Actually, I already know how to do this, and am working on implementing it into YACMiner (which can be configured to mine ultracoin).  I don't need a central database of cards, it's based on the math needed to do the calculations, and the information available to the mining software from each card. 

If you're a developer, you can look at my chacha-flexible branch on my github https://github.com/thirtybird/yacminer - it re-implements the scrypt-chacha opencl kernel without using the lookup gap as a compiler directive for the opencl kernel, and uses a parameter instead.  This means the mining software can try different lookup-gaps without recompiling the opencl kernel.  Also, over the last few months, I've changed how you configure YACMiner to be more aligned with numbers that people understand - i.e. how many megabytes of GPU memory do you want to allocate?  (Instead of "thread-concurrency"...). Once it had this number, either from BS or TC/LG, and the N factor, it can generate a chart for how many GPU hashes it can launch simultaneously, and pick the best one based on how each LG compares to the next one down.  It won't be perfect because overclocking can change the numbers, but it should eliminate HW errors for people at all N factors.  I'm calling it easy-mode, and hoping it will be within 5% of optimal settings, but I don't have any really high end cards to test it on.

GPU mining at higher N factors is great, so long as you have the information on how to do it, and I've been trying to help people get there, they also have to have realistic expectations.  Nobody is going to get 900 KH/sec mining YACoin on any card without being all HW errors.  That's a hard perception to fight - as posts age and users see these numbers from previous N factors and want to know why they can't get those numbers even though N has incremented five times since that post.

UTC will benefit from those of us who have pioneered with YACoin because we've been there, and done that, and know how to help you do the same without the pain of being the first people to do it.  Delaying N-Factor increases will certainly appeal to folks who've invested in high end Radeon R9 hardware, because they'll stay at the top of the heap, but the higher N factors will allow people to choose lower priced hardware that consumes significantly less power and still do a really good job mining.