Schnorr lot validation speed statistics

In this question, Schnorr vs ECDSA I see in Peter Wuille's response the following comment about the validation of batches of Schnorr signatures: "Improvement of verification speed, supporting the validation of batches of all the signatures in a block to the once (for a fraction of the speed of validating them individually) ".

However, check the table in the BIP here
It seems that the savings are actually a registration factor (n). In other words, for approximately 10,000 signatures, the time saving is approximately 2.5X. I know that technically that is a fraction, but it seemed that the implication was that the improvement was more dramatic.
Is there something you are forgetting? Also, what is the real time in the clock cycles that was a starting point for a Schnorr signature verification?
Thank you.