But when running a ruby simulation, the cpu used is TimingSimpleCPU right? So, actually I am creating a checkpoint with that mode. Even in that case, I cannot restore the checkpoint with O3.
So it seems that the only way to restore a Timing/O3 cpu is by having a checkpoint that was created using Timing/O3 cpu, right? Are you planning to support creating a checkpoint with TimingSimpleCPU and then load it with O3 CPU? Thanks for your time. Regards, -Stavros. On Jun 3, 2012, at 3:50 PM, Nilay Vaish wrote: > On Sun, 3 Jun 2012, Volos Stavros wrote: > >> Hi Nilay, >> >> I am aware of this feature. I do restore the checkpoint using >> --restore-with-cpu. >> >> When setting that parameter to timing, it works. However, the simulation >> will not be detailed. >> >> I want to run an OoO+ruby simulation as soon as the checkpoint is loaded. I >> tried several combinations: >> a) --restore-with-cpu=detailed >> b) --restore-with-cpu=timing --cpu-type=detailed >> >> All of them give the same error: >> >> fatal: Can't unserialize 'system.cpu0:locked' >> @ cycle 2573654901000 >> [paramIn:build/ALPHA_MOESI_hammer/sim/serialize.cc, line 228] >> >> I looked at configs/common/Simulation.py, but it seems that in case that >> restore cpu and cpu type are >> not the same, it switches to Atomic. >> >> So, it's not clear to me what the command should look like so as to perform >> an OoO+ruby simulation by >> loading a ruby checkpoint. >> > > It does not seem like you created the checkpoint using Timing/ O3 cpu, since > neither of them has that 'locked' variable. As of now, I don't think it is > possible to restore a checkpoint created with Atomic CPU using the Timing / > O3 CPU. > > -- > Nilay _______________________________________________ gem5-users mailing list gem5-users@gem5.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
