System-Level Power Estimation with On-Chip Bus Performance Monitoring Units in PKU-DSPII SoC

Abstract

Runtime system-level power estimation is essential for dynamic power adaptation in integrated circuits. Indirect power estimation using a CPU performance monitoring unit (PMU) is widely used in modern microprocessors for its low cost. However, the existing CPU PMUs only monitor the activities of core and cache, resulting in accuracy limitation in power estimation for systems containing heterogeneous devices. In this paper, an onchip bus PMU (OCB PMU) is proposed for PKU-DSPII SOC to achieve accurate system-level power estimation by taking range of peripheral devices into account. By monitoring the on-chip bus signals which indicate almost all the information of the system behavior, an OCB PMU is able to recognize different types of devices. By imitating the internal behavior of every device triggered by OCB signals using an energy state machine (ESM), an OCB PMU is capable of accurate power estimation of heterogeneous devices with complicated behavior. The proposed PMUs can also flexibly adapt to different SoC architectures.

Fast and Accurate Dynamic Power Estimation method based on Component

“…Dynamic power estimation techniques for FPGA designs can roughly be divided into two categories [7][8][9][10] . One category is based on low level simulation, which is employed by tools such as PowerPlay from Altera, XPower from Xilinx et al In order to estimate the dynamic power estimation based on low level simulation, simulation is performed to obtain the switching activity of the low level hardware components used in the FPGA design, such as the programming logic elements and the routing element.Using the low level simulation results, the dynamic power of the complete application is calculated.…”