Imec UK, the Cambridge branch of the Belgian R&D lab, has embarked on an innovative project funded by the UK’s Advanced Research and Invention Agency (ARIA) with £3 million (€3.6 million). The initiative aims to develop a predictive system-level software framework to optimize energy efficiency in next-generation AI chips by focusing on compute, memory, and connectivity. 🌐

The new simulation tool will integrate existing component models with cutting-edge technologies like CFETs (Complementary FETs) to enhance AI training hardware. By leveraging system technology co-optimization (STCO), the project seeks to tackle power and thermal limits, memory constraints, and bandwidth bottlenecks, which are crucial for effective AI training. 🤖

James Myers, program director at imec Cambridge UK, emphasized that this framework will facilitate early-stage exploration of optimal configurations for AI training and could influence new technology specifications. The project also aims to expand imec’s local team and enhance collaboration within the UK research ecosystem.