On Sun, Jul 15, 2012 at 12:36 PM, Nilay <[email protected]> wrote: > On Sun, July 15, 2012 2:21 pm, Steve Reinhardt wrote: > > It's very strange... the test passes on my local machine, but running it > > on > > zizzer (both the regression run and a run I did independently) gives this > > small stats difference: > > > > sim_insts 138165779 138165778 -1 > > -0.00% > > sim_ops 265203823 265203814 -9 > > -0.00% > > sim_ticks 5191766314000 5191766314000 0 > > +0.00% > > system.cpu.committedInsts 138165779 138165778 -1 > > -0.00% > > system.cpu.committedOps 265203823 265203814 -9 > > -0.00% > > > > I ran with both of the gcc versions on zizzer (4.4.5 and 4.5.2) and got > > the > > same result. My machine (where the test passed) has gcc 4.4.3, > > > > Any ideas? > > > > Can some one bisect the repo and figure out the changeset that introduced > this issue? May be some uninitialized variable is in use. > http://repo.gem5.org/gem5/rev/48eeef8a0997 >
Good thinking... I did, and it's the very latest changeset (Andreas's SimpleMemory change, http://repo.gem5.org/gem5/rev/48eeef8a0997) that introduces the problem. Steve _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
