Interesting experiment and results. As a bit of constructive feedback on the data visualization, I would say that the actual solve times should be points instead of a continuous line. The avg solve time is fine as a line, though.