From a "what you get" perspective I totally agree with you. I simplified the comparison to core count to have an easier basis to setup the rest of the analysis.

I would not be the least bit surprised if I halved those hardware numbers and it kept up just fine; mostly for the reasons you pointed out.