Boosting Productivity of Hardware Documentation Using Large Language Models

Adopting Large Language Models (LLMs) has recently gained prominence in various natural language processing tasks. As an inverse approach, this work investigates the feasibility of using LLMs to process formal hardware models and generate design documentations hereof. We automatically pre-process formalized system-level hardware specifications to create prompts for LLMs. Based on these prompts, an LLM generates an extensive, human-readable explanation of the system. Applying this workflow to a concrete example shows that LLMs are suitable for reducing documentation efforts. Nonetheless, we also note that the generated results are not always up to standard, and a manual review is still required. Compared to a fully manual documentation workflow, however, we still observe a noticeable productivity improvement.