Improving Parallel Program Performance with LLM Optimizers via Agent-System Interfaces

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Wei, Anjiang, Nie, Allen, Teixeira, Thiago S. F. X., Yadav, Rohan, Lee, Wonchan, Wang, Ke, Aiken, Alex
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912402948227072
author Wei, Anjiang
Nie, Allen
Teixeira, Thiago S. F. X.
Yadav, Rohan
Lee, Wonchan
Wang, Ke
Aiken, Alex
author_facet Wei, Anjiang
Nie, Allen
Teixeira, Thiago S. F. X.
Yadav, Rohan
Lee, Wonchan
Wang, Ke
Aiken, Alex
contents Modern scientific discovery increasingly relies on high-performance computing for complex modeling and simulation. A key challenge in improving parallel program performance is efficiently mapping tasks to processors and data to memory, a process dictated by intricate, low-level system code known as mappers. Developing high-performance mappers demands days of manual tuning, posing a significant barrier for domain scientists without systems expertise. We introduce a framework that automates mapper development with generative optimization, leveraging richer feedback beyond scalar performance metrics. Our approach features the Agent-System Interface, which includes a Domain-Specific Language (DSL) to abstract away the low-level complexity of system code and define a structured search space, as well as AutoGuide, a mechanism that interprets raw execution output into actionable feedback. Unlike traditional reinforcement learning methods such as OpenTuner, which rely solely on scalar feedback, our method finds superior mappers in far fewer iterations. With just 10 iterations, it outperforms OpenTuner even after 1000 iterations, achieving 3.8X faster performance. Our approach finds mappers that surpass expert-written mappers by up to 1.34X speedup across nine benchmarks while reducing tuning time from days to minutes.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15625
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Improving Parallel Program Performance with LLM Optimizers via Agent-System Interfaces
Wei, Anjiang
Nie, Allen
Teixeira, Thiago S. F. X.
Yadav, Rohan
Lee, Wonchan
Wang, Ke
Aiken, Alex
Machine Learning
Artificial Intelligence
Computation and Language
Distributed, Parallel, and Cluster Computing
Modern scientific discovery increasingly relies on high-performance computing for complex modeling and simulation. A key challenge in improving parallel program performance is efficiently mapping tasks to processors and data to memory, a process dictated by intricate, low-level system code known as mappers. Developing high-performance mappers demands days of manual tuning, posing a significant barrier for domain scientists without systems expertise. We introduce a framework that automates mapper development with generative optimization, leveraging richer feedback beyond scalar performance metrics. Our approach features the Agent-System Interface, which includes a Domain-Specific Language (DSL) to abstract away the low-level complexity of system code and define a structured search space, as well as AutoGuide, a mechanism that interprets raw execution output into actionable feedback. Unlike traditional reinforcement learning methods such as OpenTuner, which rely solely on scalar feedback, our method finds superior mappers in far fewer iterations. With just 10 iterations, it outperforms OpenTuner even after 1000 iterations, achieving 3.8X faster performance. Our approach finds mappers that surpass expert-written mappers by up to 1.34X speedup across nine benchmarks while reducing tuning time from days to minutes.
title Improving Parallel Program Performance with LLM Optimizers via Agent-System Interfaces
topic Machine Learning
Artificial Intelligence
Computation and Language
Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2410.15625