Currently I'm studying about branch prediction accuracy and the relation with pipelinedepth and/or using parallel fetching (f.e. in superscalar processors). Is it possible to analyse this relationship with the use of software? I was thinking about using zesto (Link) and the sim-bpred simulator (http://www.ece.uah.edu/~lacasa/tutorials/ss/ss.htm#sim-bpred). And working with metrics like IPC (#instructions per cycle). What are the (dis)advantages of using these? Or is there any other (good) way for doing analysis on this topic?
Asked
Active
Viewed 112 times
