StitchCUDA: An Automated Multi-Agents End-to-End GPU Programing Framework with Rubric-based Agentic Reinforcement Learning

• URL: http://arxiv.org/abs/2603.02637v1
• Date: Tue, 03 Mar 2026 06:04:49 GMT
• Title: StitchCUDA: An Automated Multi-Agents End-to-End GPU Programing Framework with Rubric-based Agentic Reinforcement Learning
• Authors: Shiyang Li, Zijian Zhang, Winson Chen, Yuebo Luo, Mingyi Hong, Caiwen Ding,
• Abstract summary: We propose StitchCUDA, a multi-agent framework for end-to-end GPU program generation.<n> Experiments show that StitchCUDA achieves nearly 100% success rate on end-to-end programming tasks, with 1.72x better speedup than the multi-agent baseline.
• Score: 26.264303471292845
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Modern machine learning (ML) workloads increasingly rely on GPUs, yet achieving high end-to-end performance remains challenging due to dependencies on both GPU kernel efficiency and host-side settings. Although LLM-based methods show promise on automated GPU kernel generation, prior works mainly focus on single-kernel optimization and do not extend to end-to-end programs, hindering practical deployment. To address the challenge, in this work, we propose StitchCUDA, a multi-agent framework for end-to-end GPU program generation, with three specialized agents: a Planner to orchestrate whole system design, a Coder dedicated to implementing it step-by-step, and a Verifier for correctness check and performance profiling using Nsys/NCU. To fundamentally improve the Coder's ability in end-to-end GPU programming, StitchCUDA integrates rubric-based agentic reinforcement learning over two atomic skills, task-to-code generation and feedback-driven code optimization, with combined rubric reward and rule-based reward from real executions. Therefore, the Coder learns how to implement advanced CUDA programming techniques (e.g., custom kernel fusion, cublas epilogue), and we also effectively prevent Coder's reward hacking (e.g., just copy PyTorch code or hardcoding output) during benchmarking. Experiments on KernelBench show that StitchCUDA achieves nearly 100% success rate on end-to-end GPU programming tasks, with 1.72x better speedup over the multi-agent baseline and 2.73x than the RL model baselines.

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)
• Optimizing PyTorch Inference with LLM-Based Multi-Agent Systems [1.2289544895833646]
  We present a framework for comparing multi-agent PyTorch optimization systems.<n>We show that exploit-heavy strategies perform best when paired with error-fixing agents.<n>The best implementation achieves an average 2.88x speedup on an H100 GPU.
  arXiv  Detail & Related papers  (2025-11-21T05:37:38Z)
• Eliminating Multi-GPU Performance Taxes: A Systems Approach to Efficient Distributed LLMs [61.953548065938385]
  We introduce the ''Three Taxes'' (Bulk Synchronous, Inter- Kernel Data Locality, and Kernel Launch Overhead) as an analytical framework.<n>We propose moving beyond the rigid BSP model to address key inefficiencies in distributed GPU execution.<n>We observe a 10-20% speedup in end-to-end latency over BSP-based approaches.
  arXiv  Detail & Related papers  (2025-11-04T01:15:44Z)
• Astra: A Multi-Agent System for GPU Kernel Performance Optimization [10.715861478214961]
  We introduce Astra, the first multi-agent system for GPU kernel optimization.<n>Within Astra, specialized agents collaborate through code generation, profiling, and planning to produce kernels that are both correct and high-performance.
  arXiv  Detail & Related papers  (2025-09-09T08:39:50Z)
• 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)
• 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)
• Kernel methods through the roof: handling billions of points efficiently [94.31450736250918]
  Kernel methods provide an elegant and principled approach to nonparametric learning, but so far could hardly be used in large scale problems. Recent advances have shown the benefits of a number of algorithmic ideas, for example combining optimization, numerical linear algebra and random projections. Here, we push these efforts further to develop and test a solver that takes full advantage of GPU hardware.
  arXiv  Detail & Related papers  (2020-06-18T08:16:25Z)
• MPLP++: Fast, Parallel Dual Block-Coordinate Ascent for Dense Graphical Models [96.1052289276254]
  This work introduces a new MAP-solver, based on the popular Dual Block-Coordinate Ascent principle. Surprisingly, by making a small change to the low-performing solver, we derive the new solver MPLP++ that significantly outperforms all existing solvers by a large margin.
  arXiv  Detail & Related papers  (2020-04-16T16:20:53Z)
• 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.