How to do research that studies two different computing architectures and compares them, ideal areas to cover in such research, and an Ideal theory or theories that can be used as the basis of the research
Performance evaluation is often carried out through simulations. You need to ask questions such as (a) What are some performance metrics for comparison - e.g. speedup, bandwidth, throughput, power, etc. (b) how can these be measured? (c) When actual prototyping is not possible, one has to use simulations - how do I model the architecture for simulation and defend the model? (d) Sometimes, queuing theory may be useful.