ParEVO: Synthesizing Code for Irregular Data: High-Performance Parallelism through Agentic Evolution

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yang, Liu, Nie, Zeyu, Liu, Andrew, Zou, Felix, Altinbüken, Deniz, Yazdanbakhsh, Amir, Liu, Quanquan C.
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908861490790400
author Yang, Liu
Nie, Zeyu
Liu, Andrew
Zou, Felix
Altinbüken, Deniz
Yazdanbakhsh, Amir
Liu, Quanquan C.
author_facet Yang, Liu
Nie, Zeyu
Liu, Andrew
Zou, Felix
Altinbüken, Deniz
Yazdanbakhsh, Amir
Liu, Quanquan C.
contents The transition from sequential to parallel computing is essential for modern high-performance applications but is hindered by the steep learning curve of concurrent programming. This challenge is magnified for irregular data structures (such as sparse graphs, unbalanced trees, and non-uniform meshes) where static scheduling fails and data dependencies are unpredictable. Current Large Language Models (LLMs) often fail catastrophically on these tasks, generating code plagued by subtle race conditions, deadlocks, and sub-optimal scaling. We bridge this gap with ParEVO, a framework designed to synthesize high-performance parallel algorithms for irregular data. Our contributions include: (1) The Parlay-Instruct Corpus, a curated dataset of 13,820 tasks synthesized via a "Critic-Refine" pipeline that explicitly filters for empirically performant algorithms that effectively utilize Work-Span parallel primitives; (2) specialized DeepSeek, Qwen, and Gemini models fine-tuned to align probabilistic generation with the rigorous semantics of the ParlayLib library; and (3) an Evolutionary Coding Agent (ECA) that improves the "last mile" of correctness by iteratively repairing code using feedback from compilers, dynamic race detectors, and performance profilers. On the ParEval benchmark, ParEVO achieves an average 106x speedup (with a maximum of 1103x) across the suite, and a robust 13.6x speedup specifically on complex irregular graph problems, outperforming state-of-the-art commercial models. Furthermore, our evolutionary approach matches state-of-the-art expert human baselines, achieving up to a 4.1x speedup on specific highly-irregular kernels. Source code and datasets are available at https://github.com/WildAlg/ParEVO.
format Preprint
id arxiv_https___arxiv_org_abs_2603_02510
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ParEVO: Synthesizing Code for Irregular Data: High-Performance Parallelism through Agentic Evolution
Yang, Liu
Nie, Zeyu
Liu, Andrew
Zou, Felix
Altinbüken, Deniz
Yazdanbakhsh, Amir
Liu, Quanquan C.
Machine Learning
Distributed, Parallel, and Cluster Computing
Neural and Evolutionary Computing
Performance
The transition from sequential to parallel computing is essential for modern high-performance applications but is hindered by the steep learning curve of concurrent programming. This challenge is magnified for irregular data structures (such as sparse graphs, unbalanced trees, and non-uniform meshes) where static scheduling fails and data dependencies are unpredictable. Current Large Language Models (LLMs) often fail catastrophically on these tasks, generating code plagued by subtle race conditions, deadlocks, and sub-optimal scaling. We bridge this gap with ParEVO, a framework designed to synthesize high-performance parallel algorithms for irregular data. Our contributions include: (1) The Parlay-Instruct Corpus, a curated dataset of 13,820 tasks synthesized via a "Critic-Refine" pipeline that explicitly filters for empirically performant algorithms that effectively utilize Work-Span parallel primitives; (2) specialized DeepSeek, Qwen, and Gemini models fine-tuned to align probabilistic generation with the rigorous semantics of the ParlayLib library; and (3) an Evolutionary Coding Agent (ECA) that improves the "last mile" of correctness by iteratively repairing code using feedback from compilers, dynamic race detectors, and performance profilers. On the ParEval benchmark, ParEVO achieves an average 106x speedup (with a maximum of 1103x) across the suite, and a robust 13.6x speedup specifically on complex irregular graph problems, outperforming state-of-the-art commercial models. Furthermore, our evolutionary approach matches state-of-the-art expert human baselines, achieving up to a 4.1x speedup on specific highly-irregular kernels. Source code and datasets are available at https://github.com/WildAlg/ParEVO.
title ParEVO: Synthesizing Code for Irregular Data: High-Performance Parallelism through Agentic Evolution
topic Machine Learning
Distributed, Parallel, and Cluster Computing
Neural and Evolutionary Computing
Performance
url https://arxiv.org/abs/2603.02510