> On 2010-12-21 22:18:34, Steve Reinhardt wrote: > > I agree with Gabe, the explanation in the commit message is kind of > > confusing, but the patch itself looks OK.
O3: Keep around the last committed instruction and use for squashing. Without this change 0 is always used for the youngest sequence number if a squash occured and the ROB was empty (E.g. an instruction is marked serializeAfter or a fetch stall prevents other instructions from issuing). Using 0 there is a race to rename where an instruction that committed the same cycle as the squashing instruction can have it's renamed state undone by the squash using sequence number 0. - Ali ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.m5sim.org/r/348/#review561 ----------------------------------------------------------- On 2010-12-06 16:14:46, Ali Saidi wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.m5sim.org/r/348/ > ----------------------------------------------------------- > > (Updated 2010-12-06 16:14:46) > > > Review request for Default, Ali Saidi, Gabe Black, Steve Reinhardt, and > Nathan Binkert. > > > Summary > ------- > > O3: Keep around the last committed instruction and use for squashing. > > Before this, 0 was used for the youngest sequence number to squash > and there was a race to rename. In the case of a instruction marked > serializeAfter that squashes the ROB is empty and 0 is passed to the rename. > If multiple instructions commited that same cycle rename it could undo some > renames beacuse of the 0 being passed which would undo some state. > > > Diffs > ----- > > src/cpu/o3/commit.hh 2b5fbdcbfb5d > src/cpu/o3/commit_impl.hh 2b5fbdcbfb5d > > Diff: http://reviews.m5sim.org/r/348/diff > > > Testing > ------- > > > Thanks, > > Ali > >
_______________________________________________ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev