Hi Glenn, I'm not an expert on the o3 model, but what you propose sounds like it should be able to work. Since fetch can already stall for other reasons, you would just need to find where the reasons are checked and add an if outstandingBranches > N condition check. There should be a place where you can grab the number of outstanding branches if you like.
Ali On Oct 28, 2010, at 3:26 PM, Glenn Ko wrote: > Hello, > > > I am trying to find out if there is a way to limit the number of speculation > levels or outstanding branches in the O3CPU model. > So far from my understanding, it seems like the branch predictor allows > branches to be taken while previous branches are still unresolved, without > any limits on the number. Is this correct? > If so, how would I add a limit to it? > I was thinking I could create a variable and added a check before the > prediction and stall the fetch stage if the max outstanding branches are > reached. Do you think there would be any problems? I would like to hear > your opinions. > > Thanks in advance! > > > -Glenn > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
