{"ID":5675343,"CreatedAt":"2026-07-03T01:40:09.565152011Z","UpdatedAt":"2026-07-07T01:06:03.009715918Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.02097","arxiv_id":"2607.02097","title":"WBMM: Windowed Batch Matrix Multiplication for Efficient Large Receptive Field Convolution","abstract":"Large kernel depthwise convolutions achieve strong performance but suffer from significant degradation as kernel size grows due to irregular memory access from gather-based computation; while Large Kernel Acceleration (LKA) helps on small feature maps, it becomes counterproductive on large feature maps, even slower than non-accelerated implementations. We propose Windowed Batch Matrix Multiplication (WBMM), which partitions input into contiguous windows and indexes a compact relative position bias table to construct weight matrices, enabling regular memory access via batched matrix multiplication. This yields a unique property: WBMM's throughput improves with larger windows, opposite to depthwise convolutions that degrade with larger kernels. Operator-level benchmarks show WBMM with 14x14 windows outperforms 5x5 depthwise convolution baselines in speed while providing a 7.8x larger per-layer receptive field. Combined with inter-block cross-window communication and hierarchical window reparameterization, WBMM achieves comparable or higher accuracy on ImageNet-1K, COCO, and ADE20K with 1.31-1.88x training speedup, and demonstrates consistent advantages across GPU, CPU, and edge devices without requiring specialized acceleration kernels. Our code is available at http://github.com/wansong-s/WBMM","short_abstract":"Large kernel depthwise convolutions achieve strong performance but suffer from significant degradation as kernel size grows due to irregular memory access from gather-based computation; while Large Kernel Acceleration (LKA) helps on small feature maps, it becomes counterproductive on large feature maps, even slower tha...","url_abs":"https://arxiv.org/abs/2607.02097","url_pdf":"https://arxiv.org/pdf/2607.02097v1","authors":"[\"Wan Song\",\"Wei Zhou\",\"Rui Wang\",\"Jun Yu\",\"Toru Kurihara\",\"Jiajia Xu\",\"Shu Zhan\"]","published":"2026-07-02T12:33:03Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.LG\"]","methods":"[]","has_code":false,"code_links":[{"ID":613898,"CreatedAt":"2026-07-03T01:40:09.565152011Z","UpdatedAt":"2026-07-03T01:40:09.565152011Z","DeletedAt":null,"paper_id":5675343,"paper_url":"https://arxiv.org/abs/2607.02097","paper_title":"WBMM: Windowed Batch Matrix Multiplication for Efficient Large Receptive Field Convolution","repo_url":"https://github.com/wansong-s/WBMM","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}