2.5. THE QRAM ARCHITECTURE 29
Class + Quant
Figure 2.1: The QRAM Architecture.
Once these instructions sent by the classical computer have been carried out in the quantum
processor, a measurement is performed and the result is sent back to the classical processor. And this
process may be repeated as a pipelined cycle multiple times.
Clearly, in this architecture the quantum processor incurs the same limitations as more familiar
augmentations, such as, vector and graphics processors, e.g., data transfer overhead. In addition to
these limitations, the quantum processor potentially introduces others. Depending on how the
quantum register is implemented, there may be a limit on the amount of time that a quantum
superposition can be maintained. This decoherence time represents a constraint which has no parallel
in classical algorithm design
2.5.1 ALGORITHMIC CONSIDERATIONS
As we have discussed before, a quantum algorithm that exclusively uses nontrivial quantum su-
perpositions has many potential problems because of the no-cloning and destructive measurement
restrictions. And on the other hand, a quantum algorithm that onl y uses trivial superpositions does
not take advantage of the algorithmic structure of quantum computing. Clearly, there is an issue of
algorithmic balance, and the quantum software engineer will have to determine which portions of
his code are good candidates for quantum acceleration.
The design of effective quantum algorithms will not be easy, as there are several algorithmic and
architectural considerations that needto be taken into consideration.For instance,it still remains to be
However, there is a parallel to the constraints imposed by signal propagation delays when implementing classical computing hard-