In order to solve the runtime variability problem caused by the contentions of shared hardware resources in multi-core processors for safety-critical systems, we propose a generalized measurement method based on performance monitor register (PMR) and resource stressing benchmark (RSB). The method captures the hardware events related to the execution time variability to analyze the sharing of hardware resources, the variability of execution time, and the black box or gray box behavior of the hardware platform. The proposed method can not only estimate the performance of hardware platforms, but also the resource consumption of applications, which thus can provide guidance for the worst case execution time (WCET) prediction in a pre-determined co-running context.