I confirm that "u" is ultraslow. We are working with multi-color profiles and some special methods for editing the BtoA tables.In this case, I was thinking that an ultralarge BtoA table would add some accuracy. However, given the calculation times required, I think we will give this idea a miss.
i've also been doing tests with quality of the B2A table creation (with RGB output profiles) and the difference of quality m/h/u is the resolution of the lut table during profile creation. medium quality (m) is 17x17x17, high quality (h) is 33x33x33, ultra quality (u) is 45x45x45, meaning that the table contains a quantity of x^3 (in case of RGB output) entries.
for example: MonacoProfiler usually contains following resolution/entries with highest quality option: A2B table with 17x17x17 and the B2A table with 33x33x33. in profile.exe this would mean
-qm -bh parameters. regards joe