Tags

Vulkan CTS 1.1.5.0

New tests:
 * Test image load/store with packed formats
 * Add tests for OpPtrAccessChain on Workgroup storage
 * Enable OpFMod testing for 16-bit
 * Check every available device and host time domains
 * Add more image dimensions to image clearing tests
 * Add test for concurrent draw and compute
 * Add VK_EXT_shader_subgroup_vote extension tests
 * Test vkCmdClearAttachments with unused attachments
 * Add tests without color buffer and with vertex only pipe
 * Add VK_EXT_depth_range_unrestricted tests
 * Add smoke test with unknown extension structure
 * Add VK_EXT_host_query_reset tests for transform feedback stream queries
 * Extend image store/load tests
 * Add test for mandatory device features ( chapter 35.1 )
 * Add more tests for YCbCr samplers
 * Add tests for EXT_separate_stencil_usage
 * Add tests for VK_EXT_external_memory_dma_buf
 * Add alpha to coverage test with unused attachment
 * Add tests for VK_EXT_calibrated_timestamps
 * Add SPIR-V unused variable tests
 * Add tests for VK_EXT_descriptor_indexing
 * Add tests for VK_EXT_pipeline_creation_feedback extension
 * Add multiview tests iterating with VK_EXT_vertex_attribute_divisor
 * Check vkCmdFillBuffer with sizes not multiple of 4
 * Test separate channels as input and output
 * Add tests for EXT_discard_rectangles
 * Add tests for UPDATE_AFTER_BIND bindings
 * VK_EXT_memory_budget and VK_EXT_memory_priority
 * Add Transform Feedback tests triangle-strip-with-adjacency
 * Add Transform Feedback tests for builtin variables
 * Add Transform Feedback tests
 * Add VK_EXT_pci_bus_info tests
 * Add tests for provoking vertex
 * Test multiple interpolations in a shader stage
 * Use nonsymmetric dimensions for attachment clear tests
 * Add tests for VK_KHR_imageless_framebuffer

Improvements:
 * Split shared_presentable_image tests into more dimensions to avoid timeout
 * Speed up hierarchy traversal, part 1
 * Reformat and share the subgroup vote shader source
 * Merge common util functions, part 2
 * Only test precision in compute stages
 * Early support check for tests (part 1)
 * Support validation layers in push descriptor tests
 * Autogenerate check(Device|Instance)Extensions
 * Unify getShaderStageName helper
 * Generate all DeviceFeatures related code
 * Require Python 3
 * Raise CMake required version to 3.0
 * Move to a single Vulkan mustpass
 * Allow C99, C++11
 * Auto-generate DeviceFeatures class
 * Add support for building with Visual Studio 2019

Fixes:
 * SPIR-V assembly tests require Invocations in Geometry shaders
 * Allows initialisation of float16 values for comparison tests by casting.
 * Fix more 16bit denorms in precision tests
 * Cast invocation ID correctly in allequal tests
 * Fix Android WSI shared presentable image test OOM
 * Fix Android WSI shared presentable image tests
 * Fix feature requirements of sample mask tests
 * Fix derivate dynamic tests to check shader stage support
 * Fix rounding when casting to integer
 * Fix template usage in precision tests
 * Fix memory commitment tests for devices with no lazily allocated memory
 * Fix vertex binding stride in rasterization.primitive_size.points
 * Fix Clang 8 and GCC 9 warnings
 * Remove unneeded casts from EXT subgroup vote tests
 * Remove unary plus operator tests
 * Fix build for concurrent draw tests
 * Remove duplicate floating-point precision testing
 * Check consistency of GetPhysicalDeviceSparseImageFormatProperties
 * Fix denorm code for fp16-as-32 tests
 * Add flags to check whether tests are supported
 * Fix depth stencil multisample image usage validation error
 * Fix protected memory swapchain tests
 * Fix typo in precision tests
 * Fix wrong image usage flags in api feature info tests
 * Remove unessential non-ascii symbols from source
 * Remove duplicate code from vktRenderPassTests.cpp
 * Fix synchronization issues in descriptor_indexing tests
 * Fixes DeviceFeatures long execution time
 * Fix sample read tests to use sample rate shading
 * Pull in TOT Vulkan headers
 * Fix validation errors in YCbCr array layer tests
 * Fix validation errors in buffer_device_address tests
 * Make depth less predictable in renderpass tests
 * Remove subpass clear colors initialization
 * Simplify precisions in GLSL precision tests
 * Fix random number generation in GLSL precision tests
 * Fix image_extent test to handle OOM errors
 * Improve VK_KHR_swapchain coverage
 * Compile only necessary shaders in texture tests
 * Fix build for GCC/Clang
 * Fix vkGetPhysicalDevicePresentRectanglesKHR() call
 * Require independentBlend in write mask tests
 * Use square dimensions for CUBE_COMPATIBLE 2D images
 * Add barriers to read query data after copy
 * Use separate filtering precision for each channel
 * Remove frexp tests from the list of test issues
 * Check sampleRateShading on sample shading tests
 * Add feature checks to shader builtin tests
 * Add GeometryPointSize feature check
 * Make validation function more robust
 * Check for largePoints feature and add 1px test variant
 * Check queries unavailable with reset and copy
 * Fix spelling of "preserve" in float controls tests
 * Don't pass YcbcrConversion objects when not used
 * Vary vertex buffer stride in pipeline barrier tests
 * Change spir-v optimizer recipes
 * Fix API usage errors in subgroups.* tests
 * Add *_MAX_ENUM to enum definitions
 * Fix shader common function tests
 * Fix run command for Windows
 * Update generate scripts for python3
 * Make random number usage platform independent
 * Factor out copies of makeBufferCreateInfo
 * Simplify support checking and fix typos in messages
 * Merge common util functions, part 1
 * vulkancts/README: Recommend -msse2 -mfpmath=sse for 32-bit x86 builds
 * Reduce expected precision of RGB10A2 filtering
 * Add feature checks to spirv assembly tests
 * Add float 16 calculation into FMA tests
 * Create YCbCr with PREDEFINED layout
 * Add hasDisplay check after createWsiDisplay throws
 * Missing format checks for vertex attribute and color attachments in input_ouput tests
 * Fix malformed assets, use DE_FATAL
 * Add helper function for creating Amber test cases
 * Fix buffer RAW hazards in MultiQueue synchronization tests
 * vktAmberTestCase::parse second argument is a string
 * Fix damage binary length field in shadercacheSave
 * Remove conformance version test from Android
 * Fix Android WSI incremental present image tests
 * Exclude dEQP-VK.ssbo.layout.random.scalar.75
 * Add compatible memory type check for given handle
 * Fix image layouts in descriptor indexing tests
 * Fix command line parameters processing of TestLog

