src/cpu/pred/bi_mode.cc (Diff revision 1)

Can you explain what these constants are? How did you choose them? Since you are using these in the update function as well, I suggest that you create a separate function for computing the global history index and call that function from lookup() and update().

Show all issues

1. Tony Gutierrez June 24, 2014, 11:01 a.m. (June 24, 2014, 11:01 a.m.)

   This was very specific to the particular implementation I was modeling. I reverted back to the stock bi-mode's indexing scheme, which should not affect anything.

src/cpu/pred/bi_mode.cc (Diff revision 1)

should this be const? I guess not since there is an update at the end, but perhaps it could be made more explicit

Show all issues

1. Tony Gutierrez June 24, 2014, 11:02 a.m. (June 24, 2014, 11:02 a.m.)

   That is the interface exposed by the base predictor for lookup up predictions, which is a pure virtual function. That would have to be changed first.

As Nilay pointed out, the level of comments is not overwhelming. It would be great to see a bit more of an explanation around the computations done (and the constants used).

thanks Tony. Looks fine to me

Ship It!