Intel’s Custom AI Tool Optimizes SoC Thermal Sensor Placement

An innovative augmented intelligence tool developed by Intel engineers is transforming the process of sensor placement in system-on-chip (SoC) designs, providing crucial data in just minutes. For years, determining the optimal placement of heat-sensitive sensors on Intel processors was a complex blend of science and art.

Circuit designers relied on historical data and experience to decide where to position thermal sensors on CPUs, a process that could take up to six weeks of testing and optimization. The Augmented Intelligence team, led by Dr. Olena Zhu, has developed a groundbreaking tool that streamlines this process, offering enhanced visibility into thermal risks before activating the SoC.

Mark Gallina, a principal engineer at Intel, expressed the significance of this tool, stating, “This tool has completely revolutionized the way we do thermals today. It’s so much more efficient and gives us so much more visibility to thermal risks before we turn the SoC on.” The tool has been instrumental in guiding engineers with precision and efficiency.

The augmented intelligence tool has been successfully implemented in the design of Intel Core Ultra Meteor Lake mobile processors and is set to be utilized in upcoming chips like Lunar Lake. Gallina emphasized the importance of managing thermal heat in client products like laptops, where burst frequencies play a critical role.

Intel engineers must meticulously analyze various workloads to pinpoint thermal hotspots accurately. The integration of AI models into the design process has significantly enhanced efficiency, allowing for quicker decision-making and optimization. Olena highlighted the power of augmented intelligence in data manipulation, enabling engineers to navigate complex design challenges more effectively.