Vulkan CTS 1.1.4.2

Fixes:

 * Handle 1.0 vs 1.1 differences for VK_KHR_device_group
 * Check for toplevel sparse binding support bit in sparse pipeline tests
 * Add a checker for vkQueuePresent returning VK_SUBOPTIMAL_KHR
 * Fix more sampleShading issues in dEQP-VK.pipeline.multisample_shader_builtin.*
 * Fix for SSBO offset alignment
 * Split incremental_present tests into more dimensions to avoid timeout
 * Fixes protected memory swapchain image format oom
 * Split compute workload into multiple submissions
 * Require variablePointersStorageBuffer feature in pointer_parameter tests
 * Fixes precise occlusion query tests
 * Waiver for device-scope core memory model tests.
 * Add GeometryPointSize feature check
 * Add feature checks to shader builtin tests
 * Check for imageCubeArray feature in image tests
 * Check for required features in memory model tests
 * Fix image sizes for formats that need a multiple of 2.
 * Check for largePoints feature and add 1px test variant
 * Remove incorrect unroll_dont_unroll test
 * Don't set sampleShadingEnable in dEQP-VK.pipeline.framebuffer_attachment.*
 * Fix stack underflow in incremental present test

Vulkan CTS 1.1.4.1

New tests:
 * Tests for VK_KHR_uniform_buffer_standard_layout

Fixes:
 * Fixes linear support check for pipeline tests
 * Check for depthBiasClamp feature before using it
 * Check for imageCubeArray support before using VK_IMAGE_VIEW_TYPE_CUBE_ARRAY
 * Reduce expected precision of RGB10A2 filtering
 * Fix memory initialisation in subgroups tests
 * Fixes pipeline tests to check for maxSamplerLodBias
 * Add float 16 calculation into FMA tests
 * Create YCbCr with PREDEFINED layout
 * Check for geometry shader support when testing adjacency topologies
 * Add hasDisplay check after createWsiDisplay throws
 * Add feature checks to spirv assembly tests
 * Fix uninitialized variable reported by gcc 8.3
 * Fix element size for R16G16B16A16_UNORM.
 * Fixed buffer memory type for protected test
 * Check for geometry shader support when testing adjacency topologies
 * Missing format checks for vertex attribute and color attachments in input_ouput tests

Improvements:
 * Require Python 3

Vulkan CTS 1.1.4.0

New tests:

 * Add tests for VK_EXT_host_query_reset
 * Add scalar uint with signed int compares as Amber cases
 * Add scissor tests
 * Additional coverage for color resolve
 * Add tests for OpFUnord with NaN
 * Add alpha to coverage test with no color attachment
 * Add tests for VK_NV_cooperative_matrix
 * Add test for concurrent compute
 * Add SPIR-V ids abuse tests
 * Add testing for VK_EXT_ycbcr_image_arrays
 * Add tests for pointSize with clamping
 * Add UTF8 abuse tests for extension and layer names
 * VK_EXT_buffer_device_address tests
 * Add tests for descriptor copy
 * Add layered rendering test for secondary cmd buffer
 * Add tests checking if no depth is written when depth test is disabled
 * Add testing for VK_EXT_depth_clip_enable
 * Add more merge cache tests
 * Memory model tests for transitive av/vis ops
 * Add tests for clearing cube attachments with non zero base array layer
 * Add tests of VK_EXT_conditional_rendering for secondary command buffer
 * Add missing PointSize to some geometry tests
 * Added variants of PointCoord test with uniforms
 * Extend api.version_check.entry_points test
 * Test swizzles of 2D texture lookup coordinates
 * Add tests for vkAcquireNextImage2KHR
 * Added test for backwards subpass self dependency

Fixes:

 * Avoid check vulkan core features in compute tests
 * Fix image layouts in SPIR-V image sampler tests
 * Fixed dEQP-VK.binding_model.descriptor_copy.* issues.
 * Optimize format flag fetch function
 * Limit transient images tests
 * Reduce memory usage in point size clamping tests
 * Disable pipeline_create_only when running Amber tests
 * Check shaderFloat16 support in 16bit_storage.*64*
 * Fix uninitialized variable reported by gcc 8.3
 * Fix build by adding DE_UNREF
 * Invalidate memory before checking results
 * Fix validation errors in ycbcr.query.* tests
 * Don't give samplers in YCbCr descriptor updates
 * Simplify the combination passed to vkCreateImage
 * Amber: Initialize the delegate to NULL
 * Update Amber so it detects Vulkan on Android
 * Send external api base extensions when not core
 * Drop handwritten toString code
 * Fix validation errors in dEQP-VK.tessellation.*
 * Use valid deviceMask in vkAcquireNextImage2KHR
 * Fix device mask in swapchain tests
 * Disable SPIR-V ids optimization
 * Add control for subpixel bits to reference rasterizer
 * Define point sizes in dEQP-VK.query_pool.* tests
 * Fix zero height clear rectangle
 * Fix spelling of error message
 * Fix stack underflow in incremental present test
 * Check if CTS has platform display for wsi-type
 * Change render size of shaderexecutor to 128
 * Fix calls to vkBindImageMemory2
 * Fix validation errors in subgroups tests
 * Reenable coverage of opaque type indexing
 * Remove incorrect vertexPipelineStoresAndAtomics dependencies
 * Correct support conditions for float16s
 * Remove one method of specifying features in vktSpvAsm
 * pipeline timestamps test: fix too-large VkBufferCopy
 * Set correct range for OpVectorTimesScalar tests
 * Allow a tolerance for results of sqrt
 * Fix incorrect use of copyImageToBuffer in TextureRenderer::renderQuad()
 * Fix image aspects in multiview tests
 * Fix memory requirements for draw_clear tests
 * Fix damage binary length field in shadercacheSave

