Chip Design Automation: 85% Faster Execution, 49% Better Performance

Designing the placement of devices inside a semiconductor chip is a time-intensive task, often spanning weeks to months due to the intricate nature of the manufacturing process. The variability in design quality and time required, contingent upon the engineer's expertise level, underscores the necessity for automating optimal semiconductor chip design and production to achieve cost reduction and operational efficiency.

MakinaRocks harnessed reinforcement learning for data-driven process optimization and virtual simulation. Our team developed and automated algorithms, employing reinforcement learning models to optimize floorplan layouts and the arrangement of devices on chips.

In comparison to manual labor by specialized personnel, our approach yielded notable improvements in power consumption, performance, and chip area utilization, slashing processing time by approximately 85%. This streamlined tasks for field engineers while enhancing productivity and efficiency by eliminating repetitive processes.