Record-Remix-Replay: Hierarchical GPU Kernel Optimization using Evolutionary Search

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Nichols, Daniel, Parasyris, Konstantinos, Melone, Caetano, Ben-Nun, Tal, Georgakoudis, Giorgis, Menon, Harshitha
Format:	Preprint
Published:	2026
Subjects:	Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning Performance
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866914468908236800
author	Nichols, Daniel Parasyris, Konstantinos Melone, Caetano Ben-Nun, Tal Georgakoudis, Giorgis Menon, Harshitha
author_facet	Nichols, Daniel Parasyris, Konstantinos Melone, Caetano Ben-Nun, Tal Georgakoudis, Giorgis Menon, Harshitha
contents	As high-performance computing and AI workloads become increasingly dependent on GPUs, maintaining high performance across rapidly evolving hardware generations has become a major challenge. Developers often spend months tuning scientific applications to fully exploit new architectures, navigating a complex optimization space that spans algorithm design, source implementation, compiler flags and pass sequences, and kernel launch parameters. Existing approaches can effectively search parts of this space in isolation, such as launch configurations or compiler settings, but optimizing across the full space still requires substantial human expertise and iterative manual effort. In this paper, we present Record-Remix-Replay (R^3), a hierarchical optimization framework that combines LLM-driven evolutionary search, Bayesian optimization, and record-replay compilation techniques to efficiently explore GPU kernel optimizations from source-level implementation choices down to compiler pass ordering and runtime configuration. By making candidate evaluation fast and scalable, our approach enables practical end-to-end search over optimization dimensions that are typically treated separately. We show that Record-Remix-Replay can optimize full scientific applications better than traditional approaches over kernel parameters and compiler flags, while also being nearly an order of magnitude faster than modern evolutionary search approaches.
format	Preprint
id	arxiv_https___arxiv_org_abs_2604_11109
institution	arXiv
publishDate	2026
record_format	arxiv
spellingShingle	Record-Remix-Replay: Hierarchical GPU Kernel Optimization using Evolutionary Search Nichols, Daniel Parasyris, Konstantinos Melone, Caetano Ben-Nun, Tal Georgakoudis, Giorgis Menon, Harshitha Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning Performance As high-performance computing and AI workloads become increasingly dependent on GPUs, maintaining high performance across rapidly evolving hardware generations has become a major challenge. Developers often spend months tuning scientific applications to fully exploit new architectures, navigating a complex optimization space that spans algorithm design, source implementation, compiler flags and pass sequences, and kernel launch parameters. Existing approaches can effectively search parts of this space in isolation, such as launch configurations or compiler settings, but optimizing across the full space still requires substantial human expertise and iterative manual effort. In this paper, we present Record-Remix-Replay (R^3), a hierarchical optimization framework that combines LLM-driven evolutionary search, Bayesian optimization, and record-replay compilation techniques to efficiently explore GPU kernel optimizations from source-level implementation choices down to compiler pass ordering and runtime configuration. By making candidate evaluation fast and scalable, our approach enables practical end-to-end search over optimization dimensions that are typically treated separately. We show that Record-Remix-Replay can optimize full scientific applications better than traditional approaches over kernel parameters and compiler flags, while also being nearly an order of magnitude faster than modern evolutionary search approaches.
title	Record-Remix-Replay: Hierarchical GPU Kernel Optimization using Evolutionary Search
topic	Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning Performance
url	https://arxiv.org/abs/2604.11109