Using Clogon's data and the previous aspd formula, under same conditions: 20dex and increasing AGI by 10 intervals.
Here is the comparison between the distributions. Stating the obvious, Gravity is following jRO's suggestion plan.
Previous one had been a linear formula while the current one closely resembles a polynomial 2nd degree formula.
The return from AGI in the new formula is significantly better < 185. It overlapped between 186 and 187 aspd, the turning point.
As it approaches the high end, under the old formula: 210agi would give 193aspd, currently: 210agi only give 190aspd.
Which arises some curiosity, I think Gravity just perfected a formula where the max-aspd for lv99 limited ~190aspd without an added boundary.