First of all, you CAN'T encode with an arbitrary CQ value, and do a comparison.
You want to do a real comparison

then learn how to do file prediction, and when you match the same file size with those old templates to the same file size with the current templates, then you'll see that the current parameters perform way better than the old templates.
What are you trying to compare

You need a reference, and the reference for comparison is to match same file sizes for all tests and samples.
I hope you understand what I mean
-kwag