> At some point in the past we were having trouble with that though, and > as a workaround we started doing the > restore-into-simple-cpu-for-one-cycle thing, and we've never gone back > to revisit it. If you think about it, life would be easier if you > didn't have to construct this ugly two-headed monster of a config just > to restore from a checkpoint. >
On a related "big picture" note, I think there are simulation features that might be cool (or work more seamlessly) if we got "general" checkpointing for any model working. I can imagine a situation where you use the SimpleCPU model to fast forward to a interesting simulation point, use a detailed model for awhile, then switch back to the SimpleCPU model and so on/so forth (e.g. I only want to run detailed execution on the interrupt handlers). >From this discussion, I assume if I wanted to do that I would need to constantly set up the "restore-into-simple-cpu-for-one-cycle thing" for every simple->detailed and detailed->simple switch of CPUs. -- - Korey
_______________________________________________ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev