What is PWM input with very high resolution?

If you want to measure the PWM on/off ratio to 1 part in 2^30, that's
about 1 part in 1E9. Not particularly difficult or expensive, but the
measurement might take a bit of time. You need a fairly stable
timebase (but nowhere near 1 part in 1E9) and to accumulate the on and
off times. If you used a timebase in the 50-100MHz range you could get
a measurement in less than a minute.
Best regards,
Spehro Pefhany

As suggested, it depends on how quickly you want the answer.

If this is a stable (slowly changing) PWM, then you can sample over
many cycles and total the HI and LOWs (assumes uncorrelated sampling)

If you want to see state of the art, look at Agilent (ex HP) Frequency
Counters. These deliver 10-12 Digits per second, and resolve to 150ps

You could even do this Analog, these days, the TI ADS1282
specs 31 bits of Analog Precision, (no missing codes).

Also, keep in mind how the PWM Originated, as there will be
a jitter/noise figure for the signal you are measuring.

-jg

"> You could even do this Analog, these days, the TI ADS1282

If the OP doesn't even understand the concept of pwm resolution, how is he
going to make use of a 31 bit A/D ?

Quick, tell me who is selling a 31 bit A/D, and for how much?  :-)

High resolution PWM sampling could mean, that someone can measure the duty
cycle with high resolution. E.g. with a resolution of 2^30 it would be
possible to measure differences of 100%/2^30 in the duty cycle, which I
think is impossible or at least very very expensive.

Could it be they simply mean the number of bits used for the data
acquisition ?

2^30 is meaningless/pointless. PWM resolution of 2^16 is quite practical.
Resolution is the ability to measure the difference between 2^30 and
(2^30) -1

Is there a reason why you can't ask them yourself?

The nature of your question suggests that either
you or they are at least a little bit confused.