Improvements:

 * Raise VS supported version to 2015
 * Update default Android API level to 28
 * Add Amber as a test engine
 * Move VkRunner tests to Amber
 * Make renderdoc_app.h optional at build time
 * Improve shader cache hashing

Vulkan CTS 1.1.3.2

Fixes:

 * Add handle compatibility check in win32 keyed mutex tests
 * Do not expect bit-exact results on float
 * cmake: Add support for PNG_ARM_NEON_OPT NDK-r19
 * Be more careful about Block decorations on structs
 * Fix the limitation in allocation test
 * Fix buffer RAW hazards in synchronization tests
 * Handle BUFFER_DYNAMIC and UPDATE_AFTER_BIND interaction
 * Fix buffer RAW hazards in MultiQueue synchronization tests
 * Don't disown Win32 handle after import

Vulkan CTS 1.1.3.1

New tests:

 * Add tests for VK_KHR_surface_protected_capabilities

Fixes:
 * Exclude few rasterization tests
 * Fix SpvWorkgroupMemoryTests float32 result check
 * Fixes YCbCr create flag requirements for image format properties tests
 * Fix RAW and WAW hazards in pipeline barrier tests.
 * Fix buffer RAW hazards in synchronization tests

 * Fix image type for uniform texel buffers
 * Check if CTS has platform display for wsi-type
 * Fix swapchain mutable tests buffer allocations
 * Check shaderInt64 feature support in memory tests
 * Fragment input components limit includes position
 * Fix mismatch between VS outputs and GS inputs
 * Fragment input components limit includes position
 * Revert "Fix subpixel precision for line geometry generation"
 * Fix memory allocation size in subgroups tests
 * Fix spirv validation issue in pointer_parameter
 * NaN handling in float16 tests
 * Add ASTC LDR error colour quality warning
 * Fix the device_group test failure by allowing same number of tries to create object as device test.
 * Reduce devices created in device_group tests

Vulkan CTS 1.1.3.0

New tests:
 * Tests for VK_KHR_depth_stencil_resolve
 * Tests for VK_EXT_conditional_rendering
 * Tests for VK_EXT_vertex_attribute_divisor
 * Tests for different data types for workgroup memory
 * Tests to intentionally mismatch opImageType
 * Tests using VkRunner, precompiling the scripts
 * Add subpass external dependency to pushDescriptor tests
 * Add tests for varying naming
 * Added tests for OpCompositeInsert
 * Enable remaining OpFUnord tests
 * Add test veryifying if proper 'count' is written in wsi getters
 * Add barrier at start of each CB to order wrt previous CBs
 * Enable previously excluded acos test
 * Add tests for derivatives in dynamic but quad-uniform control flow
 * Add tests for booleans produced from different bit-sizes
 * Add tests for late fragment test dependencies
 * Extend push desriptors test to 128 compute dispatches
 * Extend transcoding tests to 1D/3D images
 * Test 2D texture swizzles with VkComponentMappings
 * Test depth/stencil resolve attachments
 * Added tests for out-of-order attribute locations
 * test layered resolves with a single region convering all layers.
 * Test storing protected data in workgroup memory
 * Added tests for non-linear attribute locations
 * Add corner case values for float32 to int16 conversion tests.
 * Add tests for subpass dependencies
 * Test clip/cull distance reading in fragment shader
 * New tests for descriptor dynamic offset
 * Add dependency checking to dEQP-VK.api.info.*.extensions
 * Added several abuses of OpName and OpMemberName
 * dEQP-VK: Add tests for blitting to/from linear images
 * Test OpPtrAccessChain with base alignment of < 16
 * Test render pass attachments not referenced by subpasses
 * Test cube image blits with VK_IMAGE_CREATE_CUBE_COMPATIBLE flag.
 * Simultaneous depth+stencil components blit test
 * Ensure there's no KHX extensions
 * Added test set for unnormalized coord samplers
 * Add secondary command buffers to pipeline barrier tests
 * Create Compute test for OpCompositeExtract
 * New test for copying 16bit input to two outputs.
 * Test push descriptors

