Merge branch 'main' into ue4-nastys-patch-env

Common/MVKCommonEnvironment.h

@@ -88,6 +88,10 @@ extern "C" {
 #endif
 
 /** Building with Xcode versions. iOS version also covers tvOS. */
+#ifndef MVK_XCODE_15
+#   define MVK_XCODE_15             ((__MAC_OS_X_VERSION_MAX_ALLOWED >= 140000) || \
+                                    (__IPHONE_OS_VERSION_MAX_ALLOWED >= 170000))
+#endif
 #ifndef MVK_XCODE_14_3
 #	define MVK_XCODE_14_3			((__MAC_OS_X_VERSION_MAX_ALLOWED >= 130300) || \
 									(__IPHONE_OS_VERSION_MAX_ALLOWED >= 160400))

Common/MVKOSExtensions.h

@@ -84,6 +84,9 @@ double mvkGetTimestampPeriod();
  */
 double mvkGetElapsedMilliseconds(uint64_t startTimestamp = 0, uint64_t endTimestamp = 0);
 
+/** Returns the current absolute time in nanoseconds. */
+uint64_t mvkGetAbsoluteTime();
+
 /** Ensures the block is executed on the main thread. */
 void mvkDispatchToMainAndWait(dispatch_block_t block);
 
@@ -164,3 +167,9 @@ uint64_t mvkGetUsedMemorySize();
 
 /** Returns the size of a page of host memory on this platform. */
 uint64_t mvkGetHostMemoryPageSize();
+
+#pragma mark -
+#pragma mark Threading
+
+/** Returns the amount of avaliable CPU cores. */
+uint32_t mvkGetAvaliableCPUCores();

Common/MVKOSExtensions.mm

@@ -40,6 +40,7 @@ MVKOSVersion mvkOSVersion() {
 
 static uint64_t _mvkTimestampBase;
 static double _mvkTimestampPeriod;
+static mach_timebase_info_data_t _mvkMachTimebase;
 
 uint64_t mvkGetTimestamp() { return mach_absolute_time() - _mvkTimestampBase; }
 
@@ -50,6 +51,8 @@ MVKOSVersion mvkOSVersion() {
 	return (double)(endTimestamp - startTimestamp) * _mvkTimestampPeriod / 1e6;
 }
 
+uint64_t mvkGetAbsoluteTime() { return mach_continuous_time() * _mvkMachTimebase.numer / _mvkMachTimebase.denom; }
+
 // Initialize timestamping capabilities on app startup.
 //Called automatically when the framework is loaded and initialized.
 static bool _mvkTimestampsInitialized = false;
@@ -58,9 +61,8 @@ MVKOSVersion mvkOSVersion() {
 	_mvkTimestampsInitialized = true;
 
 	_mvkTimestampBase = mach_absolute_time();
-	mach_timebase_info_data_t timebase;
-	mach_timebase_info(&timebase);
-	_mvkTimestampPeriod = (double)timebase.numer / (double)timebase.denom;
+	mach_timebase_info(&_mvkMachTimebase);
+	_mvkTimestampPeriod = (double)_mvkMachTimebase.numer / (double)_mvkMachTimebase.denom;
 }
 
 void mvkDispatchToMainAndWait(dispatch_block_t block) {
@@ -138,3 +140,10 @@ uint64_t mvkGetUsedMemorySize() {
 
 uint64_t mvkGetHostMemoryPageSize() { return sysconf(_SC_PAGESIZE); }
 
+#pragma mark -
+#pragma mark Threading
+
+/** Returns the amount of avaliable CPU cores. */
+uint32_t mvkGetAvaliableCPUCores() {
+    return (uint32_t)[[NSProcessInfo processInfo] activeProcessorCount];
+}

Docs/MoltenVK_Runtime_UserGuide.md

@@ -341,6 +341,7 @@ In addition to core *Vulkan* functionality, **MoltenVK**  also supports the foll
 - `VK_KHR_shader_draw_parameters`
 - `VK_KHR_shader_float_controls`
 - `VK_KHR_shader_float16_int8`
+- `VK_KHR_shader_non_semantic_info`
 - `VK_KHR_shader_subgroup_extended_types` *(requires Metal 2.1 on Mac or Metal 2.2 and Apple family 4 on iOS)*
 - `VK_KHR_spirv_1_4`
 - `VK_KHR_storage_buffer_storage_class`
@@ -350,7 +351,9 @@ In addition to core *Vulkan* functionality, **MoltenVK**  also supports the foll
 - `VK_KHR_timeline_semaphore`
 - `VK_KHR_uniform_buffer_standard_layout`
 - `VK_KHR_variable_pointers`
+- `VK_EXT_4444_formats` *(requires 16-bit formats and either native texture swizzling or manual swizzling to be enabled)*
 - `VK_EXT_buffer_device_address` *(requires GPU Tier 2 argument buffers support)*
+- `VK_EXT_calibrated_timestamps` *(requires Metal 2.2)*
 - `VK_EXT_debug_marker`
 - `VK_EXT_debug_report`
 - `VK_EXT_debug_utils`
@@ -375,6 +378,8 @@ In addition to core *Vulkan* functionality, **MoltenVK**  also supports the foll
 - `VK_EXT_shader_atomic_float` *(requires Metal 3.0)*
 - `VK_EXT_shader_demote_to_helper_invocation` *(requires Metal Shading Language 2.3)*
 - `VK_EXT_shader_stencil_export` *(requires Mac GPU family 2 or iOS GPU family 5)*
+- `VK_EXT_shader_subgroup_ballot` *(requires Mac GPU family 2 or Apple GPU family 4)*
+- `VK_EXT_shader_subgroup_vote` *(requires Mac GPU family 2 or Apple GPU family 4)*
 - `VK_EXT_shader_viewport_index_layer`
 - `VK_EXT_subgroup_size_control` *(requires Metal 2.1 on Mac or Metal 2.2 and Apple family 4 on iOS)*
 - `VK_EXT_surface_maintenance1`

