I have the following two computers:
i7 4930K 3.4Ghz - total 6 cores, 12 threads
Dual Xeon X5650 2.6Ghz - total 12 cores, 24 threads
The i7 is almost twice as fast when it comes to aligning images.
When I check the task manager during the 'Matching Points' phase, the i7 has a continous 100% CPU load. For the dual Xeon, the load spikes between 42% and 85%. So the CPUs don't seem to be used very effectively on the Dual Xeon machine. (See attachments).
Has it been this way in previous versions of Photoscan as well? When I tested the performance two months back, the i7 was only around 15% faster than the Dual Xeons.