Metraipolis › archive › dc_236d9595bf85

Weights That Stay: On-Chip Megakernels for Serial Heads in Speculative Decoding — arXiv:2611.00506

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.00506
License (as archived)
CC0-1.0
Archived
2026-11-03T18:26:35.084Z
SHA-256 of the archived text
f79188545db6e69fde60a6179c67c06743cd4e142fd6ad359a4c5b45863be014
Stable URL of this page
https://metraipolis.com/documento/dc_236d9595bf85

How to cite

Weights That Stay: On-Chip Megakernels for Serial Heads in Speculative Decoding — arXiv:2611.00506 — arXiv (en), https://arxiv.org/abs/2611.00506 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_236d9595bf85 (Metraipolis archive dc_236d9595bf85, sha256 f79188545db6e69fde60a6179c67c06743cd4e142fd6ad359a4c5b45863be014)
BibTeX
@misc{dc236d9595bf85,
  title = {Weights That Stay: On-Chip Megakernels for Serial Heads in Speculative Decoding — arXiv:2611.00506},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.00506},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_236d9595bf85; sha256 f79188545db6e69fde60a6179c67c06743cd4e142fd6ad359a4c5b45863be014},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00506. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Parallel speculative drafters use a lightweight serial head to recover token dependencies within a block generated in parallel. Each head step updates only a small state, yet projects that state through the same vocabulary-sized matrix. Step-wise execution repeatedly transfers this large, immutable projection, making the head a meaningful source of latency despite its low arithmetic cost. We present Weights That Stay, an execution design that keeps projection shards on chip for the entire serial draft walk. A persistent megakernel couples weight residency with GPU-side token selection and state advancement, reusing resident weights while exchanging only compact state between steps. The design supports full or partial residency and accommodates both Markov and recurrent heads on H100 and B200. Integrated into SGLang, Weights That Stay improves throughput for Markov drafting on Qwen3-4B and Qwen3-8B, with gains of up to 12.3%. At the head level, the DSpark walk is up to 11.6$\times$ faster than the fastest measured SGLang reference path. These results show that aligning weight residency with the full draft walk translates cross-step reuse into end-to-end serving gains.

Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.DC, nuovi lavori del giorno).
Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.00506 — 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 2026-11-03T18:10:22.796Z dalla pagina https://arxiv.org/list/cs.DC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).