Short answer: One isn't related to the other.
Electronic flash has a characteristic discharge curve (which you can see here).
Modern flash designs 'cut off' the discharge curve at the appropriate time to deliver anything less than their full output. Note that because the intensity of the burst of light the flash delivers is not uniform during the exposure, cutting a flash's total output by a factor of x does not cut the (time) duration of the flash by the same amount.
As you can see from the chart linked to above, a 1/32 output cuts off quite a lot of the discharge curve, so going to 1/64 is pretty admirable from a feature design perspective.
As a side note, different discharge intensities have a different peak spectral output (color balance), so cutting off the flash early necessarily affects the color of the light that has been discharged. Since no one wants yellow flash, or worse, yellow flash sometimes, these factors must also be compensated for as the flash durations shorten.
When using E-TTL, the flash does not control the duration at all. Rather the camera
continuously monitors the scene exposure before and during the flash output computes the cutoff from the pre-flash reading, using the camera's ambient meter.
As a result, there really is no correlation between the manual features of a flash (eg. 1/32 or 1/64 power) and what levels E-TTL can achieve (which, from the flash's perspective, is simply outputting max power until told to stop).
If your hand was fast enough, you could arbitrarily decide to block the flash's output or, if you were really fast, yank the batteries out of your flash. If you could do something like that in ~1/20,000th of a second then you could achieve similar results to E-TTL.
It's easy to see that the light output would be a function of how quickly you could disable the flash output, and not dependent in any way on the manual feature set provided by your flash.
Here is a good article with much more gory detail on Canon flash technology.
Hope that is helpful,