aiqbench analysis
Kernel Foundry addresses the critical gap in AI-assisted code generation by combining large language models with evolutionary search to produce correct and hardware-efficient GPU kernels. This approach leverages a centralized experience library of optimization knowledge, significantly outperforming standard baselines on correctness benchmarks while preventing common model hallucinations that bypass computation logic. The work represents a significant advancement in automated machine learning (AutoML) for high-performance computing infrastructure rather than consumer applications or general software updates.
Kernel Foundry: A Diagnosis-driven Evolutionary Kernel Optimizer with Multi-Experts
arXiv:2605.30359v2 Announce Type: replace-cross Abstract: Generating high-performance GPU kernels remains challenging due to the need for both correctness and hardware-aware optimization. While large language models (LLMs) show promise in code generation, they often fail to produce kernels that are both correct and efficient. We propose Kernel Foundry, a diagnosis-driven evolutionary framework for automatic GPU kernel optimization. Our method combines expert-guided, retrieval-augmented initialization with a multi-island evolutionary search, where candidate kernels are iteratively refined using structured diagnostic feedback. A centralized experience library accumulates reusable optimization knowledge to guide subsequent evolution, while explicit mechanisms prevent cheating behaviors that bypass kernel-level computation. Experiments on KernelBench show that our method consistently improves both correctness and performance over strong baselines, achieving up to 100% correctness on Level~2.
Abstract:Generating high-performance GPU kernels remains challenging due to the need for both correctness and hardware-aware optimization. While large language models (LLMs) show promise in code generation, they often fail to produce kernels that are both correct and efficient. We propose Kernel Foundry, a diagnosis-driven evolutionary framework for automatic GPU kernel optimization. Our method combines expert-guided, retrieval-augmented initialization with a multi-island evolutionary search, where candidate kernels are iteratively refined using structured diagnostic feedback. A centralized experience library accumulates reusable optimization knowledge to guide subsequent evolution, while explicit mechanisms prevent cheating behaviors that bypass kernel-level computation. Experiments on KernelBench show that our method consistently improves both correctness and performance over strong baselines, achieving up to 100% correctness on Level~2.
Neural and Evolutionary Computing (cs.NE); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Performance (cs.PF); Software Engineering (cs.SE); Systems and Control (eess.SY)
arXiv:2605.30359 [cs.NE]
(or arXiv:2605.30359v2 [cs.NE] for this version)
Full text under the publisher's open licence · 148 words
Related coverage
News
Quantum Computing Report · Jun 7, 2026
Performance Profiling on AMD GPUs – Part 5: Profiling-Driven Kernel Optimization with an AI Code-Assist Tool
AMD · Jul 16, 2026
From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon
BAIR Blog · Jul 29, 2026
Google's AlphaEvolve Reaches General Availability with Evolutionary Code Optimization as a Service
InfoQ · Jul 19, 2026