Skip to content
AI News HubLIVE
Original source2 min read

CUDA-Harness: Harnessing Agentic CUDA Kernel Generation and Optimization from Natural Language

Summary

CUDA-Harness is a new framework for generating and optimizing high-performance CUDA kernels from natural language. It introduces Intermediate-Structured Generation to connect high-level semantics with low-level kernel generation, uses Synthesis-Based Verification to mitigate reward hacking, and proposes Feedback-Adaptive Evolution to prioritize correctness while optimizing performance. Experiments show its effectiveness and generalization across LLMs, hardware, and C-to-CUDA transpilation.

SourcearXiv Computational LinguisticsAuthor: Qi Fan, An Zou, Yehan Ma
CUDA-Harness: Harnessing Agentic CUDA Kernel Generation and Optimization from Natural Language
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 30 Aug 2026]

Title:CUDA-Harness: Harnessing Agentic CUDA Kernel Generation and Optimization from Natural Language

View a PDF of the paper titled CUDA-Harness: Harnessing Agentic CUDA Kernel Generation and Optimization from Natural Language, by Qi Fan and 2 other authors

View PDF HTML (experimental)

Abstract:Developing high-performance CUDA kernels demands specialized knowledge in algorithm implementation, correctness validation, and hardware-aware parallel optimization, creating a substantial expertise barrier and making generating CUDA kernels directly from natural language (Text2CUDA) essential. Meanwhile, the general-purpose code generation capability of Large Language Models (LLMs) prompts a series of works exploring LLM-based CUDA kernel generation. They mainly focus on transpilation from high-level frameworks such as PyTorch to CUDA (Torch2CUDA) rather than Text2CUDA, where models must understand the high-level input semantics and handle low-level kernel implementation and validation. Additionally, these methods are vulnerable to reward hacking due to reliance on predefined test inputs. In this paper, we propose CUDA-Harness, a framework for harnessing agentic CUDA kernel generation and optimization from natural language. Specifically, we introduce Intermediate-Structured Generation to connect high-level semantic understanding with low-level kernel generation. To dilute reward hacking in Text2CUDA, we construct Synthesis-Based Verification to provide isolated test data and progressive validation. Furthermore, we propose Feedback-Adaptive Evolution, a kernel evolution strategy that prioritizes correctness while optimizing performance. Finally, through extensive experiments, we demonstrate the effectiveness of CUDA-Harness, with further evaluations illustrating generalization across LLMs, hardware platforms, and to C-to-CUDA transpilation.

Subjects:

Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Programming Languages (cs.PL); Software Engineering (cs.SE)

Cite as: arXiv:2609.00058 [cs.CL]

(or arXiv:2609.00058v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2609.00058

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Qi Fan [view email] [v1] Sun, 30 Aug 2026 13:51:43 UTC (356 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled CUDA-Harness: Harnessing Agentic CUDA Kernel Generation and Optimization from Natural Language, by Qi Fan and 2 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CL

new | recent | 2026-09

Change to browse by:

cs cs.AI cs.MA cs.PL cs.SE

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • CUDA-Harness enables Text2CUDA, directly generating CUDA kernels from natural language, tackling the expertise barrier.
  • The framework includes three core components: Intermediate-Structured Generation, Synthesis-Based Verification, and Feedback-Adaptive Evolution.
  • Synthesis-Based Verification reduces reward hacking by providing isolated test data.
  • Experiments demonstrate effectiveness across various LLMs and hardware, with extension to C-to-CUDA transpilation.

Highlights and analysis are generated automatically and may contain errors. Check the original source.