EvoEngineer: Mastering Automated CUDA Kernel Code Evolution with Large Language Models

• URL: http://arxiv.org/abs/2510.03760v1
• Date: Sat, 04 Oct 2025 10:00:25 GMT
• Title: EvoEngineer: Mastering Automated CUDA Kernel Code Evolution with Large Language Models
• Authors: Ping Guo, Chenyu Zhu, Siyuan Chen, Fei Liu, Xi Lin, Zhichao Lu, Qingfu Zhang,
• Abstract summary: Large Language Models (LLMs) for automating kernel optimization promise promise.<n>General-purpose LLM code evolution methods cannot meet strict correctness requirements of kernel optimization.<n>EvoEngineer provides guidance for designing and adapting optimization strategies to achieve a balance between performance and correctness.<n>Our method achieves a maximum speedup of textbf36.75$times among all operations over PyTorch kernels and delivers the highest speedup on textbf28 (textbf56.0%) of 50 operations that achieve over textbf2times$ acceleration.
• Score: 27.430839306140157
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: CUDA kernel optimization has become a critical bottleneck for AI performance, as deep learning training and inference efficiency directly depends on highly optimized GPU kernels. Despite the promise of Large Language Models (LLMs) for automating kernel optimization, this field suffers from a fragmented ecosystem of isolated and incomparable approaches with unclear problem formulations. Furthermore, general-purpose LLM code evolution methods cannot meet strict correctness requirements of CUDA kernel optimization. We address these fundamental challenges by first formalizing CUDA kernel optimization as a code optimization task with a clear objective, constraints, and evaluation metrics. We then establish the first systematic LLM-based code evolution framework, EvoEngineer, that provides guidance for designing and adapting optimization strategies to achieve a balance between performance and correctness. Finally, we implement a kernel optimization system based on this framework and conduct extensive experiments on 91 real-world CUDA kernels. Our results demonstrate that EvoEngineer achieves a principled balance between performance and correctness, with the highest averaged median speedup of \textbf{2.72}$\times$ over baseline CUDA kernels and a code validity rate of \textbf{69.8}\%, outperforming existing methods on both dimensions. Our method achieves a maximum speedup of \textbf{36.75}$\times$ among all operations over PyTorch kernels and delivers the highest speedup on \textbf{28} (\textbf{56.0\%}) of 50 operations that achieve over \textbf{2$\times$} acceleration.

