arXiv：cs.LG· Zhenduo Zhao, Qihui Zhou, Mingcong Song, Zhiyi Chen, Chuangguan Ye, Fengfan Hou, Zequn Gong, Jing Li, Hongjie Si, Guoping Long·· 3 小时前
QUILT：通过共享查询执行重新思考稀疏注意力预填充
QUILT: Rethinking Sparse-Attention Prefill through Shared Query Execution
AI 导读
QUILT 是一种工作负载感知的稀疏注意力执行机制，通过联合处理相邻查询、复用共享 KV 条目来减少冗余内存流量与计算。它引入 Shift-and-Compare Set Decomposition（SCSD），将不规则集合操作转为规则数据并行原语并与注意力计算流水线化。
正文
Abstract:Sparse attention reduces the cost of long-context attention, but existing kernels typically process queries independently, repeatedly loading and dequantizing KV entries shared across queries. We observe substantial overlap in the KV entries selected by neighboring queries, creating opportunities for cross-query reuse. We present QUILT, a workload-aware sparse-attention execution mechanism that jointly processes neighboring queries and reuses shared KV entries to reduce redundant memory traffic and computation. QUILT introduces Shift-and-Compare Set Decomposition (SCSD), which transforms irregular set operations into regular data-parallel primitives suitable for modern accelerators, and pipelines SCSD with attention computation to hide its overhead. Cascaded sharing captures reuse hierarchically at multiple granularities. A tile-aware execution strategy balances sharing granularity with hardware tile utilization and selectively removes low-importance query-specific tails to eliminate underutilized tiles. We evaluate QUILT on LongBench using GLM-5.3 and DeepSeek-3.2 under both tensor and sequence parallelism. Compared with the state-of-the-art sparse-attention kernel, QUILT reduces average kernel latency by up to 55.1% and processed KV data by up to 55.9%, while reducing time-to-first-token (TTFT) latency by up to 36.8% with negligible accuracy degradation.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.11134 [cs.LG] |
| (or arXiv:2610.11134v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11134 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Qihui Zhou [view email]
[v1]
Thu, 8 Oct 2026 02:58:29 UTC (826 KB)
来源：arXiv：cs.LG · arxiv.org