Systematic Approach in Intel SoC (System on Chip) Thermal Solution Design using CFD (Computational Fluid Dynamics) Simulation

This paper focus on Intel System on Chip (SoC) thermal analysis challenges and novel methods in addressing Intel SoC under Internet of Things Group (IoTG) unique workloads. The workload challenges are coming from embedded. industrial and PC client market segments; where customers from these segments have different workloads hence different power dissipation (both dynamic and static power) within SoC Core and IPs. Conventional Design Methodology and Practices - prototyping and testing. are time consuming and don’t scale well with fundamentally different and diverse Internet of Things (IoT) workloads. IoTG Markets has shorter Product Life Cycle (PLC) and needed Targeted Thermal Solution. thus requires a quick-turn around time for evaluating these solutions. IoTG emphasize on pre silicon Computational Fluid Dynamics (CFD) simulation and Co-development strategy to design and develop thermal solution and specifications. The thermal analysis was done in an incremental fashion. from lowest component level within the SoC. then platform and System level ingredients - considering heat flux. extended temperature. mutual heating and etc. This incremental process helped facilitate early validation of design decisions at every level (component. platform and system) and timely discovery of potential refinement leading to an optimal solution. Component level feasibility focuses on SoC power maps and heat-sink design. Platform level feasibility expands into form factor requirement and on-board component placements. System level feasibility envelope everything to fully capture the details of customer boundary conditions like operating ambient. system chassis and applications. Highlighting an example using CFD simulation start with component level analysis. then move on to platform level where board level components are introduced; and finally. system level where boundary conditions with specific use conditions are modeled. Iterations with pre-silicon use cases power assumption to assess thermal solution performance. Through this presentation. we would like to share a bottom up simulation design approach in solving complex thermal system.