Metraipolis › archive › dc_a4dfc0733952
Compass: Dissecting Communication and Computation Operators for Efficient LLM Training — arXiv:2609.10549
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.10549
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T05:44:59.912Z
- SHA-256 of the archived text
2fa74bd1797b9d7239ec9614286539f09a360ff13194927b293ca3fc8e57a7b6- Stable URL of this page
https://metraipolis.com/documento/dc_a4dfc0733952
How to cite
Compass: Dissecting Communication and Computation Operators for Efficient LLM Training — arXiv:2609.10549 — arXiv (en), https://arxiv.org/abs/2609.10549 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a4dfc0733952 (Metraipolis archive dc_a4dfc0733952, sha256 2fa74bd1797b9d7239ec9614286539f09a360ff13194927b293ca3fc8e57a7b6)
BibTeX
@misc{dca4dfc0733952,
title = {Compass: Dissecting Communication and Computation Operators for Efficient LLM Training — arXiv:2609.10549},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.10549},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a4dfc0733952; sha256 2fa74bd1797b9d7239ec9614286539f09a360ff13194927b293ca3fc8e57a7b6},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10549. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Overlapping communication and computation operators is a common practice to hide communication overheads, accelerating large language models (LLMs) training on GPU clusters. Existing systems achieve this through either intra-operator fusion (IntraFusion), which packs operators into a single large kernel, or inter-operator decomposition (InterDecom), which splits a tensor into multiple parts for pipelined execution. However, current IntraFusion methods underutilize network topology, causing suboptimal bandwidth usage on multi-GPU systems, while InterDecom struggles to determine the optimal number of decomposed parts for peak performance. To address these issues, we introduce Compass, which employs systematic optimization and comprehensive modeling. First, we design a novel IntraFusion algorithm leveraging double-ring communications to maximize bandwidth utilization in hybrid NVLink-PCIe systems, achieving 1.5x-2.5x speedups. Second, we develop a decomposition model that mathematically derives the optimal tensor decomposition degree for InterDecom, improving performance by up to 1.3x. Finally, we develop a unified performance framework that accurately determines the best strategy for different scenarios. We validate Compass through extensive evaluation across 288 configurations and end-to-end experiments on real-world applications. The results demonstrate that Compass consistently selects the optimal strategy, achieving up to a 1.42x end-to-end speedup compared to the Megatron-LM baseline. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-11 in cs.PF). Categorie arXiv: Performance (cs.PF). Pagina del lavoro: https://arxiv.org/abs/2609.10549 — 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 2026-10-16T05:27:15.554Z dalla pagina https://arxiv.org/catchup/cs.PF/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).