I upsized an image from my 5D Mkll straight out of camera raw to double it's size with the new choice of bicubic smoother preserve details and that too 1.5 minutes which is tolerable but I the applied the new smart sharpen and that took about 4 minutes. Smart sharpen in CS6 is pretty quick but this is snail pace.
5D MkIII image.
Win 8 laptop, 24GB RAM, NVidia 660m, SSD scratch drive, 2.4GHz i7 (actually runs at about 3.2 GHz in normal use):
Upsize 15 sec, Smart Sharpen @ defaults 31 sec.
CS6 on same machine, running the same operations (the old algorithms, obviously), upsize 1 sec., Smart Sharpen 11 sec.
Win 7 x64 desktop, 8GB RAM, NVidia 500 Ti, non-SSD scratch drive (Ps runs from an SSD), 3.0 GHz i7:
Upsize 20 sec., Smart Sharpen @ defaults 38 sec.
CS6 upsize <1 sec., Smart Sharpen 14 sec.
I also compared the old "Bicubic Smoother" algorithm, and it runs in the same time in CC as in CS6. So, yes, the new algorithms are considerably more compute-intensive than the old ones. All cores are running close to maximum on the CC operations, which they aren't in CS6, so number of cores definitely makes a difference, as Jim's results indicate.