----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.m5sim.org/r/925/#review1713 -----------------------------------------------------------
I think this is a great idea and something I'd have liked to have a long time ago. I have two complaints though. The first is that it doesn't go far enough. I'd like to see this stat ported to all the CPUs and be a first class stat up there with sim_inst and sim_ticks. When you make a change in the microcode of an instruction in x86, for instance, you can be fairly confident you haven't broken anything if the number of microops change but the number of architecture level instructions hasn't. That would be useful no matter what CPU you're on. Second, you're naming looks wrong. This stat doesn't count macroops, it counts not microops which include non-microcoded instructions. I think that's actually the more useful of the two options anyway, so it's really just the name that I think should change. Perhaps we should go as far as making this the new sim_insts and come up with a new name for what sim_insts is currently counting. - Gabe On 2011-12-01 00:37:23, Ali Saidi wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.m5sim.org/r/925/ > ----------------------------------------------------------- > > (Updated 2011-12-01 00:37:23) > > > Review request for Default, Ali Saidi, Gabe Black, Steve Reinhardt, and > Nathan Binkert. > > > Summary > ------- > > O3: Add stat that counts committed macro-instructions > > > Diffs > ----- > > src/cpu/o3/commit.hh d4e08e83f4fa > src/cpu/o3/commit_impl.hh d4e08e83f4fa > > Diff: http://reviews.m5sim.org/r/925/diff > > > Testing > ------- > > > Thanks, > > Ali > > _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
