<MPSImageAllocator >(3) MetalPerformanceShaders.framework <MPSImageAllocator >(3)

<MPSImageAllocator >

#import <MPSImage.h>

Inherits <NSObject>, and <NSSecureCoding>.


(MPSImage *__nonnull) - imageForCommandBuffer:imageDescriptor:kernel:
(MPSImageBatch *__nonnull) - imageBatchForCommandBuffer:imageDescriptor:kernel:count:

A class that allocates new MPSImage or MPSTemporaryImage Sometimes it is prohibitively costly for MPS to figure out how big an image should be in advance. In addition, you may want to have some say over whether the image is a temporary image or not. In such circumstances, the MPSImageAllocator is used to provide the developer with an opportunity for just in time feedback about how the image should be allocated.

Two standard MPSImageAllocators are provided: MPSImageDefaultAllocator and MPSTemporaryImageDefaultAllocator. You may of course provide your own allocator instead.

Example:

// Note: MPSImageDefaultAllocator is already provided
//       by the framework under that name.  It is provided here
//       as sample code for writing your own variant.
-(MPSImage * __nonnull)  imageForCommandBuffer: (__nonnull id <MTLCommandBuffer>) cmdBuf

imageDescriptor: (MPSImageDescriptor * __nonnull) descriptor
kernel: (MPSKernel * __nonnull) kernel {
MPSImage * result = [[MPSImage alloc] initWithDevice: cmdBuf.device
imageDescriptor: descriptor ];
// make sure the object sticks around at least as lomg as the command buffer
[result retain];
[cmdBuf addCompletedHandler: ^(id <MTLCommandBuffer> c){[result release];}];
// return autoreleased result
return [result autorelease]; }; -(BOOL) supportsSecureCoding{ return YES; } -(void)encodeWithCoder:(NSCoder * __nonnull)aCoder {
[super encodeWithCoder: aCoder];
// encode any data owned by the class at this level } -(nullable instancetype) initWithCoder: (NSCoder*__nonnull) aDecoder {
self = [super initWithCoder: aDecoder ];
if( nil == self )
return self;
// use coder to load any extra data kept by this object here
return self; }

Please see [MPSImage defaultAllocator] and [MPSTemporaryImage defaultAllocator] for implentations of the protocol already provided by MPS.

When considering whether to write your own MPSImageAllocator, you should know the existing MPSImage and MPSTemporaryImage default allocators are optimized to make image batch allocation much faster than one MPSImage at a time in a loop. When possible, it can be better to use the MPS provided allocators and override the behavior in a padding policy instead, if the changes can be contained in the MPSImageDescriptor. This will help reduce CPU encode time. However, custom padding policies can inhibit optimizations in the MPSNNGraph, particularly node fusion, resulting in more work for the GPU. In cases where the custom padding method does not change filter properties but only adjusts the result image (e.g. adjust result feature channel format) then MPSNNPaddingMethodCustomWhitelistForNodeFusion may be used to signal that node fusion is acceptable.

- (MPSImageBatch * __nonnull MPSImageAllocator) imageBatchForCommandBuffer: (__nonnull id< MTLCommandBuffer >) cmdBuf(MPSImageDescriptor *__nonnull) descriptor(MPSKernel *__nonnull) kernel(NSUInteger) count [optional]

Efficiently create an array of MPSImages with a common descriptor See class description for sample implementation

Parameters:

cmdBuf The MTLCommandBuffer on which the image will be initialized. cmdBuf.device encodes the MTLDevice.
descriptor A MPSImageDescriptor containing the image format to use. This format is the result of your MPSPadding policy.
kernel The kernel that will overwrite the image returned by the filter. Note that the MPS implementations of this protocol don't need this field. It is provided for your convenience.
count The number of images in the batch

Returns:

A valid MPSImage or MPSTemporaryImage. It will be automatically released when the command buffer completes.

- (MPSImage * __nonnull MPSImageAllocator) imageForCommandBuffer: (__nonnull id< MTLCommandBuffer >) cmdBuf(MPSImageDescriptor *__nonnull) descriptor(MPSKernel *__nonnull) kernel [required]

Create a new MPSImage See class description for sample implementation

Parameters:

cmdBuf The MTLCommandBuffer on which the image will be initialized. cmdBuf.device encodes the MTLDevice.
descriptor A MPSImageDescriptor containing the image format to use. This format is the result of your MPSPadding policy.
kernel The kernel that will overwrite the image returned by the filter. Note that the MPS implementations of this protocol don't need this field. It is provided for your convenience.

Returns:

A valid MPSImage or MPSTemporaryImage. It will be automatically released when the command buffer completes.

Generated automatically by Doxygen for MetalPerformanceShaders.framework from the source code.

Mon Jul 9 2018 Version MetalPerformanceShaders-119.3