[gem5-users] GEM5 timing accuracy in system emulation

[Humberto "SilverOne" Carvalho]

Hi,

I'm wondering if GEM5 is able to provide good accuracy in terms of time
*cycles) when running in system mode. In particular, if GEM5 models
memory/instruction barriers and other high cost instructions such as
context switching and their effects on the pipeline/memory system in terms
of time (cycles).

I've seen benchmarks for user mode but unfortunately i was not able to find
anything for kernel mode.

Best Regards,
SilverOne