Skip to content

Releases: KhronosGroup/VK-GL-CTS

Vulkan CTS 1.2.3.0

18 Jun 16:07
Compare
Choose a tag to compare

Vulkan CTS 1.2.3.0

New tests:

  • Test multiple fragment output locations
  • Try spec constants with more sizes and packing
  • Add tests for VK_EXT_filter_cubic
  • Add SampleMask write tests
  • Add drm format modifiers tests
  • Test using ViewportIndex from frag shaders only
  • Test OpArrayLength with stride larger than element size
  • Add an Amber test which uses textureOffset()
  • Test negative value conversion to b10g11r11 float format.
  • Extend Android Hardware Buffer external memory format tests
  • Test variableMultisampleRate with non-empty framebuffer
  • Test mixed sample count in empty subpass and framebuffer
  • Add fp16 float control tests without 16bit_storage requirement
  • Test interpolation functions with single components
  • Add tests for VK_AMD_shader_trinary_minmax
  • Test OpCompositeInsert with OpUndef
  • Test strided stores in SSBOs
  • Add a batch of GraphicsFuzz tests
  • Check vkGetInstanceProcAddr works with NULL instance
  • Add VK_AMD_shader_image_load_store_lod tests
  • Enable MSAAx16 non-stict wide lines rasterization tests
  • Test nearest sampling edge cases
  • Test large indirect draw count and small parameter
  • Added tests for line continuity
  • Add test api.tooling_info.validate_instance_layers
  • Test execution modes set only in Tess Control Shader
  • Add tests for VK_EXT_tooling_info
  • Add test case for negative int div and
  • Test ViewportIndex SPIR-V builtin in frag shaders
  • Add tests covering variableMultisampleRate
  • Test Stream, XfbBuffer and XfbStride decorations on structs
  • Test exact sampling with linear filtering over a solid color
  • Add dual-source blending tests
  • Add tests for scissor constraint
  • Add tests for VK_AMD_shader_explicit_vertex_parameter
  • Test not/xor SPIR-V operation sequence
  • Test filtering of YCbCr formats
  • Test winding order of primitive topologies
  • Test multiple sets with dynamic offsets
  • VK_EXT_extended_dynamic_state tests

Improvements:

  • Update Android API levels and grant permissions
  • Adjust framework to removal of non-API enums
  • Waiver mechanism
  • Remove TODO, FIXME that were left behind
  • Framework definitions for VK_KHR_ray_tracing

Fixes

  • Fix VkPhysicalDeviceMemoryBudgetPropertiesEXT
  • Fix compilation on clang 11
  • Fix Fedora 32 gcc/clang warnings
  • Fix warnings found by compiling with MS VC++ 2019
  • Fix BindImageMemoryInfo in *render.device_group2
  • Correct few pipeline_barrier tests
  • Do not use 2D array d/s attachment created from 3D texture
  • Fix overflow in VK_EXT_subgroup_size_control tests.
  • Add conformance tests for new buffer offset rules
  • Remove non-compute variants of common function tests
  • Fixes clipping tests supported check
  • Allow RTZ rounding in spv_assemblywrites_two_buffers tests
  • Allow Bresenham for single sample nonstrict lines
  • Split spirv_assembly tests workload
  • Add --deqp-print-validation-errors cmd line option
  • Correct few tests from indirect_dispatch group
  • Correct sparse_image_format_properties2 tests
  • Clarify integer formats conversions in blit_image tests
  • Correct external_memory_host tests
  • Correct opatomic_storage_buffer_volatile tests
  • fix input dependency for renderpass depth/stencil format tests
  • Split subgroupBroadcast tests
  • Make ImageSamplingInstance loop over pNexts
  • Fix of a build failure on QNX
  • Replace LAST with MAX_ENUM in mapBorderColor
  • Make checkSupportImageSamplingInstance loop through pNexts
  • Fix verification in multisample image store tests
  • Use mandatory formats in stencil export tests
  • Correct Flush(/Invalidate)MappedMemoryRange
  • Fix validation problems with renderpass tests
  • Clean ExternalMemoryUtil
  • Test memory leaks in vkEnumeratePhysicalDevices
  • Apply nonUniformEXT after constructing a combined image/sampler
  • Improve image logging in some renderpass tests
  • Add Mandatory extension features
  • Fix format features required with samplerYcbcrConversion
  • Fix number of pipelines/framebuffers/renderpasses created in dynamic offset tests
  • Avoid fragment shader in some depth/stencil tests
  • Fix 16bit_storage dependency in some spirv instruction fp16 tests
  • Add shaderSubgroupExtendedTypes to mandatory features
  • Fix format features required with samplerYcbcrConversion
  • Remove duplicated code from isinf/isnan tests
  • Single random float generator for precision tests
  • Add check for YCbCr Chroma Reconstruction Explicit format support
  • Fix a typo for Linux run of deqp-vk
  • Add missing writes to PointSize in subgroups tests
  • Fixes timeline semaphore test queues
  • Fix image layouts in graphics spirv_assembly tests
  • Fix maxMipMap calculation for images
  • Fix image layouts in graphics spirv_assembly tests
  • Fix descriptorCount for dEQP-VK.ycbcr.conversion
  • Clear unused image aspects in buffer to depth/stencil copy tests
  • Add storageBuffer16BitAccess to mandatory features
  • Add missing barrier before host access in draw tests
  • Don't request DebugReportRecorder with no layers
  • Require VK_EXT_filter_cubic for cubic filtering
  • Fixes execServer with target API 28
  • Fixes advanced blending tests support check

