CPU cycle during inference?

Is there anyway to count the number of CPU cycles during inference using TVM_runtime? I am currently using hwcounter python library but it does’nt seem to be accurate.