Training a Neural Network with Metal Performance Shaders

At a glance

Project structure

Source bundle/
├── MPSTrainingClassifier/
│   ├── main.mm
│   ├── MNISTClassifierGraph.h
│   ├── MNISTClassifierGraph.mm
│   └── Controls.h
├── Common/
│   ├── MNISTDataSet.mm
│   ├── DataSources.h
│   ├── DataSources.mm
│   ├── MNISTDataSet.h
│   ├── Helpers.h
│   └── Helpers.mm
├── Configuration/
│   └── SampleCode.xcconfig
└── MPSTrainingHelloWorld.xcodeproj/
    └── .xcodesamplecode.plist

Structure observations

• Architecturally prominent files are ranked from entry points and role-named declarations; resource-only paths are omitted.
• Primary languages: C/Objective-C header, Objective-C++.
• The verified tree contains 3 project/configuration file(s) and 4 source declaration(s).

Overall architecture

flowchart LR
    N1["main"]
    N2["Metal / MetalPerformanceShaders APIs"]
    N1 --> N2

Reference code

MPSTrainingClassifier/main.mm:156 — architecture anchor

int main(int argc, const char * argv[]) {
    // ...
}

Interpretation

The arrows summarize the source-visible entry, role-named types or folders, and framework direction; when nodes come from structural folders, the sequence is a high-level interpretation rather than proof that every adjacent node calls the next. Ownership is claimed only where the next section cites a stored property or assignment. The diagram is intentionally limited to the dominant path into Metal Performance Shaders.

Ownership and state

classDiagram
    main --> MetalPerformanceShadersAPIs : uses

Ownership evidence

Common/MNISTDataSet.mm:8 — stored dependency or nearest verified ownership anchor

#import "MNISTDataSet.h"

Composition arrows indicate a source-visible construction expression or locally owned value state; aggregation means the owner stores or receives a dependency without proving exclusive lifetime ownership.

Class and protocol design

Common/DataSources.h:34 — representative type boundary

@interface ConvDataSource : NSObject<MPSCNNConvolutionDataSource>{
    // ...
}
// ...
@end    /* ConvDataSource */

No local protocol conformance is claimed as protocol-oriented design; external framework conformances are listed only as dependencies.

Access control

Reference code

Common/DataSources.h:34 — representative boundary

@interface ConvDataSource : NSObject<MPSCNNConvolutionDataSource>{
    // ...
}
// ...
@end    /* ConvDataSource */

Swift declarations without a modifier are internal; explicit private, fileprivate, private(set), public, or open entries above are interpreted by language semantics. Objective-C/C samples instead rely on header and implementation boundaries, which are not equivalent to Swift lexical privacy.

Logic ownership and placement

Design patterns

Naming conventions

• Types: DataSource: ConvDataSource.
• Protocols: no local protocol declaration in the scanned source.
• Methods: gunzippedData, init, getRandomTrainingBatchWithDevice, initWithKernelWidth, dataType, descriptor, weights, biasTerms.
• Files: Common/MNISTDataSet.mm, Common/MNISTDataSet.h, MPSTrainingClassifier/MNISTClassifierGraph.h, MPSTrainingClassifier/MNISTClassifierGraph.mm.

Architecture takeaways

• main is the main source-visible entry or composition anchor for this sample.
• Framework work reaches Metal, MetalPerformanceShaders, random through a deliberately small high-level chain; the detailed API graph remains inside the cited implementation files.
• Stored-property evidence identifies lifecycle collaboration; it does not by itself prove exclusive object ownership.
• Access-control conclusions separate verified language visibility from the likely design rationale.
• The source does not justify labeling the design protocol-oriented.

Source map

• Official Apple documentation