Vulkan CTS 1.2.2.1

18 May 15:31
Compare
Choose a tag to compare

Vulkan CTS 1.2.2.1

Fixes:

  • Add missing barrier to HOST in memory_barrier test
  • Fix wide line rasterization tests
  • Avoid exceeding maxPerStageResources in descriptor_indexing tests
  • Remove event tests which use illegal behaviour
  • Fixes FDM subsampled tests
  • Prevent odd width single plane 422 images in dEQP-VK.ycbcr.query.*
  • Queries test should handle devices with disabled inheritedQueries
  • Add memory barrier to fragment shader interlock tests.
  • Fix framework for R64 int images test
  • Add missing extensions VK_GOOGLE
  • Fix invalid SPIR-V in GraphicsFuzz tests
  • remove opconstantnull physical storage pointer tests
  • Half-float texture interpolation tolerance is too small.
  • Use vkBindImageMemory2 to bind disjoint images

Vulkan CTS 1.2.2.0

26 Mar 17:30
Compare
Choose a tag to compare

Vulkan CTS 1.2.2.0

New tests:

  • Add more tests for VK_BUFFER_CREATE_DEVICE_ADDRESS_CAPTURE_REPLAY_BIT
  • Add tests for non strict lines
  • Test OpUndef with OpConstantComposite and OpSpecConstantComposite
  • Add depth and stencil formats to buffer copy tests
  • Add concurrent query pool tests
  • Test SampledImage as function argument
  • Added a batch of coverage tests found by GraphicsFuzz
  • Test additional atomic image SPIR-V operations
  • Validate samplerless descriptor writes
  • Test Centroid SPIR-V decoration on struct members
  • Test OpMemberDecorateString from SPIR-V 1.4
  • Test samplers with unnormalized coordinates
  • Test exact sampling cases
  • Improve vkGet*ProcAddr testing
  • Test shift ops in OpSpecConstantOp with more bit widths
  • Extend tests for VK_EXT_fragment_density_map
  • Add image clearing tests with multiple subresource ranges
  • Test OpImageWrite with the Sample image operand
  • Add geometry layered tests for 3D depth/stencil views
  • Add tests for VK_EXT_subgroup_size_control
  • Add tests for subgroup LOD
  • Test creation of overly large buffers
  • Run SPIR-V builtin precision tests on 64-bit floats
  • Test SpecConstantOp/CompositeInsert in arrays and structs
  • Add tests of VkSwapchain acquiring too many images
  • Extend multiview/occlusion queries tests
  • Added DMA buffer external memory cases to cross instance sharing tests
  • Add SPIR-V Invariant decoration tests

Fixes:

  • Require 16-bit storage in some 8-bit storage tests
  • Fix barrier code in dEQP-VK.api.buffer_marker.* tests
  • Fix dEQP-VK.subgroups.ballot.framebuffer.* to use core API
  • Verify results in ballot bitcount tests
  • Allow 1ULP of tolerance in ldexp precision tests
  • Fix 32-bit statistic query tests using 64 bits
  • Restore disabled texture function tests
  • Check if display is available for Vulkan.
  • Fix subgroups tests using basePipelineHandle with index != -1
  • Don't enable both buffer device address extensions
  • Don't remove 16-bit denorms from fp64 tests
  • Stricter codomain for atan tests
  • Use common code for SG arithmetic-like tests
  • Loosen basic subgroup elect test vertex checks
  • Stricter codomain in atan2 precision tests
  • Simplify the test code for ballot bit counts
  • Use float16 precision for testing mediump
  • Remove duplicated code that handles Android hardware buffer
  • Fix image checking for sparse image binding tests
  • ldexp tests do not test corner cases correctly
  • Fix texture shadow test using image views with zero aspect
  • Fixes 16bit storage check in spirv_assembly tests
  • More vertex-invocation fixes for synchronisation tests

