Metraipolis › archive › dc_ded9230a0193
Speeding up Log-Sum-Exp: Kernel Fusion at the Memory Wall, Integer Arithmetic at the Compute Wall — arXiv:2610.33647
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.33647
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-27T08:14:22.165Z
- SHA-256 of the archived text
5c05a2d246e26434f712ceca55c367fb55082d0c90ca27c9b66516f98db2a5ed- Stable URL of this page
https://metraipolis.com/documento/dc_ded9230a0193
How to cite
Speeding up Log-Sum-Exp: Kernel Fusion at the Memory Wall, Integer Arithmetic at the Compute Wall — arXiv:2610.33647 — arXiv (en), https://arxiv.org/abs/2610.33647 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ded9230a0193 (Metraipolis archive dc_ded9230a0193, sha256 5c05a2d246e26434f712ceca55c367fb55082d0c90ca27c9b66516f98db2a5ed)
BibTeX
@misc{dcded9230a0193,
title = {Speeding up Log-Sum-Exp: Kernel Fusion at the Memory Wall, Integer Arithmetic at the Compute Wall — arXiv:2610.33647},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.33647},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ded9230a0193; sha256 5c05a2d246e26434f712ceca55c367fb55082d0c90ca27c9b66516f98db2a5ed},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.33647. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
While the Log-Sum-Exp (LSE) function is a foundational numerical primitive for stable log-domain computation in modern machine learning, its fast and efficient computation on generic platforms (\ie, CPUs and GPUs) has been underexplored, potentially creating execution bottlenecks for large workloads. On GPU, the de facto implementation this http URL dispatches a chain of nine sub-kernels per call, incurring redundant High Bandwidth Memory (HBM) traffic and launch overhead. On CPU, exp and log operations are hardware-hungry, limiting the computing efficiency. To address these bottlenecks, we propose two drop-in kernels to significantly accelerate LSE computation on both GPUs and CPUs: FuseLSE, a single-pass fused kernel that evaluates LSE's exp and log functions on the GPU's Special Function Unit (SFU); and IntLSE, which further replaces the LSE computation with approximated integer arithmetic for CPU computation. Our experiments, validated by benchmarks across workload sizes and hardware platforms, show that both kernels are ${\sim}10-11\times$ faster than PyTorch at small workloads and ${\sim}6\times$ faster on larger ones on GPUs, where the speed-up is attributed to reduced launch overhead and HBM traffic. On CPU, where no SFU is available, LSE computation dominates the cost, allowing IntLSE to gain ${\sim}3-6\times$ execution speed over FuseLSE.
Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.DC, nuovi lavori del giorno).
Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC).
Pagina del lavoro: https://arxiv.org/abs/2610.33647 — 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 2026-10-27T07:56:15.133Z dalla pagina https://arxiv.org/list/cs.DC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).