Docs/Whats_New.md

@@ -19,7 +19,13 @@ MoltenVK 1.2.5
 Released TBD
 
 - Add support for extensions:
+	- `VK_KHR_shader_non_semantic_info`
+	- `VK_EXT_4444_formats`
+	- `VK_EXT_calibrated_timestamps`
 	- `VK_EXT_shader_demote_to_helper_invocation`
+	- `VK_EXT_shader_subgroup_ballot`
+	- `VK_EXT_shader_subgroup_vote`
+- Add support for `VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN`.
 - Ensure non-dispatch compute commands don't interfere with compute encoding state used by dispatch commands.
 - Support `VK_PRESENT_MODE_IMMEDIATE_KHR` if `VkPresentTimeGOOGLE::desiredPresentTime` is zero.
 - Support maximizing the concurrent executing compilation tasks via `MVKConfiguration::shouldMaximizeConcurrentCompilation`

ExternalRevisions/SPIRV-Cross_repo_revision

@@ -1 +1 @@
-55750be7886a96008b964e75e1eb4a5a6c369a2a
+aafcc207ea82308722124db2575aa95f42cb99c9

MoltenVK/MoltenVK/API/mvk_datatypes.h

@@ -227,6 +227,15 @@ MTLTextureSwizzle mvkMTLTextureSwizzleFromVkComponentSwizzle(VkComponentSwizzle
 /** Returns all four Metal texture swizzles from the Vulkan component mapping. */
 MTLTextureSwizzleChannels mvkMTLTextureSwizzleChannelsFromVkComponentMapping(VkComponentMapping vkMapping);
 
+/** Maps a clear color according to the specified VkComponentSwizzle. */
+float mvkVkClearColorFloatValueFromVkComponentSwizzle(float *colors, uint32_t index, VkComponentSwizzle vkSwizzle);
+
+/** Maps a clear color according to the specified VkComponentSwizzle. */
+uint32_t mvkVkClearColorUIntValueFromVkComponentSwizzle(uint32_t *colors, uint32_t index, VkComponentSwizzle vkSwizzle);
+
+/** Maps a clear color according to the specified VkComponentSwizzle. */
+int32_t mvkVkClearColorIntValueFromVkComponentSwizzle(int32_t *colors, uint32_t index, VkComponentSwizzle vkSwizzle);
+
 
 #pragma mark Mipmaps
 

MoltenVK/MoltenVK/Commands/MVKCmdDraw.h

@@ -91,6 +91,7 @@ class MVKCmdDraw : public MVKCommand {
 						uint32_t firstInstance);
 
     void encode(MVKCommandEncoder* cmdEncoder) override;
+	void encodeIndexedIndirect(MVKCommandEncoder* cmdEncoder);
 
 protected:
 	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
@@ -120,6 +121,7 @@ class MVKCmdDrawIndexed : public MVKCommand {
 
 protected:
 	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
+	void encodeIndexedIndirect(MVKCommandEncoder* cmdEncoder);
 
 	uint32_t _firstIndex;
 	uint32_t _indexCount;
@@ -146,6 +148,7 @@ class MVKCmdDrawIndirect : public MVKCommand {
 
 protected:
 	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
+	void encodeIndexedIndirect(MVKCommandEncoder* cmdEncoder);
 
 	id<MTLBuffer> _mtlIndirectBuffer;
 	VkDeviceSize _mtlIndirectBufferOffset;
@@ -167,7 +170,15 @@ class MVKCmdDrawIndexedIndirect : public MVKCommand {
 						uint32_t count,
 						uint32_t stride);
 
+	VkResult setContent(MVKCommandBuffer* cmdBuff,
+						id<MTLBuffer> indirectMTLBuff,
+						VkDeviceSize indirectMTLBuffOffset,
+						uint32_t drawCount,
+						uint32_t stride,
+						uint32_t directCmdFirstInstance);
+
 	void encode(MVKCommandEncoder* cmdEncoder) override;
+	void encode(MVKCommandEncoder* cmdEncoder, const MVKIndexMTLBufferBinding& ibbOrig);
 
 protected:
 	MVKCommandTypePool<MVKCommand>* getTypePool(MVKCommandPool* cmdPool) override;
@@ -176,4 +187,5 @@ class MVKCmdDrawIndexedIndirect : public MVKCommand {
 	VkDeviceSize _mtlIndirectBufferOffset;
 	uint32_t _mtlIndirectBufferStride;
 	uint32_t _drawCount;
+	uint32_t _directCmdFirstInstance;
 };