I answered my own question using the benchmark project on this thread:
https://forums.creativecow.net/thread/2/1019120
I used the linked AE CS6 project to test rendering speed on both cards:
Common:
AMD Phenom II X4 965 Black Edition OC’d to 3.9ghz
8GB DDR3 RAM
Windows 8 64-bit
First video card:
GeForce GTX 560 Ti (Fermi) 2GB 256-bit GDDR5
core clock 880MHz
384 CUDA cores
AVI render time: 9 min 55 sec
Second video card:
GeForce GTX 650 Ti 2GB 128-bit GDDR5
core clock 1071MHz
768 CUDA cores
AVI render time: 4 min 14 sec
Interesting find. Across the board with game performance, the 256-bit memory in the GTX 560 Ti sports a noticeable performance improvement over the GTX 650 Ti with only 128-bit memory. But when it came to rendering this project (as AVI), the newer card was twice as fast. Maybe it does have a lot to do with the CUDA cores in this case.
Hope this helps someone else!