Compare runs

trappy.compare_runs() is a function to compare two or more runs.

First some hacks to allow importing trappy from here

In [1]:
import sys

In [2]:
%pylab inline
import trappy

Populating the interactive namespace from numpy and matplotlib

map_label is a dict that maps cpumasks to a name. This name will be used as a label for the load and frequency plots.

In [3]:
map_label = {"00000000,00000039": "LITTLE", "00000000,00000006": "big"}

actor_order is an array that with the order in which the actors appear in the thermal_power_allocator tracepoint.

In [4]:
actor_order = ["GPU", "big", "LITTLE"]

runs is an array of tuples consisting of a name and the path to the directory where the trace.dat or trace.txt can be found. The path can be relative to the location of the notebook or an absolute path. We use a relative path in this example so that it works regardless of the path of trappy in your system. Note that you can specify the path to the directory containing the trace.dat or trace.txt or you can specify the path to the file if you have called it something else (e.g. /work/ipa/latest_trace.dat)

In [5]:
runs = [("good", "./trace.txt"), ("experiment", ".")]

In [6]:
trappy.compare_runs(actor_order, map_label, runs)