Fixes:
 * Fixes multiview query tests
 * Fix dependency on unspecified rounding mode
 * Correct support conditions for float16s
 * Clear Core Validation Errors for Protected Memory
 * Fix dup tests to not free in-use semaphores/fences
 * Fix 0b prefix in conditional rendering tests
 * Fix image aspects in multiview tests
 * Fix memory requirements for draw_clear tests
 * Fix further aspect masks in renderpass2 tests
 * Simplify YCbCr result verification
 * Fix initial YCbCr image layout
 * Fix validation of occlusion query tests
 * Fix more memory flush/invalidation cases
 * Validation fixes for device_init tests
 * Be more careful about Block decorations on structs
 * Remove duplicate tests under dEQP-VK.renderpass*
 * Fix API usage errors in fragment operation tests
 * Fixed incorrect load/store ops for depth in multiview test
 * Reduce devices created in device_group tests
 * Remove duplicate identical make*MemoryBarrier
 * Allow SRGB conversion for store tests to generate reference image.
 * Fix subgroup shuffle tests to do testing
 * Remove duplication from subgroup shuffle tests
 * Set the correct "unused" attachment initialLayout
 * Validation: Use correct aspects in renderpass2
 * Fix leak in shaderCacheLoad()
 * Correct image mutable tests on null driver
 * Remove duplicate float16Int8Features
 * Remove StorageUniform16 from SPIR-V conversion tests
 * Fix anonymous namespace in UBO matrix padding test
 * Properly align and pad vec16x2 values in UBO array
 * Require VK_KHR_shader_float16_int8 in float16 opconstant tests
 * Fixes for Protected Memory
 * Further validation fixes for renderpass tests
 * Fix invalid color values in image.mutable tests
 * Use the correct type for image creation flags
 * Fixes AHB image tests feature check
 * Added framework function for image to buffer copy
 * Correct invalidate->flush after writing to host-mapped memory
 * Specify PointSize when drawing point lists
 * Utilize copyBufferToImage framework function where possible
 * Fixed unused attachment tests to properly use renderpass2
 * SubgroupBallot: Don't test 0 <= id
 * Rationalise the 'check*' functions in subgroups
 * Remove one copy of the clustered test sources
 * Further validation fixes for renderpass tests
 * Use GENERAL image layout for images used as STORAGE image
 * Properly align and pad vec16x2 values in UBO array
 * Fix for unnormalized coordinate test using mipmaps
 * Fix API usage errors in dEQP-VK.geometry.basic.*
 * Fix swapchain creation in dEQP-VK.wsi.*.swapchain.*
 * Require VK_KHR_shader_float16_int8 in float16 opconstant tests
 * Import dedicated allocs w/ importDedicatedMemory
 * Properly analyze edge0>=edge1 cases in smoothstep
 * Remove NaN from OpFOrd* and OpFUnord* tests
 * Validation fixes for device group tests
 * Fix validation in dynamic offset tests
 * Mark storage buffer as readonly
 * Fix validation in renderpass tests
 * Fix precision mismatches in GLSL tests
 * Place OpVariable at the beginning of the block
 * Fix barrier range in UpdateBuffer tests
 * Fix validation in memory tests
 * Fix memory flush alignments in binding model tests
 * Fix validation in command buffer tests
 * Fix validation in memory binding tests
 * Fix queue validation in synchronization tests
 * Fix non-valid barriers in dEQP-VK.tessellation.*
 * Check sparse support in memory requirements tests
 * Don't pass tess and geom flags when not needed
 * Fix testing of various device limits
 * Fix api.smoke.* tests for validation
 * Fix unused function warnings in the CTS code
 * Add missing invalidates in renderpass tests
 * Fix synchronisation in ycbcr tests
 * Use a valid final layout in renderpass tests
 * Fix valid usage for SampleLocations extension
 * Fix some pipeline and memory barriers
 * Fix DeviceFeatures infinite loop in pNext chain
 * Fix array stride in uniform buffer for VK_KHR_16bit_storage
 * Drop superfluous Clip/CullDistance capabilities
 * Remove invalid execution mode in smoke test
 * Fix array stride in uniform buffer for VK_KHR_16bit_storage
 * Fix asserts in image.mutable tests
 * Remove over-zealous sanity check for memory types
 * dEQP-VK: Stop requiring the destination format to support linear sampling for blit tests
 * Fix copy/paste errors in opundef assembly tests
 * Fixes geometry shader invocation dependency
 * Fix malformed assets, use DE_FATAL
 * Add Flat decorations to integer fragment shader inputs
 * Fix result checking in multisample resolve tests
 * Fix asserts and check for base image format support in image.mutable tests
 * Fix NaN handling in funord tests
 * Fix validation failure in float_controls
 * Fixes ycbcr image properties requirement
 * Fix image type in Uniform texel buffer tests
 * Protected Memory Fixes - Barrier Layouts & Flags
 * Update VK_KHR_vulkan_memory_model to spec version 3
 * Fixes depth load_op in multiview tests
 * Truncate random coordinates in late fragment tests
 * Touch watchdog more often for precision tests.
 * Added PointSize in shader used for drawing points
 * Add AHB to Android allowed extensions list
 * Fix logging of 8-bit integer values
 * Fix incorrect use of copyImageToBuffer in TextureRenderer::renderQuad()
 * Allow coarse derivatives to be calced by fine func
 * Added missing writes to gl_Position in vertex shaders
 * Depth attachment not stored in late_fragment_tests

Improvements:
 * Compile shaders for 1.0-compatible tests as SPIRV 1.0, not max supported
 * Handle gcc 7 warnings
 * Build prebuilts for HLSL shaders
 * Remove one method of specifying features in vktSpvAsm
 * Cause vkrunner tests to fail if scripts not found
 * Allow a tolerance for results of sqrt
 * Add framework function createInstanceWithExtensions
 * Add support for RenderDoc in Vulkan tests
 * Added script for vulkan.h.in generation
 * Use precompiled headers on MSVC
 * Remove test lib dependencies on deqp-vk-package.
 * Add a flushAlloc helper to fix valid usage
 * Add SPIR-V optimization options to vk-build-programs
 * Improve wsi.incremental_present.* for low memory

Vulkan CTS 1.1.2.3

New tests:
 * Add convert tests for 8-bit integers
 * Add float <--> int conversion tests
 * Add operation tests for 8-bit integers
 * Add fp16 <-> fp64 conversion tests
 * Tests for VK_KHR_shader_float_controls
 * FP16 precision tests VK_KHR_shader_float16_int8
 * SPIR-V tests for VK_KHR_shader_float16_int8
 * Add tests for VK_EXT_scalar_block_layout
 * Add tests for OpSpecConstantOp INotEqual

Improvements:
 * Refactor: Compatible compute and graphics VerifyIO

Fixes:
 * Properly align and pad vec16x2 values in UBO array
 * Properly analyze edge0>=edge1 cases in smoothstep
 * Remove NaN from OpFOrd* and OpFUnord* tests
 * Fix array stride in uniform buffer for VK_KHR_16bit_storage
 * Don't include source files in other source files
 * Fix missing self dependency in renderpass tests
 * Add watchdog touches to dEQP-VK.api.descriptor_pool#repeated_reset_long
 * Fix swapchain creation in dEQP-VK.wsi.*.swapchain.*
 * Fix case when unsuported create flag was used for AHB
 * Correct vertex input tests on null driver
 * Correct image properties tests on Null driver
 * Fix non-constant Id in subgroupBroadcast
 * Remove texel_view uncompressed float formats
 * Fix: alloc_callback_fail.render_pass test
 * Fix: adds checking the maxPerStageResources limit
 * Fix pointer storage class for spirv_assembly tessellation tests.
 * Fix incorrect decoration on PushConstant
 * Remove unsupported image load store tests
 * Fix image feature checks for swapchain_mutable.*

Vulkan CTS 1.1.2.2

New tests:
 * Add tests for VK_KHR_swapchain_mutable_format

Improvements:
 * Update glslang to past the Turing extensions.
 * Update glslang to fix memory model spirv validation
 * fetch_sources: can remove a tag before fetching

