Metraipolis › archive › dc_538197d2a85d
Accelerating Transfer-Learning-Based Autotuning with Predictive LLVM IR Performance Ranking — arXiv:2609.15807
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.15807
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T23:46:18.739Z
- SHA-256 of the archived text
97b26b35736d8b95b36468fcebe2050b252373e3fdfe3878520745e3ba3dc739- Stable URL of this page
https://metraipolis.com/documento/dc_538197d2a85d
How to cite
Accelerating Transfer-Learning-Based Autotuning with Predictive LLVM IR Performance Ranking — arXiv:2609.15807 — arXiv (en), https://arxiv.org/abs/2609.15807 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_538197d2a85d (Metraipolis archive dc_538197d2a85d, sha256 97b26b35736d8b95b36468fcebe2050b252373e3fdfe3878520745e3ba3dc739)
BibTeX
@misc{dc538197d2a85d,
title = {Accelerating Transfer-Learning-Based Autotuning with Predictive LLVM IR Performance Ranking — arXiv:2609.15807},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.15807},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_538197d2a85d; sha256 97b26b35736d8b95b36468fcebe2050b252373e3fdfe3878520745e3ba3dc739},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.15807. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
As the complexity of High Performance Computing (HPC) ecosys- tems continually increases, achieving optimal performance becomes a challenge. Traditional performance autotuning techniques pro- vide promising means to navigate this complexity, these techniques remain computationally intensive and require many evaluations to find optimal configurations. This work proposes an autotuning framework that designs a machine learning-based ensemble LLVM Intermediate Representa- tion (IR) ranker, Neural Configuration Scorer (NCS). NCS ranks the performance of IRs sampled by a transfer-learning-based autotuner, improving the efficiency of the tuning process by reducing tuning overheads and circumventing subpar evaluations. By leveraging knowledge from related tasks, we are able to effectively exploit the transfer relationship to access high-performing configurations in fewer samples than traditional techniques that rely upon itera- tive refinement. Our framework can achieve similar performance improvements as state-of-the-art autotuning techniques with up to 61.67% fewer evaluations, averaging 27.85% fewer evaluations across various HPC benchmarks. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-15 in cs.PF). Categorie arXiv: Performance (cs.PF) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.15807 — 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 2026-10-15T23:13:49.408Z dalla pagina https://arxiv.org/catchup/cs.PF/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).