Related papers

• CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation [51.72529978689561]
  Agent is a large-scale agentic reinforcement learning system that develops kernel expertise through three components.<n>Agent delivers 100%, 100%, and 92% faster rate over torchcompile on KernelBench.
  arXiv  Detail & Related papers  (2026-02-27T18:58:05Z)
• KernelBlaster: Continual Cross-Task CUDA Optimization via Memory-Augmented In-Context Reinforcement Learning [3.4998382481249286]
  We release KernelBlaster as an open-source agentic framework, accompanied by a test harness, verification components, and a reproducible evaluation.<n>Our method achieves mean speed-ups of 1.43x, 2.50x, and 1.50x on KernelBench Levels 1, 2, and 3, respectively.
  arXiv  Detail & Related papers  (2026-02-15T19:48:43Z)
• DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels [17.979042914049842]
  Diffusion large language models (dLLMs) have emerged as a compelling alternative to autoregressive (AR) LLMs.<n>CuKe is an augmented dataset optimized for high-performance kernels.<n>DICE is a series of diffusion large language models designed for kernel generation.
  arXiv  Detail & Related papers  (2026-02-12T08:45:13Z)
• CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization [36.794824560677064]
  CudaForge is a training-free multi-agent workflow for kernel generation and optimization.<n>By leveraging base models like OpenAI-o3, CudaForge achieves 97.6% correctness generated kernels and an average 1.68$times$ speedup.
  arXiv  Detail & Related papers  (2025-10-23T22:52:00Z)
• Adaptive Kernel Design for Bayesian Optimization Is a Piece of CAKE with LLMs [31.28294215349796]
  We propose a Context-Aware Kernel Evolution (CAKE) to enhance BO with large language models (LLMs)<n>Our fresh CAKE-based BO method consistently outperforms established baselines across a range of real-world tasks.
  arXiv  Detail & Related papers  (2025-09-22T16:39:12Z)
• Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization [25.135006275638172]
  We introduce robust-kbench, a new benchmark for rigorous evaluation of kernel performance and correctness across varied scenarios.<n>We also present a comprehensive agentic framework that translates torch code to kernels and iteratively improve their runtime setting.<n>Our approach produces kernels outperforming torch implementations for practical applications, including forward and backward passes.
  arXiv  Detail & Related papers  (2025-09-16T11:08:30Z)
• CUDA-LLM: LLMs Can Write Efficient CUDA Kernels [9.287036563375617]
  Large Language Models (LLMs) have demonstrated strong capabilities in general-purpose code generation.<n>We propose a novel framework called textbfFeature SearchReinforcement (FSR) FSR jointly optimize compilation and functional correctness.
  arXiv  Detail & Related papers  (2025-06-10T10:51:03Z)
• EPS-MoE: Expert Pipeline Scheduler for Cost-Efficient MoE Inference [49.94169109038806]
  This paper introduces EPS-MoE, a novel expert pipeline scheduler for MoE that surpasses the existing parallelism schemes.<n>Our results demonstrate at most 52.4% improvement in prefill throughput compared to existing parallel inference methods.
  arXiv  Detail & Related papers  (2024-10-16T05:17:49Z)
• Harnessing Deep Learning and HPC Kernels via High-Level Loop and Tensor Abstractions on CPU Architectures [67.47328776279204]
  This work introduces a framework to develop efficient, portable Deep Learning and High Performance Computing kernels. We decompose the kernel development in two steps: 1) Expressing the computational core using Processing Primitives (TPPs) and 2) Expressing the logical loops around TPPs in a high-level, declarative fashion. We demonstrate the efficacy of our approach using standalone kernels and end-to-end workloads that outperform state-of-the-art implementations on diverse CPU platforms.
  arXiv  Detail & Related papers  (2023-04-25T05:04:44Z)
• Efficient Convex Algorithms for Universal Kernel Learning [46.573275307034336]
  An ideal set of kernels should: admit a linear parameterization (for tractability); dense in the set of all kernels (for accuracy) Previous algorithms for optimization of kernels were limited to classification and relied on computationally complex Semidefinite Programming (SDP) algorithms. We propose a SVD-QCQPQP algorithm which dramatically reduces the computational complexity as compared with previous SDP-based approaches.
  arXiv  Detail & Related papers  (2023-04-15T04:57:37Z)
• Learning Performance-Improving Code Edits [107.21538852090208]
  We introduce a framework for adapting large language models (LLMs) to high-level program optimization. First, we curate a dataset of performance-improving edits made by human programmers of over 77,000 competitive C++ programming submission pairs. For prompting, we propose retrieval-based few-shot prompting and chain-of-thought, and for finetuning, these include performance-conditioned generation and synthetic data augmentation based on self-play.
  arXiv  Detail & Related papers  (2023-02-15T18:59:21Z)
• Learning to Superoptimize Real-world Programs [79.4140991035247]
  We propose a framework to learn to superoptimize real-world programs by using neural sequence-to-sequence models. We introduce the Big Assembly benchmark, a dataset consisting of over 25K real-world functions mined from open-source projects in x86-64 assembly.
  arXiv  Detail & Related papers  (2021-09-28T05:33:21Z)
• Global Optimization of Gaussian processes [52.77024349608834]
  We propose a reduced-space formulation with trained Gaussian processes trained on few data points. The approach also leads to significantly smaller and computationally cheaper sub solver for lower bounding. In total, we reduce time convergence by orders of orders of the proposed method.
  arXiv  Detail & Related papers  (2020-05-21T20:59:11Z)
• PolyScientist: Automatic Loop Transformations Combined with Microkernels for Optimization of Deep Learning Primitives [55.79741270235602]
  We develop a hybrid solution to the development of deep learning kernels. We use the advanced polyhedral technology to automatically tune the outer loops for performance.
  arXiv  Detail & Related papers  (2020-02-06T08:02:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.