Fixes:
 * Add 1.1.2.2 to known conformance versions
 * Place OpVariable at the beginning of the block
 * Fix 16-bit storage IO f32-to-f16 conversion tests
 * Remove invalid execution mode in smoke test
 * Fix uninitialised variable warning for some compilers
 * Fix incomplete VkImageSubresourceRange
 * Fix spelling on NotSupported error message
 * Make VK_KHR_shader_subgroup_vote tests work with sparse dispatch
 * Fix unaligned host accesses in UBO and SSBO tests
 * Add flush to binding descriptor set random tests
 * Fix the device_group test failure by allowing same number of tries to create object as device test.
 * Import dedicated allocs w/ importDedicatedMemory
 * Fix asserts and check for base image format support in image.mutable tests
 * Fix asserts in image.mutable tests
 * Correct copy_and_blit on Null driver
 * Don't exhaust 32-bit host visible address space

OpenGL CTS 4.6.0.2

Fixes:
 * CoverageGL21 does not reset GL_READ_BUFFER for FBO test
 * Use non-arrayed varying name for TCS blocks
 * Add screenLayout to AndroidManifest configChanges
 * Scale texture sizes to fit within available memory
 * Free framebuffer objects after use.
 * Do not expect TCS shader with XFB variables to fail linkging in GL
 * bufSize of glGetInternalformativ does not follow spec
 * Fix resource leaks in arrays_of_arrays
 * Don't used a fragment output in the shader storage buffer object test
 * TextureBarrier required for texture_view.coherency
 * Interface mismatch in KHR-GL45.enhanced_layouts.xfb_block_stride
 * Format mismatch between bound texture and image layout specifier
 * Max compiler threads only specifies a suggestion
 * Exclude texture_view.view_classes
 * Sparse Texture tests not following spec
 * Wrong shaders in glcParallelShaderCompileTests.cpp
 * Fixes wrong MIN_MAP_BUFFER_ALIGNMENT limit
 * This change fixes an issue in BindBufferOffsetEXT
 * Added glViewport(0,0,2,2) call to TestRenderToStencil() and TestBlitStencil() functions
 * Scale cubemap size to fit memory limits
 * Remove references to not existing directory from CMakeLists
 * Fix the precision of shader structure test
 * Fixes copy_tex_image_conversions_forbidden

Vulkan CTS 1.1.2.1

New tests:
 * Tests for VK_KHR_driver_properties
 * Tests for VK_KHR_shader_atomic_int64
 * Tests for VK_KHR_vulkan_memory_model
 * Tests for VK_EXT_inline_uniform_block
 * Tests for GL_EXT_shader_16bit_storage
 * Tests for GL_EXT_shader_8bit_storage
 * Tests for random descriptor set tests

Improvements:
 * Shader validation: respect Vulkan version, relaxed layout requirement

Fixes:
 * Enable Android-O AHB support in Android-P build
 * VK_KHR_incremental_present: Fix usage flags.
 * Fix runtime array with workgroup storage class
 * Fixes AHB image tests feature check
 * Fix timeout in dEQP-VK.wsi.android test
 * Reduces percentage of heap used in memory tests
 * Fix pipeline barrier issues in protected memory tests
 * Fixes WSI protected swapchain allocation tests
 * 2Darray compatible flag not allowed for 2D images
 * Consider rounding mode differences in mipmap tests
 * Test cube image blits with VK_IMAGE_CREATE_CUBE_COMPATIBLE flag.
 * Fix RobustnessVertexAccessInstance bounds check

OpenGL ES CTS 3.2.5.1

Fixes:
 * Use sRGB encoded reference values for 8-bit Display P3 tests
 * Add screenLayout to AndroidManifest configChanges
 * Report NotSupported for GLES1 tests on non-ES1 run
 * Allow reading GL_BYTE surfaces to GL_RGBA output.
 * Fix the memory stomp running *core.internalformat.texture2d.depth_stencil*
 * Add specific RGB10A2 data for BlendHSLHue and BlendHSLSaturation
 * Fix: rendering complete times not monotonic
 * Check the bits of channel when calling glCopyTexImage2D
 * Modify the negative test related to glShaderBinary
 * Wrong statues effects KHR-GLES32.core.internalformat.* tests
 * Throw NotSupported if MAX_FRAGMENT_IMAGE_UNIFORMS = 0
 * Allow glFramebufferTexture2D with level!=0 in ES2 tests with ES3
 * Update transform feedback tests
 * Call eglSwapBuffers() to ensure wayland surface is attached
 * Fix dEQP-GLES3.functional.fbo.multiview.*
 * Loosen threshold in Texture3D tests
 * Some glCopyTexSubImage3D related tests conflict with GLES3.2 spec
 * Fix swap_buffers_with_damage.resize* hang on X11
 * Correct getProcAddress() handling in robustness tests
 * Remove shaders.negative.unused_uniform_precision_matching
 * Shaders Derivate Tests: fix threshold
 * Exclude tests causing OOM
 * Allow glReadPixels() differences due precision errors
 * Robust_buffer_access_behavior.texel_fetch relies on undefined behavior
 * Increase threshold GLES3 blend tests
 * Do not fail if no config is available
 * Get an EGLConfig supporting pbuffers
 * Remove undefined EGL test cases
 * wayland: search for GLESv1 toggle DEQP_SUPPORT_GLES1 on if found
 * generatemipmap_zero_level_array_compressed may reuse stale object
 * Skip BC/DXT formats in Texture3D tests
 * GL_SRG8_ALPHA8 also supported in OpenGL ES 3.0+
 * GLES 2/3: Fix masked color clear tests.

