GPU SpMV Optimization — Literature Review

This section summarizes the key papers that inform our GPU kernel design. Each summary includes the core contribution, mathematical framework, and relevance to our implementation.

• Paper Summaries
  • 1. Bell & Garland (SC ‘09) — Foundational CSR SpMV on GPUs
  • 2. Greathouse & Daga (SC ‘14) — CSR-Adaptive
  • 3. Liu & Vinter (ICS ‘15) — CSR5
  • 4. Chu et al. (HPDC ‘23) — Ampere-Aware Optimization
  • 5. Merrill & Garland (SC ‘16) — Merge-Based SpMV
  • 6. Niu et al. (IPDPS 2021) — Tiled SpMV
  • 7. Gao et al. (2024) — Systematic Literature Survey
  • Summary Table: Papers vs. Optimizations
  • References
• GPU Kernels
  • 1. v1 — Basic CSR Row-Parallel Kernel
  • 2. v2 — Optimized Kernel
  • 3. Performance Results
  • References

Indices and tables

• Index

• Module Index

• Search Page