On the x-axis we have the number of instructions (the average one from the summary of the COMP 4c block of each event) On the y-axis we have the corresponding execution time (the average one from the summary of the COMP 4b block of each event) These two average values allow you to plot a point for each subblock in the trace. Then the idea is to plot also the variability for each point. As a first rough estimation, we could plot whiskers that represent the min and max for each metric (instructions on x, time on y). a better approximation would be to use the min and max after removing the outliers (using values from the most populated bins). A even better one would be to compute the standard deviation from the histogram contents.

Plot between EventID_Subblock and Number of instructions per second from traces acquired with calibration feature enabled

Comparison between COMP 4b blocks of Time Dependent and Time Independent Calibrated Traces

Ratios between avg, min and max of timed and time independent communication time deltas of 5 different traces of each

Ratios between avg, min and max of time dependent traces