> question 1: > How can I configure a detailed cpu to have a perfect branch > predictor? I looked at bpred_unit_impl.hh and there are only two types > of predictor(local and tournament.) Does this mean I am going to add a > new object into M5(write C++ code for perfect_branch_predictor, then > instantiate it using python)? Now that we do execute-in-execute, this may be impossible to do. Kevin or Steve should chime in.
> How about a perfect L2 cache? Should I just increase the L2 cache > size to say 100MB to reduce the miss rate? Just get rid of the L2 cache and make the main memory have the performance of a cache. Nate _______________________________________________ m5-users mailing list m5-users@m5sim.org http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
