I am trying to do some basic simulations on cache performance and tried to simulate bzip2 benchmark from SPEC2006 but it never stops (at least I run it 8 hours).
From other side when I run it for limited number of instructions (1Mln) the L1 ICache has very low miss rate (0.000005)
I also check the test stats for bzip2 (spec2000 )provided by gem5 (tests/long) and found out that icache miss rate is 0 (system.cpu.icache.overall_miss_rate::total 0.000000, size=131072, assoc = 2) which is I believe impossible.
so my question is how to check why simulation of bzip2 never stops?
and how I can check that statistics for the icache is correct?
update: I tried to simulate another simple matrix multiplication program.
Program itself runs on my computer 7 mins.
Left it running whole night on gem5 but it didn't finish the simulation.
THis is my command line: build/ARM/gem5.opt -d matrixTiming configs/example/se.py --caches --l1d_size=32kB --l1d_assoc=1 --l1i_size=32kB --l1i_assoc=1 --cacheline_size=64 --cpu-type=TimingSimpleCPU -c tests/test-progs/matrixmul/bin/ARM/linux/mmMethod3_arm.o
Any idea what is the reason, or how i can find it.
Thanks for the help