Vulkan CTS 1.1.1.3

Fixes:
 * Fix pipeline barrier issues in protected memory tests
 * 2Darray compatible flag not allowed for 2D images
 * Add screenLayout to AndroidManifest configChanges
 * Add ASTC LDR error colour quality warning
 * Fixes WSI protected swapchain allocation tests
 * Consider rounding mode differences in mipmap tests
 * Test cube image blits with VK_IMAGE_CREATE_CUBE_COMPATIBLE flag.
 * Fix RobustnessVertexAccessInstance bounds check
 * Fixes tessellation triangle check
 * Reduces percentage of heap used in memory tests
 * Fix subpixel precision for line geometry generation
 * Fixes memory allocation tests
 * Incorrect enum values for VK_KHR_create_renderpass2
 * Improve wsi.incremental_present.* for low memory
 * Fix issues in ProtectedMemCopyBufferToImage
 * Fix copy/paste errors in opundef assembly tests
 * Fixes geometry shader invocation dependency
 * Remove unconditional depth_clamping for 1.1.1
 * Fix build when SPIRV-Tools is not present
 * Invalidate memory in draw tests
 * Request queue with sparse support only when needed
 * Don't fail DisplayPlaneProperties2 with no display
 * Fix 8bit uniform buffers arrays strides
 * Fix memory allocation tests
 * Allow two images worth of alignment slop in protected heap
 * Protected memory image access tests not flushing staging buffer
 * gl_PrimitiveIDIn tests have sketchy loop bounds.
 * Drop linear filter requirement from YCbCr tests
 * Compile shaders for 1.0-compatible tests as SPIRV 1.0, not max supported
 * Memory binding tests not invaidating host memory before results check.
 * Require extension for the deviceFeatures2 test

Vulkan CTS 1.1.2.0

New tests:
 * New tests for pointers as function parameters
 * Add tests for vkCmdClearColorImage on linear images
 * Cross-stage interface tests
 * Added test for unused framebuffer attachment
 * Added tests for incompatible render pass
 * Test discontiguous descriptor binding values
 * Add extra multiview tests
 * Add more component qualifier array tests for fragment outputs
 * Add OpVariable initialization tests
 * Added test for destroying pipeline object early
 * Added image blit tests with multi-part image lock
 * Add API info tests for YCbCr formats
 * Add tests for max fragment input component limit
 * Add AHardwareBuffer external memory test for format/usage flags.
 * Add tests for packed vertex formats
 * Added tests for rasterizer discard
 * Add tests for compute pipeline derivatives
 * Test cross invocation communication
 * Test omitting blend state with unused attachments
 * Add test for VK_EXT_shader_stencil_export
 * New tests for VK_NV_shader_subgroup_partitioned
 * Add SPIR-V tests for 16, 32 and 64-bit integers
 * Test copying data from/to tightly sized buffers
 * Add tests for render targets with write masks
 * Add test for dynamically indexing output variable components
 * Add AHardwareBuffer support to external memory tests
 * Add 16bit storage access chain tests
 * Add tests for 16-bit OpSwitch
 * Add SPIR-V tests for return values of atomic functions
 * Test using gl_Layer in Vertex and Tessellation shaders
 * Test OpTypeImage with different depth property values
 * Test copying data between compressed and uncompressed images
 * Add 16bit to 16bit StorageInputOutput16 tests
 * Fix image validation errors in SPIR-V tests
 * Add operation tests for 16-bit integers
 * Test using gl_ViewportIndex in tessellation shader
 * Add tests for freeing descriptor sets followed by pool reset
 * Add tests for checking YCbCr format enums in Vulkan 1.1
 * Test viewport with zero height
 * Alignment tests of 16bit types in structs
 * Add missing SConvert, FConvert and UConvert tests
 * Create Compute tests for OpName
 * Create Graphics tests for OpName
 * Test updating set with an empty binding
 * Test discontiguous descriptor sets
 * Test creating device when out of host memory
 * Add operation tests for 64-bit integers
 * Extend some of the 16bit storage tests
 * Add sparse image support for pipeline multisample tests.
 * Extend multiview tests
 * Add test for specific depth stencil state
 * Test texture gather with separate sparse binding queue
 * Add tests for sparse compressed textures.
 * Subgroup tests: Using framebuffer in tess geometry tests
 * Add timestamp tests that use two command buffers
 * Test using a pipeline whose render pass has been destroyed
 * Add descriptor set layout lifetime tests
 * Depth/stencil multiview tests
 * Test for return values of vkGet  *ProcAddr
 * Test QueueSubmit with null SubmitInfo and fence
 * Add tests for VK_EXT_post_depth_coverage
 * Add tests for OpSwitch operator
 * Test using gl_ViewportIndex in a vertex shader
 * Test waiting for fences without waitAll
 * Test push constant ranges with overlapping shader stages
 * Fix test with invalid viewport size
 * Test pipeline layout lifetime
 * Add pipeline cache corner cases
 * Add test for VkSetHdrMetadata
 * Tests for layered resolve attachments
 * Test creating empty descriptor set layouts
 * Extend coverage in subgroups: vote, builtinMaskVar
 * Add tests for vkGetDeviceQueue2
 * Robust buffer access through variable pointers
 * Added pipeline creation test matching a crashing layer test
 * Extend combined image sampler tests
 * Add tests for texture *ClampARB and sparseTexture *ClampARB.

 Improvements:
 * Added shader cache to speed up runtime
 * Improve wsi.incremental_present.* for low memory
 * Ensure that SPIR-V Headers are always present
 * Use NDK's CMake toolchain for NDK r15+
 * Drop support for NDK r11
 * Add option to embed validation layers in Android package
 * Log SPIR-V dissasembly when validation fails
 * Compile shaders for 1.0-compatible tests as SPIRV 1.0, not max supported
 * Enable shader cache by default
 * MacOS WSI test support
 * Integrated SPIRV-Tools-opt to conformance test
 * AHardwareBuffer make the system API dynamic
 * Add more context for SPIR-V build log outputs.
 * Use utility function for VkViewport and VkRect2D creation
 * Use common utility function for render pass creation
 * Use common utility function for graphics pipeline creation
 * Use common utility function for render pass begin and end
 * Use common utility function for command buffer begin and end
 * Use common utility function for queue submit and wait
 * Added optimization avoiding shader disassembly

