On Mon, Apr 12, 2010 at 5:36 AM, Korey Sewell <[email protected]> wrote:
>> My question is why "sim_insts" is very different from
>> "itb.fetch_hits + itb.fetch_misses". Does anyone know the underlying
>> reasons?
> if I'm not mistaken, the SimpleCPU also models a very light form of
> stalling. Whether that be from TLB misses, branches, etc. you just
> need to double check the "tick()" code to be sure.

THere are stalls in TimingSimpleCPU but not in AtomicSimpleCPU.

Note that sim_insts + dtb.data_misses + itb.fetch_misses == cpu.numCycles.

Steve
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to