your math is wrong. this chip cannot do miracles it can only detect maximum of 4 transitions of the quadrature encoder.
how it works:
PPR value is how many pulses per revolution you get from ONE channel.
if your encoder has 2500PPR it means that:
channel A will provide 2500 pulses per revolution and
channel B will provide 2500 pulses per revolution
the only trick is that the phases of the two signals are shifted.
if you take a closer look at ONE PERIOD of say channel A
(period in this case is one whole "1" and one whole "0" of the A channel)
and try to figure out how many UNIQUE states you can create
by combining this with signal from channel B, you will see only 4 states.
many encoder interfaces don't go this route, they simply count pulses
depeding on which channel leads at the moment. This chip does
extract all 4 states withing ONE encoder pulse.
So if your encoder has 2500 RPM, using this chip you can get
2500x4 = 10000 pulses per revolution - and that's it
(see quote, you ware multiplying this one more time by 4).
Datasheet says that max pulse rate is CLOCK/8.
with 8MHz clock, you can detect max 1000000 pulses (1MHz).
if you have encoder with 500 PPR, chip can resolve position up to
500x4=2000 pulses per revolution.
1MHz/2000PPR = 1000000 (pulse per sec) / 2000 (pulse per rev) =
500 Revolution per second.
there are 60 sec in a minute so we multiply this by 60 to convert
the value to RPM:
500 (rev per second) = 500 (rev per 1/60 min) =500x60=30000RPM
so the datasheet example is correct.