MetalPerformanceShadersGraph iOS xcode26.4 b2

#MetalPerformanceShadersGraph.framework

diff -ruN /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraph.h /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraph.h
--- /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraph.h	2025-11-09 05:21:48
+++ /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraph.h	2026-02-16 20:12:02
@@ -65,6 +65,8 @@
     MPSGraphReducedPrecisionFastMathNone                                               MPS_SWIFT_NAME(none) = 0,
     /// Execute winograd transform intermediate as FP16.
     MPSGraphReducedPrecisionFastMathAllowFP16Conv2DWinogradTransformIntermediate                            = 1 << 1,
+    /// Allow conversion of operands to FP19 or TF32 from FP32 when needed by dropping 13 mantissa bits.
+    MPSGraphReducedPrecisionFastMathAllowConvertingOperandsFromFP32ToFP19                                   = 1 << 2,
     /// Curated list allowing intermediates for multi-pass GPU kernels to be FP16.
     MPSGraphReducedPrecisionFastMathAllowFP16Intermediates                                                  = MPSGraphReducedPrecisionFastMathAllowFP16Conv2DWinogradTransformIntermediate,
     /// Default selection.
@@ -114,6 +116,9 @@
 
 /// Turns off type inference and relies on type inference during runtime.
 -(void) disableTypeInference;
+
+/// Turns off Automatic Layout Conversion (for conv like operations) for GPU.
+-(void) disableAutoLayoutConversion MPS_AVAILABLE_STARTING(macos(26.3), ios(26.3), tvos(26.3));
 
 /// The optimization level for the graph execution, default is MPSGraphOptimizationLevel1.
 @property (readwrite, nonatomic) MPSGraphOptimization optimizationLevel MPS_AVAILABLE_STARTING(macos(12.3), ios(15.4), tvos(15.4));
diff -ruN /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphArithmeticOps.h /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphArithmeticOps.h
--- /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphArithmeticOps.h	2025-11-09 04:55:39
+++ /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphArithmeticOps.h	2026-02-17 23:31:03
@@ -1027,6 +1027,33 @@
 MPS_AVAILABLE_STARTING(macos(14.0), ios(17.0), tvos(17.0))
 MPS_SWIFT_NAME( complexTensor(realTensor:imaginaryTensor:name:) );
 
+/// Returns a real-valued tensor from a complex-valued tensor with real and imaginary planes separated.
+///
+/// Creates a complexToPlanar operation and returns the result tensor.
+/// In case the input tensor is a complex tensor the result tensor is twice as wide
+/// as the input tensor in the last dimension, and its datatype will be the underlying datatype of
+/// the input tensor - for example `<3xcomplex<f16>>` becomes `<6xf16>`.
+/// In case the input is not complex-valued, this op simply returns the input tensor.
+/// For complex input:
+/// ```md
+/// { resultTensor[...,i] = realPart(inputTensor[...,i])
+/// { resultTensor[...,i+DimSize(inputTensor,-1)] = imagPart(inputTensor[...,i])
+/// ```
+/// For real-valued input:
+/// ```md
+/// resultTensor = inputTensor
+/// ```
+///
+/// - Parameters:
+///   - tensor: The input tensor.
+///   - name: An optional string which serves as an identifier for the operation..
+/// - Returns: A valid `MPSGraphTensor` object containing the elementwise result of the applied operation.
+-(MPSGraphTensor *) planarTensorWithComplexTensor:(MPSGraphTensor *) tensor
+                                             name:(NSString * _Nullable) name
+MPS_AVAILABLE_STARTING(macos(26.3), ios(26.3), tvos(26.3))
+MPS_SWIFT_NAME( planarTensor(tensor:name:) );
+
+
 @end
 
 NS_ASSUME_NONNULL_END
diff -ruN /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphConvolutionOps.h /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphConvolutionOps.h
--- /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphConvolutionOps.h	2025-11-09 05:16:41
+++ /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphConvolutionOps.h	2026-02-16 20:12:01
@@ -333,7 +333,7 @@
 /// - Parameters:
 ///   - incomingGradient: Incoming loss gradient tensor
 ///   - weights: Forward pass weights tensor
-///   - outputShapeTensor: 4D Int32 or Int64 tensor. Shape of the forward pass source tensor
+///   - outputShapeTensor: 4-element 1D Int32 or Int64 tensor. Shape of the forward pass source tensor
 ///   - forwardConvolutionDescriptor: Forward convolution 2D op ``descriptor``
 ///   - name: The name for the operation.
 /// - Returns: A valid MPSGraphTensor object
diff -ruN /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphCore.h /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphCore.h
--- /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphCore.h	2025-11-09 04:24:50
+++ /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphCore.h	2026-02-16 20:12:01
@@ -14,7 +14,9 @@
 
 NS_ASSUME_NONNULL_BEGIN
 
+#if !MPSGRAPH_TOOLCHAIN_BUILD
 MPS_CLASS_AVAILABLE_STARTING(macos(14.0), ios(17.0), tvos(17.0))
+#endif
 /// The common base class for all Metal Performance Shaders Graph objects.
 ///
 /// Only the child classes should be used.
diff -ruN /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphExecutable.h /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphExecutable.h
--- /Applications/Xcode_26.3.0.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphExecutable.h	2025-11-09 05:16:41
+++ /Applications/Xcode_26.4.0-beta2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/System/Library/Frameworks/MetalPerformanceShadersGraph.framework/Headers/MPSGraphExecutable.h	2026-02-16 19:30:55
@@ -131,7 +131,7 @@
 /// Use this method to choose when specialization happens, else it occurs at encode time automatically.
 ///
 /// - Parameters:
-///   - device:Ooptional MPSGraph device to compile with.
+///   - device:Optional MPSGraph device to compile with.
 ///   - inputTypes: Input types expected to be passed to the executable.
 ///   - compilationDescriptor: Compilation descriptor to be used to specialize, since the executable was created with a compilationDescriptor already this one overrides those settings to the extent it can.
 -(void) specializeWithDevice:(MPSGraphDevice * _Nullable) device