If you are trying to find out *why* ASIC's are so much faster the simple answer is: Because a CPU/GPU it a general purpose device they spend a lot of clock cycles just moving data around inside their registers as they execute the various code operations, data input and output is just a small part of the time being spent.

ASIC's don't do they that, they pipeline the data through the hard-coded logic elements. When new work is presented to the chip 1-2 clock cycles are all that is needed to load the work into the pipeline and from there it is 1 cycle to read data AND process And output result. Throw in the fact that even the ancient S1 had 256 cores in each chip and the reason for their speed should be obvious.