Vulkan CTS 1.2.1.1

12 Mar 15:32
Compare
Choose a tag to compare

Vulkan CTS 1.2.1.1

New tests:

  • Add tests for VK_KHR_shader_non_semantic_info

Fixes:

  • Add 8bit support checks to subgroup tests
  • Check if display is available for Vulkan.
  • Fix memory leak in pipeline.executable_properties tests
  • Correctly handle normalized formats in FloatFormat
  • Protected Memory allocation tests require compatible device
  • Return Quality Warning for inaccurate filtering
  • Fix dEQP-VK.geometry.layered.2d_array.readback
  • Protected Memory allocation tests require compatible device
  • Only require geometryStreams feature in XFB query tests if needed
  • Fix transform feedback triangle strip verification
  • Fix dEQP-VK.api.image_clearing.*.clear_depth_stencil_image
  • Fix init in vktApiBufferMarkerTests.cpp
  • Fix use of pointer to stack allocated struct

OpenGL ES CTS 3.2.6.1

30 Jan 17:29
Compare
Choose a tag to compare

OpenGL ES CTS 3.2.6.1

Fixes:

  • Remove more tests due to GLSL ES 1.0/3.x differences
  • KHR-GLES31.core.constant_expressions.* add T&G tests to the test tree
  • Remove tests that check non-zero values for reservedMustBeZero on all APIs
  • Don't check color outcome for tests that expect undefined behaviour
  • Replace imageAtomicExchange with imageAtomicCompSwap in vertex shader
  • Unsized internalformats not valid GetInternalFormat

Vulkan CTS 1.2.1.0

17 Jan 08:20
Compare
Choose a tag to compare

Vulkan CTS 1.2.1.0

Vulkan CTS 1.1.6.3

10 Dec 20:26
Compare
Choose a tag to compare

Vulkan CTS 1.1.6.3

New tests:

  • Tests for VK_KHR_buffer_device_address

Fixes:

  • Bump timeout for the basic fence tests
  • Temporarily remove lines_wide tests
  • Do the 'from UNDEFINED' layout transition only on the first pass
  • Add two more formats with proper threshold values.
  • Extend rasterization testing, modify verification
  • Check memory model support in volatile atomic tests
  • Add feature checks to subgroups tests
  • Fix SFR swapchain test's implicit layout transitions
  • Apply the same relaxed precision fix for interface_blocks.* to basic_type.*
  • Remove verification of string bytes after null termination.
  • Fix cbuffer packing test
  • Re-align importing dedicated memory to spec
  • Add memory barrier to VK transform_feedback tests
  • Split descriptor set indexing render pass workload
  • Reduce dEQP-VK.pipeline.render_to_image allocs to within platform limits

Vulkan CTS 1.1.6.2

19 Nov 12:16
Compare
Choose a tag to compare

Vulkan CTS 1.1.6.2

New tests:

  • Tests for VK_KHR_performance_query

Fixes:

  • Do not reset perf queries in multi submitted cmdbuffer
  • Check renderpass2 support in image clearing tests
  • Fix check of api features
  • Fix descriptor_indexing.input_attachment subpass description
  • Fix missing image memory barrier in WSI display timing tests

Vulkan CTS 1.1.6.1

11 Nov 08:41
Compare
Choose a tag to compare

Vulkan CTS 1.1.6.1

New tests:

  • Tests for VK_KHR_separate_depth_stencil_layouts

Fixes:

  • Add missing write to PointSize
  • Add timelineSempahore to mandatory ext features
  • Fix check in testMandatoryExtensions
  • Remove deprecated std::bind2nd() call
  • Add scissor matching renderArea to descriptor indexing tests
  • Check available functionality with context method
  • Fix incorrect assumption for pipeline cache hit
  • Fixes VS num instances assumption

Vulkan CTS 1.1.6.0

24 Oct 15:18
Compare
Choose a tag to compare

Vulkan CTS 1.1.6.0

New tests:

