The New Compiler Stack: A Survey on the Synergy of LLMs and Compilers

TL;DR

This survey investigates the emerging field of LLM-enabled compilation by proposing a four-dimensional taxonomy (Design Philosophy, LLM Methodology, Level of Code Abstraction, Task Type) and grounding it in a curated corpus of 159 primary studies. It analyzes three architectural roles for LLMs—Selector, Translator, and Generator—and two methodological approaches—Training-Required and Training-Free—while mapping tasks across NL/PL/ASM representations and intra- vs cross-level transformations. The paper highlights three main advancements: democratizing compiler development, enabling novel optimization strategies, and broadening the compiler’s scope to include tasks like transpilation and repair; it also confronts challenges in correctness, scalability, and interpretability, arguing for hybrid, self-improving, and interactive developer-tool paradigms as promising directions. Together, these contributions provide a roadmap for researchers and practitioners to harness LLMs for intelligent, adaptive, and synergistic compilation tools. The emphasis on standardized benchmarks, robust verification, and cross-domain collaboration underpins the practical impact of this work on future compiler design and software engineering workflows.

Abstract

This survey has provided a systematic overview of the emerging field of LLM-enabled compilation by addressing several key research questions. We first answered how LLMs are being integrated by proposing a comprehensive, multi-dimensional taxonomy that categorizes works based on their Design Philosophy (Selector, Translator, Generator), LLM Methodology, their operational Level of Code Abstraction, and the specific Task Type they address. In answering what advancements these approaches offer, we identified three primary benefits: the democratization of compiler development, the discovery of novel optimization strategies, and the broadening of the compiler's traditional scope. Finally, in addressing the field's challenges and opportunities, we highlighted the critical hurdles of ensuring correctness and achieving scalability, while identifying the development of hybrid systems as the most promising path forward. By providing these answers, this survey serves as a foundational roadmap for researchers and practitioners, charting the course for a new generation of LLM-powered, intelligent, adaptive and synergistic compilation tools.

Figures (6)

• Figure 1: Literature Search and Selection Protocol Overview
• Figure 2: Overview of LLM-as-Selector methodology
• Figure 3: Overview of LLM-as-Translator methodology
• Figure 4: Overview of LLM-as-Generator methodology
• Figure 5: Code level view of LLM Compiler