AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization

TL;DR

AccelOpt is a self-improving LLM-based system that autonomously optimizes AI accelerator kernels, demonstrating significant throughput improvements and cost-effectiveness on AWS Trainium hardware.

Contribution

It introduces a novel self-improving LLM agentic system for kernel optimization that does not require expert knowledge, with a new benchmark suite and open-source code.

Findings

Improves average throughput from 49% to 61% on Trainium 1.

Achieves kernel optimization comparable to Claude Sonnet 4 at 26x lower cost.

Demonstrates continuous improvement over time with open-source models.

Abstract

We present AccelOpt, a self-improving large language model (LLM) agentic system that autonomously optimizes kernels for emerging AI acclerators, eliminating the need for expert-provided hardware-specific optimization knowledge. AccelOpt explores the kernel optimization space through iterative generation, informed by an optimization memory that curates experiences and insights from previously encountered slow-fast kernel pairs. We build NKIBench, a new benchmark suite of AWS Trainium accelerator kernels with varying complexity extracted from real-world LLM workloads to evaluate the effectiveness of AccelOpt. Our evaluation confirms that AccelOpt's capability improves over time, boosting the average percentage of peak throughput from $49\%$ to $61\%$ on Trainium 1 and from $45\%$ to $59\%$ on Trainium 2 for NKIBench kernels. Moreover, AccelOpt is highly cost-effective: using open-source models, it matches the kernel improvements of Claude Sonnet 4 while being $26\times$ cheaper. The code is open-sourced at https://github.com/zhang677/AccelOpt.