[ ] Roofline Model for Bottleneck representation of both algorithms
[ ] A bunch of interesting statistical visualizations in the "Statistical Analysis of Sketch Estimators" paper
[ ] Show a snippet of critical code, then show the resulting VTune profile, discuss the problem+optimization, then show snippets of the result
[ ] A multi-diagram showing the CPU Time distribution per different sample distribution / buckets size
[x] A quick diagram showing the throughput for depending on sample
[x] A graph displaying throughput scaling depending on sample size(and distribution?) per algorithm. And then another one after optimizations
[ ] A simple graph showing only improvement in percent for AGMS-Stock vs AGMS-Optimized and F-AGMS-Stock vs F-AGMS-Optimized (averaged across all distributions and all RowXBucket sizes)