LLM-DSE: Searching Accelerator Parameters with LLM Agents

TL;DR

LLM-DSE introduces a multi-agent framework utilizing large language models to optimize high-level synthesis directives for domain-specific accelerators, significantly improving performance and efficiency over existing methods.

Contribution

The paper presents a novel multi-agent framework combining LLMs with design space exploration to optimize HLS directives more effectively and adaptively.

Findings

Achieves 2.55x performance gains over state-of-the-art methods.

Uncovers novel hardware designs with reduced runtime.

Validates effectiveness through ablation studies.

Abstract

Even though high-level synthesis (HLS) tools mitigate the challenges of programming domain-specific accelerators (DSAs) by raising the abstraction level, optimizing hardware directive parameters remains a significant hurdle. Existing heuristic and learning-based methods struggle with adaptability and sample efficiency. We present LLM-DSE, a multi-agent framework designed specifically for optimizing HLS directives. Combining LLM with design space exploration (DSE), our explorer coordinates four agents: Router, Specialists, Arbitrator, and Critic. These multi-agent components interact with various tools to accelerate the optimization process. LLM-DSE leverages essential domain knowledge to identify efficient parameter combinations while maintaining adaptability through verbal learning from online interactions. Evaluations on the HLSyn dataset demonstrate that LLM-DSE achieves substantial $2.55\times$ performance gains over state-of-the-art methods, uncovering novel designs while reducing runtime. Ablation studies validate the effectiveness and necessity of the proposed agent interactions. Our code is open-sourced here: https://github.com/Nozidoali/LLM-DSE.