CS-PQ: Cache-Friendly SIMD Product Quantization for Large-Scale ANNS Index Construction
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910254438023168 |
|---|---|
| author | Ma, Y. T. Huang, K. C. Jiang, X. K. Wang, M. L. Yao, X. Chen, R. H. Zhang, G. Shao, Z. L. |
| author_facet | Ma, Y. T. Huang, K. C. Jiang, X. K. Wang, M. L. Yao, X. Chen, R. H. Zhang, G. Shao, Z. L. |
| contents | Product Quantization (PQ) construction is deeply integrated into vector index construction for Approximate Nearest Neighbor Search (ANNS). The rapid growth in vector dimensionality and volume has significantly increased the computational cost of PQ. Existing GPU-based PQ accelerations are ill-suited for PQ construction due to its "one-to-one" execution pattern (one compute, one data load, i.e., data transfer overhead dominates). Although CPU-based solutions are prevalent, they are essentially general-purpose designs that fail to capture the intrinsic characteristics of PQ construction.In this paper, we propose CS-PQ, a Cache-friendly, SIMD-optimized PQ framework based on modern CPUs. CS-PQ introduces a vector-oriented SIMD paradigm that decouples quantization granularity from SIMD width by vectorizing across PQ centroids rather than subvector dimensions. It further restructures the execution pipeline to improve cache locality and reformulates PQ computation to eliminate redundant operations while preserving correctness. Experiments on large-scale datasets show that CS-PQ achieves up to 10.7 times speedup over state-of-the-art CPU-based PQ implementations without sacrificing ANNS accuracy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25521 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CS-PQ: Cache-Friendly SIMD Product Quantization for Large-Scale ANNS Index Construction Ma, Y. T. Huang, K. C. Jiang, X. K. Wang, M. L. Yao, X. Chen, R. H. Zhang, G. Shao, Z. L. Databases Product Quantization (PQ) construction is deeply integrated into vector index construction for Approximate Nearest Neighbor Search (ANNS). The rapid growth in vector dimensionality and volume has significantly increased the computational cost of PQ. Existing GPU-based PQ accelerations are ill-suited for PQ construction due to its "one-to-one" execution pattern (one compute, one data load, i.e., data transfer overhead dominates). Although CPU-based solutions are prevalent, they are essentially general-purpose designs that fail to capture the intrinsic characteristics of PQ construction.In this paper, we propose CS-PQ, a Cache-friendly, SIMD-optimized PQ framework based on modern CPUs. CS-PQ introduces a vector-oriented SIMD paradigm that decouples quantization granularity from SIMD width by vectorizing across PQ centroids rather than subvector dimensions. It further restructures the execution pipeline to improve cache locality and reformulates PQ computation to eliminate redundant operations while preserving correctness. Experiments on large-scale datasets show that CS-PQ achieves up to 10.7 times speedup over state-of-the-art CPU-based PQ implementations without sacrificing ANNS accuracy. |
| title | CS-PQ: Cache-Friendly SIMD Product Quantization for Large-Scale ANNS Index Construction |
| topic | Databases |
| url | https://arxiv.org/abs/2605.25521 |