Researchers have developed MAGE (Macro Placement Agentic Engine), a novel multimodal multi-agent framework designed to refine macro placement in physical design flows for integrated circuits. This framework utilizes natural-language directives and a six-phase workflow that combines structured floorplanning rules with visual checks, rather than relying on labeled placement data. MAGE demonstrated significant improvements in timing and routability metrics, outperforming commercial placers and human experts on specific benchmarks, while also enhancing metrics for human-like placement. AI
IMPACT This framework could significantly improve the efficiency and quality of physical design in chip manufacturing by automating complex placement tasks.
RANK_REASON The cluster contains a research paper detailing a new framework for chip design. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
