🇨🇳 简体中文

📋 Changelog

All notable changes to the GPU SpMV project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.0.0] - 2025-04-16

🎉 First Stable Release

✨ Added

Core Features

  • Full CSR (Compressed Sparse Row) sparse matrix format support
  • Full ELL (ELLPACK) sparse matrix format support
  • Four CUDA kernels: Scalar CSR, Vector CSR, Merge Path, ELL
  • Automatic kernel selection based on matrix statistics
  • Texture cache support with execution context reuse

Performance & Testing

  • Bandwidth metrics calculation
  • Benchmarking framework
  • GPU vs CPU performance comparison

Applications

  • PageRank graph algorithm implementation

Engineering Quality

  • RAII resource management (CudaBuffer, CudaTimer, SpMVExecutionContext)
  • Semantic error code system
  • Comprehensive Google Test suite
  • CMake Presets build configuration
  • GitHub Actions CI/CD

🔒 Security

  • Integer overflow protection
  • Memory safety checks

🚀 Performance

  • ELL Column-major coalesced memory access
  • Warp-level shuffle reduction
  • Load balancing optimization (Merge Path)

[0.1.0] - 2025-03-01

🚀 Initial Release

  • Basic project structure
  • Basic CSR matrix implementation
  • Simple SpMV GPU kernel
  • CMake build configuration

Version History

Version Date Status Highlights
1.0.0 2025-04-16 Stable First stable release, complete feature set
0.1.0 2025-03-01 Archived Initial prototype

Migration Guide

Upgrading to 1.0.0

No breaking changes from pre-release versions.

  1. Use Named Constants 
    1
2
3
4
5

    // Old
config.block_size = 256;
   
// New
config.block_size = spmv::DEFAULT_BLOCK_SIZE;
  2. Use Execution Context for Reuse 
    1
2
3
4
5
6
7
8
9
10

    // Old
for (int i = 0; i < iterations; i++) {
    spmv_csr(csr, d_x, d_y, &config, cols);
}
   
// New
SpMVExecutionContext context;
for (int i = 0; i < iterations; i++) {
    spmv_csr(csr, d_x, d_y, &config, cols, &context);
}

Future Roadmap

Planned [1.1.0]

  • COO format support
  • Hybrid CSR/ELL format
  • Multi-GPU support
  • Batched SpMV operations

Under Consideration

  • BFloat16 precision support
  • Automatic format selection tuning
  • Python bindings
**[← Performance](performance.en)** · **[ 🏠 Home →](index.en)**