is there any difference between two separate Titan Black cards and one Titan Z in the perspective of CUDA?
(except the lower frequencies of the Titan Z)
The GPU-Z screenshot indicates that the Titan Z acts like two separate Titan Black cards in SLI mode.
If I assume correctly, then it would not be possible to have only one capable CUDA device with 30 multiprocessors and 5760 ALUs?
The CUDA deviceQuery app will output two separate CUDA devices and I will have to change my CUDA apps to support multi GPU?
Maybe someone can post the deviceQuery output of the Titan Z?