arXiv:2606.30949v1 Announce Type: new Abstract: High-Level Synthesis (HLS) provides a fast path from concepts to silicon, but converting real-world software into synthesizable HLS code remains challenging due to restrictive language support and the gap between software and hardware programming practices. Existing automated and LLM-based refactoring approaches partially address this problem, yet they often lack flexibility, struggle to scale, and incur high computational costs. We introduce AgRefactor, an LLM-based multi-agent workflow for refactoring software into HLS-compatible programs. AgRe
The increasing demand for specialized hardware acceleration, particularly for AI, is driving innovation in High-Level Synthesis (HLS) and automated code generation.
Improving the efficiency and accessibility of HLS can significantly accelerate hardware development cycles and reduce the barriers to custom silicon design for software-centric companies.
The introduction of multi-agent LLM workflows like AgRefactor aims to automate complex software-to-hardware refactoring, addressing key limitations of prior approaches.
- · Hardware design companies
- · AI accelerator developers
- · Software developers targeting HLS
- · LLM developers
- · Manual HLS refactoring service providers
- · Traditional hardware description language (HDL) specialists
HLS adoption rates increase, leading to faster iteration on custom silicon designs.
Reduced design barriers empower a broader range of companies to develop specialized hardware, diversifying the compute supply chain.
The enhanced capability for software-defined hardware accelerates the development of novel AI architectures and applications, leading to new compute paradigms.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI