Hmm. That's a good suggestion, but it looks like the regStats stuff is called from python, and my kung fu isn't strong enough to go messing with that stuff just for a quick hack.
Gabe Ali Saidi wrote: > As long as the initialization code gets run before all the regStats calls are > done it should be ok. If it's temporary you could just modify the static > simobject function that calls regstats on all simobjects to also call > whatever it is that you want. This goes back to Matt's question. We really > don't have a good way to instrument individual instructions. We never destroy > a static inst when we created it right? Perhaps we could have each static > inst hold an execution count that the code could then aggregate by opcode at > the end of the simulation. > > Ali > > On Oct 24, 2010, at 2:29 PM, Gabe Black wrote: > > >> I'm going to instrument M5 to count the number of stupd microops that >> get executed during a run, and I thought it would be nice to use the >> fancy stats system we have. Because I want an aggregate of all the times >> a stupd is executed, I'd like to set up a static stat I can increment >> that isn't associated with any particular simobject. Is there a way to >> do that? I know there are regStats functions on the various simobjects, >> but I don't know if that's the only way to use them. Thanks! >> >> Gabe >> _______________________________________________ >> m5-dev mailing list >> [email protected] >> http://m5sim.org/mailman/listinfo/m5-dev >> >> > > _______________________________________________ > m5-dev mailing list > [email protected] > http://m5sim.org/mailman/listinfo/m5-dev > _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
