type or the script will be rejected as invalid.
ssbo _binding_ _size_Sets the number of elements in the SSBO at binding to size. The buffer will be created with a default format of char. This default format will be overridden by a call to ssbo subdata or probe ssbo.
ssbo _binding_ subdata _type_ _offset_ _values_+Sets the value of the buffer at binding and offset. The type is from the Data Types section below. The values must be a non-zero multiple of the requested type. The offset must be a multiple of the type size in bytes.
When setting data into a single binding, each call to ssbo subdata must use the same type or the script will be rejected as invalid.
patch parameter vertices _count_Sets the number of control points for tessellation patches to count. Defaults to 3.
tolerance tolerance0 [tolerance1 tolerance2 tolerance3]The tolerance command sets the amount of fuzzyness used when using the ~= comparator. If a single tolerance value is set it is used for every comparison. If all four values are set then each vecN command will use the first N tolerance values. Each column of a matMxN will also use the first N tolerances. A tolerance maybe either a number or a percentage 0.01%.
clear color _r_ _g_ _b_ _a_Sets the clear color. Defaults to (0, 0, 0, 0).
clear depth _value_Sets the depth clear value. The value is a float and defaults to 1.0.
clear stencil _value_Sets the stencil clear value. The value is an integer and defaults to 0.
clearClears the framebuffer.
There are a number of pipeline flags which can be set to alter execution. Each draw call uses the pipeline configuration that was specified prior to the draw call.
The pipeline commands with their accepted data are:
primitiveRestartEnable <bool>depthClampEnable <bool>rasterizerDiscardEnable <bool>depthBiasEnable <bool>logicOpEnable <bool>blendEnable <bool>depthTestEnable <bool>depthWriteEnable <bool>depthBoundsTestEnable <bool>stencilTestEnable <bool>topology <VkPrimitiveTopology>polygonMode <VkPolygonMode>logicOp <VkLogicOp>frontFace <VkFrontFace>cullMode <VkCullMode>depthBiasConstantFactor <float>depthBiasClamp <float>depthBiasSlopeFactor <float>lineWidth <float>minDepthBounds <float>maxDepthBounds <float>srcColorBlendFactor <VkBlendFactor>dstColorBlendFactor <VkBlendFactor>srcAlphaBlendFactor <VkBlendFactor>dstAlphaBlendFactor <VkBlendFactor>colorBlendOp <VkBlendOp>alphaBlendOp <VkBlendOp>depthCompareOp <VkCompareOp>front.compareOp <VkCompareOp>back.compareOp <VkCompareOp>front.failOp <VkStencilOp>front.passOp <VkStencilOp>front.depthFailOp <VkStencilOp>back.failOp <VkStencilOp>back.passOp <VkStencilOp>back.depthFailOp <VkStencilOp>front.reference <uint32_t>back.reference <uint32_t>colorWriteMask <VkColorComponent bitmask>[test] clear color 1 0.4 0.5 0.2 clear relative probe rect rgba (0.0, 0.0, 1.0, 1.0) (1.0, 0.4, 0.5, 0.2)
intuintint8_tuint8_tint16_tuint16_tint64_tuint64_tfloatdoublevecvec[234]dvecdvec[234]ivecivec[234]uvecuvec[234]i8veci8vec[234]u8vecu8vec[234]i16veci16vec[234]u16vecu16vec[234]i64veci64vec[234]u64vecu64vec[234]matmat[234]x[234]dmatdmat[234]x[234]PATCH_LISTPOINT_LISTGL_LINE_STRIP_ADJACENCYGL_LINE_STRIPGL_LINESGL_LINES_ADJACENCYGL_PATCHESGL_POINTSGL_TRIANGLE_STRIPGL_TRIANGLE_FANGL_TRIANGLESGL_TRIANGLES_ADJACENCYGL_TRIANGLE_STRIP_ADJACENCYLINE_LISTLINE_LIST_WITH_ADJACENCYLINE_STRIPLINE_STRIP_WITH_ADJACENCYTRIANGLE_FANTRIANGLE_LISTTRIANGLE_LIST_WITH_ADJACENCYTRIANGLE_STRIPTRIANGLE_STRIP_WITH_ADJACENCYbyteubyteshortushortintuinthalffloatdoubleintuintfloatdoublevecvec2vec3vec4dvecdvec2dvec3dvec4uvecuvec2uvec3uvec4ivecivec2ivec3ivec4robustBufferAccessfullDrawIndexUint32imageCubeArrayindependentBlendgeometryShadertessellationShadersampleRateShadingdualSrcBlendlogicOpmultiDrawIndirectdrawIndirectFirstInstancedepthClampdepthBiasClampfillModeNonSoliddepthBoundswideLineslargePointsalphaToOnemultiViewportsamplerAnisotropytextureCompressionETC2textureCompressionASTC_LDRtextureCompressionBCocclusionQueryPrecisepipelineStatisticsQueryvertexPipelineStoresAndAtomicsfragmentStoresAndAtomicsshaderTessellationAndGeometryPointSizeshaderImageGatherExtendedshaderStorageImageExtendedFormatsshaderStorageImageMultisampleshaderStorageImageReadWithoutFormatshaderStorageImageWriteWithoutFormatshaderUniformBufferArrayDynamicIndexingshaderSampledImageArrayDynamicIndexingshaderStorageBufferArrayDynamicIndexingshaderStorageImageArrayDynamicIndexingshaderClipDistanceshaderCullDistanceshaderFloat64shaderInt64shaderInt16shaderResourceResidencyshaderResourceMinLodsparseBindingsparseResidencyBuffersparseResidencyImage2DsparseResidencyImage3DsparseResidency2SamplessparseResidency4SamplessparseResidency8SamplessparseResidency16SamplessparseResidencyAliasedvariableMultisampleRateinheritedQueriesA1R5G5B5_UNORM_PACK16A2B10G10R10_SINT_PACK32A2B10G10R10_SNORM_PACK32A2B10G10R10_SSCALED_PACK32A2B10G10R10_UINT_PACK32A2B10G10R10_UNORM_PACK32A2B10G10R10_USCALED_PACK32A2R10G10B10_SINT_PACK32A2R10G10B10_SNORM_PACK32A2R10G10B10_SSCALED_PACK32A2R10G10B10_UINT_PACK32A2R10G10B10_UNORM_PACK32A2R10G10B10_USCALED_PACK32A8B8G8R8_SINT_PACK32A8B8G8R8_SNORM_PACK32A8B8G8R8_SRGB_PACK32A8B8G8R8_SSCALED_PACK32A8B8G8R8_UINT_PACK32A8B8G8R8_UNORM_PACK32A8B8G8R8_USCALED_PACK32B10G11R11_UFLOAT_PACK32B4G4R4A4_UNORM_PACK16B5G5R5A1_UNORM_PACK16B5G6R5_UNORM_PACK16B8G8R8A8_SINTB8G8R8A8_SNORMB8G8R8A8_SRGBB8G8R8A8_SSCALEDB8G8R8A8_UINTB8G8R8A8_UNORMB8G8R8A8_USCALEDB8G8R8_SINTB8G8R8_SNORMB8G8R8_SRGBB8G8R8_SSCALEDB8G8R8_UINTB8G8R8_UNORMB8G8R8_USCALEDD16_UNORMD16_UNORM_S8_UINTD24_UNORM_S8_UINTD32_SFLOATD32_SFLOAT_S8_UINTR16G16B16A16_SFLOATR16G16B16A16_SINTR16G16B16A16_SNORMR16G16B16A16_SSCALEDR16G16B16A16_UINTR16G16B16A16_UNORMR16G16B16A16_USCALEDR16G16B16_SFLOATR16G16B16_SINTR16G16B16_SNORMR16G16B16_SSCALEDR16G16B16_UINTR16G16B16_UNORMR16G16B16_USCALEDR16G16_SFLOATR16G16_SINTR16G16_SNORMR16G16_SSCALEDR16G16_UINTR16G16_UNORMR16G16_USCALEDR16_SFLOATR16_SINTR16_SNORMR16_SSCALEDR16_UINTR16_UNORMR16_USCALEDR32G32B32A32_SFLOATR32G32B32A32_SINTR32G32B32A32_UINTR32G32B32_SFLOATR32G32B32_SINTR32G32B32_UINTR32G32_SFLOATR32G32_SINTR32G32_UINTR32_SFLOATR32_SINTR32_UINTR4G4B4A4_UNORM_PACK16R4G4_UNORM_PACK8R5G5B5A1_UNORM_PACK16R5G6B5_UNORM_PACK16R64G64B64A64_SFLOATR64G64B64A64_SINTR64G64B64A64_UINTR64G64B64_SFLOATR64G64B64_SINTR64G64B64_UINTR64G64_SFLOATR64G64_SINTR64G64_UINTR64_SFLOATR64_SINTR64_UINTR8G8B8A8_SINTR8G8B8A8_SNORMR8G8B8A8_SRGBR8G8B8A8_SSCALEDR8G8B8A8_UINTR8G8B8A8_UNORMR8G8B8A8_USCALEDR8G8B8_SINTR8G8B8_SNORMR8G8B8_SRGBR8G8B8_SSCALEDR8G8B8_UINTR8G8B8_UNORMR8G8B8_USCALEDR8G8_SINTR8G8_SNORMR8G8_SRGBR8G8_SSCALEDR8G8_UINTR8G8_UNORMR8G8_USCALEDR8_SINTR8_SNORMR8_SRGBR8_SSCALEDR8_UINTR8_UNORMR8_USCALEDS8_UINTX8_D24_UNORM_PACK321- https://github.com/Igalia/vkrunner/blob/d817f8b186cccebed89471580a685dc80a330946/README.md