Metraipolis › archive › dc_24560924fee9

QUILT: Rethinking Sparse-Attention Prefill through Shared Query Execution — arXiv:2610.11134

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.11134
License (as archived)
CC0-1.0
Archived
2026-10-09T06:21:10.823Z
SHA-256 of the archived text
8f1e3c5abb9f0abbd53784d56bdd5b5d5b6fc994d451135ebce35ed198651af0
Stable URL of this page
https://metraipolis.com/documento/dc_24560924fee9

How to cite

QUILT: Rethinking Sparse-Attention Prefill through Shared Query Execution — arXiv:2610.11134 — arXiv (en), https://arxiv.org/abs/2610.11134 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_24560924fee9 (Metraipolis archive dc_24560924fee9, sha256 8f1e3c5abb9f0abbd53784d56bdd5b5d5b6fc994d451135ebce35ed198651af0)
BibTeX
@misc{dc24560924fee9,
  title = {QUILT: Rethinking Sparse-Attention Prefill through Shared Query Execution — arXiv:2610.11134},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.11134},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_24560924fee9; sha256 8f1e3c5abb9f0abbd53784d56bdd5b5d5b6fc994d451135ebce35ed198651af0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.11134. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.11134 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-09T05:50:31.108Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).