Fixes:
 * Remove unconditional depth_clamping
 * Vulkan AHB - fix incorrect structure definition
 * Fix copy/paste errors in opundef assembly tests
 * Remove broken/redundant barrier for multiview depth/stencil readback
 * Fixes geometry shader invocation dependency
 * Removed cache info from logged shaderstrings
 * Incorrect pipeline barrier in multiview.multisample.*
 * Fix build when SPIRV-Tools is not present
 * Invalidate memory in draw tests
 * Request queue with sparse support only when needed
 * Don't fail DisplayPlaneProperties2 with no display
 * Fix 8bit uniform buffers arrays strides
 * Fix memory allocation tests
 * Allow two images worth of alignment slop in protected heap
 * Protected memory image access tests not flushing staging buffer
 * Remove an unneded type qualifier
 * gl_PrimitiveIDIn tests have sketchy loop bounds.
 * Drop linear filter requirement from YCbCr tests
 * Fix 8bit storage spir-v tests
 * Allowed device extensions list is missing entry
 * Don't assume sampling is allowed on linear-tiled YCbCr
 * Fix invalid block layouts
 * Use local binding for renderImage layout change
 * Memory binding tests not invaidating host memory before results check.
 * Made glsl loop tests a bit more robust
 * Fix validation errors in synchronization tests
 * Fix multiview mutlisample tests
 * Fix barriers in dEQP-VK.api.fill_and_update_buffer.*
 * Add missing self-dependency in color->input self-dep tests
 * Account for noncoherent atom size in renderpass memory flush
 * Remove spurious TS/GS stage bits from renderpass test barriers
 * Fix image blit mirrored comparisons
 * Remove one copy of the ballot test source
 * Fix barrier and invalidation of the fill_buffer readback
 * Require geometry shader in multisample resolve
 * Added more early-out hardware support checks
 * Fix check for 16-bit extension for 16-bit conversion functions.
 * Fix improper use of vkCmdWaitEvents
 * dEQP-VK.renderpass.sparserendertarget: bind metadata
 * Emit defined positions from tessellation shader in subgroups tests.
 * Use default device for sparse binding tests.
 * Allow fp16 denorm inputs to be flushed to zero.
 * Fix validation errors in draw tests
 * Fix protected memory buffer tests to do proper operation
 * Fix UConvert validation error in mul_sdiv tests
 * Multiview: do not use inline renderpass with secondary command buffers
 * Fix pipeline barrier for clearing multiview attachment
 * Clear all Views of Framebuffer Attachment to reference background color
 * Fix validation errors in image tests
 * Randomize image layout in renderpass attachment allocation test
 * Check ballot requirement in subgroup basic tests
 * Fix logging issues when shader cache is enabled
 * Skip shader compile on unsupported ycbcr tests
 * Require extension for the deviceFeatures2 test
 * Require protectedMemory feature for non-zero queue create flags
 * Fix resource buffers used for the Op?Convert graphics tests
 * Use the subgroups 'supportedCheck' helpers
 * Fix stack overflow on vkGetPhysicalDeviceSparseImageFormatProperties
 * Request proper features in spirv_asembly tests
 * Fix use of sparse binding queue on default context
 * UnpackHalf2x16: consistent subnormal 16bit values
 * Fix validation errors in pipeline tests
 * Fix invalid command pool create flag in image tests
 * Fix invalid memory scope in SPIR-V tests
 * Allow pop-free behavior in line clipping tests
 * Validation errors fix in copying 3d images
 * Don't run sparse texture tests if format isn't sparse supported
 * Fix validation errors in image clearing tests
 * Fix: Remove workaround nonCoherentAtomSize
 * Account for differences in nan with fp16 comparison
 * Fix wrong layout used in vkCmdImageResolve
 * Use framebuffer with a compatible render pass
 * Fix invalid  memory semantics in barrier tests
 * Use default device for spir-v assembly tests
 * Fixed color bias and scale for BC4-5 tests
 * Removed incompatible render pass test
 * Remove max expected query result in pipeline statistics tests
 * Fix subpixel precision for line geometry generation
 * Fix issues in ProtectedMemCopyBufferToImage
 * Handle denorm flush to negative zero in unpackhalf2x16 tests.
 * Fixes tessellation triangle check
 * Reduces percentage of heap used in memory tests

Vulkan CTS 1.1.1.2

New tests:
 * Tests for VK_KHR_8bit_storage
 * Tests for VK_KHR_create_renderpass2

Fixes:
 * Check required peer memory features in sparse resources device group tests
 * Require geometry shader in multisample resolve
 * Fix protected memory buffer tests to do proper operation
 * Fixes ycbcr copy tests to ignore padding
 * Don't rely on unspecified argument evaluation order in atomic swap tests
 * Request required 16bit storage features in UConvert/SConvert tests
 * Added PointSize in shader used for drawing points
 * Fix VkPipelineLayout use after free

OpenGL ES CTS 3.2.5.0

New tests:
 * Tests for precision matching rules
 * Tests for EXT_shader_group_vote
 * Tests for readonly and writeonly buffer variables
 * Tests for npot_tex2d_render
 * Tests for multiple contexts
 * Tests for struct names hiding other struct names
 * Tests for mixing UB/SSB with and without instance names
 * Tests for GL_FRAGMENT_PRECISION_HIGH
 * Tests for negative gl_WorkGroupSize cases
 * Tests for internal formats conversions
 * Tests for packed depth and stencil
 * Tests for OES_texture_3D
 * Tests for packed pixels
 * Tests for CopyImageSubData with integer textures
 * Increased negative GLSL coverage
 * Increased coverage for robustness

Vulkan CTS 1.1.1.1

New tests:
 * Add tests for VK_KHR_draw_indirect_count

Fixes:
 * Limit protected swapchain images based on memory
 * Don't use invalid sType as canary value for arrays
 * Clear image layer 0 in memory commitment test
 * Vulkan: Fix WIN32 NativeHandle importMemory leaks
 * Do not fail plane tests when there is no display
 * Do not fail display tests when there is no display
 * Do not test CLEAR loadOp with read-only depth stencil image layouts
 * Remove 4K min size from dedicated image tests
 * Missing pipeline barriers from fragment variants of dEQP-VK.subgroups.*
 * Request GPDP2 extension if required in shared presentable image tests

OpenGL CTS 4.6.0.1

Fixes:

 * Fix glTransformFeedbackBufferRange defintion
 * Move khr_debug tests to NoContext
 * Fixes for sparse texture clamp tests.
 * SparseTextureAlloc check for GL_NV_deep_texture3D
 * Fix ClearOpsBufferStorageTestCase on ppc64le
 * Fix XCreateWindow BadMatch with depth 32 visuals
 * Fix checks for maximum number of samples
 * Loosen color target tolerances for some tests
 * Remove use of unsupported SURFACETYPE_OFFSCREEN_GENERIC
 * Implementations can use more samples than requested
 * Remove read value check against uncommitted pages
 * Add some missing GL_TEXTURE_SPARSE state setting
 * Fix typo in log message for ClearOpsBufferStorage
 * Validate instancing cases properly

Vulkan CTS 1.1.1.0

First major update since the initial 1.1.0.3 release.
This release adds 26272 new tests in total.

