extreme axis compensation in software

[Doug Fortune <pentam-at-home.com>]

Lets say hypothetically <bg> I had a way to absolutely 
measure (to 0.000 05" == .001 mm) steps vs actual travelled
distance for each axis screw. 

I understand EMC has an axis compensation table (actually 
two- going in both directions).  I presume it is meant to 
be calibrated every inch or so.  However, as the number of 
calibrations per inch increase, I presume so does the 
software overhead.

Q#1:
At what frequency-of-entry would the calculations become 
too onerous for EMC (ie every 1/10 inch, 1/1000", 
or even 1/10,000" )?

Observation:
Some machines have dual X axes (most decent routers), and some 
very high end milling machines.  The two screws will obviously 
differ in minor amounts, but certainly detectable in the 
0.000 05" range <another bg> measured absolutely from one end 
of the machine.  

Q#2:
Would it be reasonable to have two X calibration tables (one 
for the left X screw, one for the right X screw), and then have EMC 
do a linear interpolation between the two depending on where the 
tool is on the Y axis (ie at halfway across Y axis, there is 
a 50% contribribution from each X axis calibration)?

Please ignore all other issues (such as backlash and temperature 
concerns) for the sake of being focussed on the current topic.  

Doug Fortune
http://www.cncKITS.com