Per-Core Power Modeling for Heterogenous SoCs

State-of-the-art mobile platforms, such as smartphones and tablets, are powered by heterogeneous system-on-chips (SoCs). These SoCs composed of many processing elements, including multiple CPU core clusters (e.g., big.LITTLE cores), graphics units (GPUs), memory controllers other on-chip resources. On the one hand, platforms need to provide a swift response time for interactive apps high throughput graphics-oriented workloads; on power consumption must be under tight control prevent skin temperatures energy consumption. Therefore, commercial systems feature range mechanisms dynamic temperature control. However, these techniques rely simple indicators, utilization total System architects typically limited consumption, since resources share same rail. More importantly, most rails not exposed input/output pins. To address this challenge, paper presents thorough methodology model major in SoCs. The proposed models utilize wide performance counters capture workload dynamics accurately. Experimental validation Nexus 6P phone, an octa-core Snapdragon 810 SoC, showed that can estimate within 10% error margin.