New tests:
 * Add tests for VK_EXT_sampler_filter_minmax
 * Add test to stress semaphore chains
 * Add two-step variants of surface clear tests
 * Add test for sparse render target
 * Add VK_KHR_16bit_storage granularity stress test
 * Add tests for different shader varying vec width
 * Add tests for binding model with multiple descriptor sets
 * Add wide-color tests
 * Add new tests for OpSConvert and OpFConvert
 * Add geometry layered readback tests
 * Add ASTC format to compressed texture test
 * Add test for differing interpolation decorations.
 * Add new tests for VK_KHR_get_display_properties2
 * Add test for branching out of switch block using continue.
 * Add testing for sparse D/S/DS images.
 * Add depth bounds tests
 * Add test for color attachments with different sizes.
 * Add tests for verifying standard image block shape usage.
 * Add tests for VK_EXT_sample_locations
 * Add tests for aligned mip size in sparse images.
 * Add tests to use OpPhi with various variable types
 * Add test for reading samples from previous subpass.
 * Add tests for inverted depth ranges.
 * Add swapchain tests for the VK_KHR_protected_memory extension
 * Add test to stress deeply nested OpPhi
 * Add tests for component qualifier in shaders
 * Add more tests for multiview
 * Add tests for empty render pass
 * Add tests for reading compute shader builtin vars by component
 * Add tests for local workgroup size
 * Add tests for workgroup size setting combinations
 * Add tests for separate and combined images and samplers in SPIR-V

Improvements:
 * Add compressed BC format support for TCU
 * Subgroups tests upgrade: reduce time of execution
 * Subgroup Tests: Using framebuffer instead of SSBO
 * get_physical_device_properties2: add information print

Fixes:
 * Added missing gl_PointSizes when drawing points
 * Emit defined positions from tessellation shader in subgroups tests
 * Validation errors dEQP-VK.synchronization.op.multi_queue
 * Validation errors in dEQP-VK.api.image_clear*
 * Avoid out-of-range depth values
 * Fix instance-rate vertex attribute verification
 * Add missing barrier to texel_view_compatible tests
 * YCbCr image views need a VkSamplerYcbcrConversionInfo
 * Avoid hitting fp16 NaN values
 * Invalid handle in create_display_plane_surface
 * Add StorageImageExtendedFormats cap check
 * Enforce OOB requirements for binding with multiple attributes
 * Pipeline must be bound before calling vkCmdSetViewport/Scissor etc
 * Remove check for depthBounds support in frag depth tests
 * Invalidate memory in draw tests
 * Avoid assert in dEQP-VK.tessellation.fractional_spacing
 * Fix depth & stencil passOp for pipeline.render_to_image tests
 * Fix unreasonable memory expectations in pipeline.render_to_image
 * Fix imageExtent to use correct destination size
 * Fix stageMask (0x1) usage in various tests
 * Relax precision for atan2 built-in precision tests
 * Fix OpCopyObject from U32 to S32
 * Fix PIPELINE_STAGE_TOP_OF_PIPE_BIT usage in api tests
 * Fix incorrect displayMode for wsi.display test
 * Fix non-unique type declarations in tests
 * Fixed invalid OpPtrAccessChain usage in access chain tests.
 * Fix unreasonable memory expectations in pipeline.render_to_image
 * Fix memory corruption vktApiObjectManagementTests
 * Fix possibility of memory leak in subgroups tests
 * Fix regression in dEQP-VK.pipeline.multisample.sample_mask
 * Modify spirv_assembly tests adding dependency on 16bit_storage extension
 * Do not use depthClampEnable unconditionally.
 * Remove duplicated definitions in spirv_assembly tests
 * Test with depth range greater than 1.0 should require extension
 * Pipeline used in second subpass should set subpass index to 1
 * Verify VkSamplerYcbcrConversionImageFormatPropertiesKHR is properly filled
 * Use fill data that varies within 4 bytes boundaries
 * Check viewport limits against framebuffer limits
 * Fixed rounding issue in spirv_assembly tests
 * Do not fail display tests when there is no display

Vulkan CTS 1.1.0.3

First major Vulkan 1.1 CTS release. Adds ~44k unique test
cases that cover Vulkan 1.1 and Vulkan 1.0.
Improves the framework to handle both Vulkan 1.1 and Vulkan 1.0
implementations.

New tests for extensions:
 * Tests for VK_KHR_maintenance3
 * Tests for VK_KHR_device_group and VK_KHR_device_group_creation
 * Tests for VK_KHR_display

Other new tests:
 * Tests for shader subgroup operations
 * Tests for protected content
 * Tests for multiview
 * Tests for generating mipmaps with vkCmdBlitImage()
 * Tests for getDeviceMemoryCommitment
 * Tests for OpSMod/OpSRem
 * Tests for NEAREST filtered blitting
 * Tests for resolving between images of different sizes
 * Tests for Multisample Shader BuiltIn
 * Tests for occlusion query with fragment discard
 * Test for maxVertexInputAttributes
 * Tests for NMin, NMax and NClamp instructions
 * Tests for vkCmdClearColorImage with VK_REMAINING_ARRAY_LAYERS
 * Tests for OpAtomic* in graphics pipeline
 * Tests for FragDepth shader built-in

Improvements:
 * HLSL compilation support
 * Refactor large test groups
 * Refactor time consuming initializers

Removed tests:
 * OpUConvert with signed integer destination