  • Tests for SPIR-V 1.4
  • Enable explicit_lod tests
  • Add pipeline derivatives to compute subgroup tests
  • Various GraphicsFuzz tests
  • Tests for integer cooperative matrix
  • Test blitting with 1D and 3D depth/stencil images
  • Add VK_NV_ray_tracing infrastructure and test
  • Add copy tests for wide and tall images
  • Add tests for VK_KHR_shader_subgroup_extended_types
  • Test minimum alignment for texel buffers.
  • VK_EXT_line_rasterization tests
  • Test SPIRV comparison operations with 64-bit types
  • Add tests checking max varyings per shader stage
  • Add OpFRem to SPIR-V precision tests
  • Test for loop with condition always false
  • Test cbuffer packing with scalar block layout
  • Test enum with settings for separate controls
  • Add touch tests for volatile atomics
  • Test input attachments in random descriptor sets test group
  • Test buffer and image properties with external host memory
  • Add VK_EXT_shader_subgroup_ballot gl_SubGroup*MaskARB tests
  • Testing for sampleShadingEnable
  • Increase mediump precision testing
  • Add depthBounds tests for VK_EXT_depth_range_unrestricted
  • convert{su}tof and convertfto{su} tests with float16 need Float16 capability
  • Test vkCmdClearAttachments with unused depth/stencil attachments
  • Test 64-bit results with transform feedback queries
  • Test 32-bit stat queries with vkGetQueryPoolResults
  • Add QueryIndexed tests when streamId is zero
  • Test DMA buffers as external image memory
  • Test 32-bit timestamp query results
  • Add tests for VK_EXT_shader_subgroup_ballot
  • Test vkCmdCopyQueryPoolResults with transform feedback queries

Fixes:

  • Skip test with no display plane is available
  • Fix subnormal floats in extended image usage tests
  • Fix relaxed precision and no perspective dEQP tests
  • Fixed flags used to select queue in dEQP-VK.synchronization.op.multi_queue
  • Fix semaphore creation in smoke tests
  • Split descriptor set indexing render pass workload
  • Add feature requirements to existing amber files
  • Add variable pointers requirement for vktSpvAsmPtrAccessChainTests
  • Fix buffer alignment in dynamic offset tests
  • Fix broadcastFirst tests
  • remove opconstantnull physical storage pointer tests
  • Fix basicMultiFenceWaitAllFalseCase test
  • Fix 64bit type promotions in subgroup tests
  • Generate 'false' sometimes for random bool data
  • Tidy up extension headers in ballot_broadcast tests
  • Remove unnecessary extension from float16 constant tests
  • Remove 16 storage from float16 logicals tests
  • Remove 16-bit storage requirement from float16 func tests
  • Remove 16-bit storage requirement from half arithmetic tests
  • Remove unnecessary 16-bit storage requirement
  • Accept saturated values for the 32bit buffer
  • Fix API usage errors in geometry tests
  • Texel view compatible: Image: proper check for image format supports
  • 1d_image: enforce height 1 in decompress stage as well
  • 1d_image: basic: compute: texel_view_compatible: set size.y to 1
  • external fence tests: account for sync fd -1 in multiple export case
  • Fix depth store flag in seperate_channels tests
  • Float controls: graphics: fix 16bit storage validation
  • Need Float16 capability for float16 tests
  • Fix Float16 validation for 16bt storage tests
  • Float controls: break out extra functions as separate from constants
  • Account for precision loss in unorm depth pipeline
  • Float controls: compute: fix for strict 16bit storage validation
  • Add missing invalidates in depth/stencil resolve tests
  • Remove superfluous shuffle code
  • Rename subgroup.partitioned tests
  • Use templates instead of copying code
  • Fix 16bit/8bit storage validation for workgroup memory tests
  • Need Float16 in some derivative tests
  • Check if the statistics query results are linear
  • Don't generate then squash subnormal numbers
  • Fix validation VK_KHR_16bit_storage in opphi test
  • Add input attachment restrictions to makeCreateImageInfo()
  • Fix random generation for isinf/isnan tests
  • Remove float16 derivatives tests
  • Make texture aniso basic self tests less sensitive
  • Relaxes line rasterization tolerances
  • Generate a non-zero exit code on test failure
  • Use the correct imageType for image support checks.
  • Fix wrong wait stage mask
  • Fix in shadercacheLoad

Improvements:

  • Remove uses of strstr in vktSpvAsmFloatControlsTests.cpp
  • Limit the apiVersion of the instance within the CTS framework
  • Test all quad broadcasts in one test
  • Use a single test source for quad tests
  • Allow multiprocess execution for conformance submissions
  • Always require glslang and spirv-tools to be present
  • gen_ext_deps.py made independent from VK versions
  • Run SPIR-V assembly type tests in compute shaders
  • Improve verification in sampled image tests
  • Enable validation layers for custom devices and instances
  • Combine ARB and Core subgroupBroadcast code
  • Share instance/device across test case instances in dEQP-VK.synchronization.*
  • Use the default device in dEQP-VK.api.buffer.* tests
  • Cut down on repetition in the subgroup tests
  • Remove duplication from YCbCrUtils
  • Further tidy up the subgroup vote sources
  • Backport non-extension subgroup changes from next-dev