MVKCmdCopyImage uses texture view if formats different. Enable
MTLTextureUsagePixelFormatView for VK_IMAGE_USAGE_TRANSFER_SRC_BIT.
Refactor mvkMTLTextureUsageFromVkImageUsageFlags() to clarify Metal usages
and enable views on depth-stencil attachments by default.
Retain MVKSwapchainImage before passing to MTLCommandBuffer completion handler.
Redesign MVKVulkanAPIObject retain/release functionality using atomic ref count
instead of mutex locks, to simplify design and improve performance.
Refactor SPIRVToMSLConverterContext into distinct SPIRVToMSLConversionConfiguration
and SPIRVToMSLConversionResults for conversion input and output, respectively.
Update to latest SPIRV-Cross version.
Update What's New document.
On iOS GPU family 2 and earlier:
- Allow immutable samplers to be passed to SPIRV-Cross in a format
they can be output as hardcoded constexpr samplers in MSL.
- Validate depth compare samplers are only used as immutable.
MVKMVKPipelineLayout & MVKDescriptorSetLayout can take null MVKCommandEncoder
to perform a validation pass when binding or pushing.
Reduce redundancy by combining SPIRVToMSLConverterContext structs
with compatible SPIRV-Cross structs.
Update MoltenVK version to 1.0.36.
vkCmdClearAttachments(): Return encoder to previous pipeline, depth-stencil & resource state after execution.
Fix issue clearing stencil attachment via renderpass when depth attachment is not being cleared.
MVKCmdClearAttachment sets the stencil reference value to 0 and the stencil
reference value is not properly restored for subsequent draw calls.
I'm affraid there might be more similar bugs in MoltenVK.
MVKPipeline records whether Metal pipelines are successfully compiled.
Pipeline and draw commands check for valid Metal pipeline state before encoding.
Compute compilation operations can have custom names.
Name Tessellation control stage compute compiler for logging.
Set MTLFunction label immediately after creation from MVKShaderModule.
Remove _shaderModule reference from MVKShaderLibrary & MVKShaderLibraryCache.
No-op propogateDebugName() function in MVKShaderModule & MVKPipelineCache.
Dispatch messenger callbacks on any reported message or error.
Add MVKDebugUtilsMessenger class.
Add MVKVulkanAPIObject::getVkObjectType() function and override in all Vulkan API objects.
MVKDevice::getMTLPixelFormatFromVkFormat() take base object for reporting.
Instead, just avoid calling into Metal. It's perfectly valid to specify
no viewports--for example, if rasterization is disabled. Perhaps, then,
it'd be better to assert that rasterization is disabled when no
viewports are given.
vkCmdDebugMarkerBegin() and vkCmdDebugMarkerEnd() attach debug
groups to render encoder if available, or command buffer if not.
Name MTLRenderCommandEncoder first after renderpass,
if named, then command buffer, if named, then usage name.
Add MVKCommandEncoder::endRenderpass() to clean up renderpass components.
MVKShaderLibrary and MVKShaderLibraryCache track shader module to
retrieve name for MTLLibrary and MTLFunction, even when cached offline.
All external library header references consistently include framework references.
Cleanup references to external library headers that are no longer required.
Simplify and consolidate external library header paths in Xcode projects.
Add MVK_EXCLUDE_SPIRV_TOOLS build option to avoid use of SPIRV-Tools library.
Remove all other references to headers within SPIRV-Tools library.
Buffers in constant address space must be aligned to 256 bytes on macOS.
This fixes Metal validation errors:
Compute Function(cmdCopyBufferToImage3DDecompressDXTn): the offset into the buffer src that is bound at buffer index 0 must be a multiple of 256 but was set to 12928.
Compute Function(cmdCopyBufferToImage3DDecompressDXTn): the offset into the buffer src that is bound at buffer index 0 must be a multiple of 256 but was set to 78464.
Move MVKVulkanAPIObject to its own .h/mm files.
Add MVKCmdDebug.h/mm files.
Change extension on MVKExtensions.cpp and MVKBaseObject.cpp to .mm.
Remove unused command use in MVKQueue submit() and waitIdle() functions.
MVKCommandPool constructor use default isPooling value in MVKCommandTypePool constructors.
MVKSwapchainImage pass image index in constructor.
Only set MTLRenderPassDescriptor layered rendering properties if layered rendering
is supported and the framebuffer really has multiple layers.
MVKCmdClearImage don't set renderTargetArrayLength, since layers are cleared individually.
Update highest available MTLFeatureSets for use in pipelineCacheUUID.
Add MVKPhysicalDeviceMetalFeatures::multisampleLayeredRendering feature.
MVKCmdClearAttachments shader don't include [[render_target_array_index]] if multisample
and multisampleLayeredRendering is false.
Refactor validation of VkImageCreateInfo when creating an MVKImage.
Validate MVKImage multisample layered array attachments require multisampleLayeredRendering.
Generate derived External/SPIRV-Cross/mvkSpirvCrossRevisionDerived.h
header file from SPIRV-Cross_repo_revision file from within fetchDependencies.
MoltenVK includes derived header file and extracts first part of revision hash
for inclusion in pipelineCacheUUID.
Update highest available MTLFeatureSets for use in pipelineCacheUUID.
Suppress echoing of script directory changes in fetchDependencies.
Support runtime shader compilation from GLSL.
Return VK_ERROR_INVALID_SHADER_NV on shader and pipeline compilation errors.
Add MVKShaderCompilationPerformance::glslToSPRIV to track GLSL conversion performance.
Rename MoltenVKGLSLToSPIRVConverter MVKShaderStage enum to MVKGLSLConversionShaderStage
to avoid naming conflicts with MoltenVK MVKShaderStage enum.
Hologram demo load SPIR-V directly instead of using GLSL through either
MoltenVKGLSLToSPIRVConverter or VK_NV_glsl_shader extension.
Update to latest version of VulkanSamples that supports MVKGLSLConversionShaderStage.
All of the resources' `applyMemoryBarrier()` methods check if the second
scope contains host operations. If they don't, they have no effect. If
there are a lot of resources, iterating them for a method that will
ultimately do nothing is wasteful. Bail out if we can see that they will
do nothing.
While `MTLFence` could be used for synchronization within a single
queue, it doesn't prevent execution across queues. For this, an
`MTLEvent` is required. Yes, this requires Metal 2.1 (macOS 10.14, iOS
12). Older versions will continue to use the old, CPU-based
implementation.
According to the Vulkan spec, semaphores must either be already signaled
or have a signal operation in progress before waiting for them. This
implementation increments the `MTLEvent`'s expected value after encoding
a wait operation, which requires this assumption to hold.
One thorny bit is acquiring a swapchain image when it is not already in
use. In this case, any semaphore provided must be signaled right away.
To accomplish this, a command buffer is executed which performs the
signal operation on the device. Alternatively, we could use an
`MTLSharedEvent`, which would allow us to signal the event on the host.
But, this could have performance implications that may not be
acceptable, just to handle this one case.
Fixes#438.
