Skip to content

RusTorch v0.6.24 - Neural Engine Integration & Performance Enhancements

Choose a tag to compare

View all tags
@JunSuzukiJapan JunSuzukiJapan released this 28 Sep 05:40
· 19 commits to main since this release
v0.6.24
b8fc933

RusTorch v0.6.24 Release 🚀

🧠 Neural Engine Integration

  • True CoreML Neural Engine Execution: Replaced all placeholder implementations with real Neural Engine operations
  • Advanced MLMultiArray Support: Optimized tensor conversion for Apple Silicon Neural Engine
  • Matrix Operations: Enhanced matmul with Neural Engine acceleration
  • Convolution Operations: Hardware-accelerated Conv2D with Neural Engine backend
  • Activation Functions: Neural Engine optimized activation function execution

⚡ Performance Improvements

  • Metal GPU Convolution: 19% performance improvement with hardware acceleration
  • im2col + GEMM Implementation: Optimized convolution algorithm for GPU execution
  • Hybrid Device Selection: Intelligent Metal/CoreML backend selection with mac-hybrid feature
  • Zero-Copy Optimizations: Reduced memory overhead in tensor operations

🔧 Technical Enhancements

  • Enhanced Error Handling: Improved CoreML error types with detailed conversion errors
  • Compilation Fixes: Resolved Metal kernel compilation issues across platforms
  • WebAssembly Compatibility: Conditional compilation for WASM targets
  • Memory Management: Enhanced buffer management for multi-device operations

📊 Quality Assurance

  • 1139+ Tests Passing: Comprehensive test coverage across all platforms
  • Full CI/CD Validation: Docker builds, cross-platform testing, and quality checks
  • License Compliance: CC0-1.0 support and comprehensive license validation
  • Documentation Updates: Updated all version references and examples

🛠️ Developer Experience

  • Consolidated Benchmarks: Organized performance benchmarks in dedicated workspace
  • Enhanced Debug Output: Improved logging and profiling capabilities
  • Version Consistency: Updated all notebooks and documentation to v0.6.24

📚 Updated Documentation

  • Jupyter Notebooks: All multilingual demos updated with latest version
  • README Examples: Updated Cargo.toml examples and installation instructions
  • Hybrid Computing Docs: Enhanced documentation for f32 unified hybrid systems

🚀 Installation

[dependencies]
rustorch = "0.6.24"

⚡ Quick Start

# Run examples
cargo run --example neural_network_demo --no-default-features
cargo run --example device_performance_comparison --features metal

# Interactive Jupyter
curl -sSL https://raw.githubusercontent.com/JunSuzukiJapan/rustorch/main/scripts/install_jupyter.sh | bash
rustorch-jupyter

Full changelog and technical details available in the repository.

🤖 Generated with Claude Code

Assets 2
Loading