Hi All,

I am facing the following problem: when I run a simulation without
checkpoint interruptions the stats.txt gives me a system.cpu.numCycles, but
when I restore from a specific checkpoint the stats.txt gives me a
completely different value for system.cpu.numCycles. For example:

For Spec 2006 bzip2 without checkpoint gives me system.cpu.numCycles =
7,651,294,711
and for for the same simulation starting from a checkpoint gives me
system.cpu.numCycles = 1,612,799,132.

My command line parameters are:

to start the simulation

gem5/build/ARM/gem5.fast \
    --stats-file=bzip2-stats.txt \
    --dump-config=bzip2-config.ini \
    --json-config=bzip2-config.json \
    --redirect-stdout \
    --redirect-stderr \
    --stdout=bzip2-stdout.txt \
    --stderr=bzip2-stderr.txt \
gem5/configs/example/se.py \
     --cpu-type=arm_detailed -n 1 \
     --sys-clock=400MHz \
     --cpu-clock=800MHz \
     --mem-size=512MB \
     --mem-type=lpddr2_s4_1066_x32 \
     --caches --l2cache --cacheline_size=32 \
     --cmd="$spec_dir/401.bzip2/build/arm/bin/gc/bzip2.arm" \
     --options="$spec_dir/401.bzip2/data/ref/input/text.html 1" \
     --max-checkpoints=1000 \
     --take-checkpoints=0,943991943750

And to restore from N = 11, checkpoint = cpt.9439923598750

gem5/build/ARM/gem5.fast \
    --stats-file=bzip2-stats.txt \
    --dump-config=bzip2-config.ini \
    --json-config=bzip2-config.json \
    --redirect-stdout \
    --redirect-stderr \
    --stdout=bzip2-stdout.txt \
    --stderr=bzip2-stderr.txt \
gem5/configs/example/se.py \
     --cpu-type=arm_detailed -n 1 \
     --sys-clock=400MHz \
     --cpu-clock=800MHz \
     --mem-size=512MB \
     --mem-type=lpddr2_s4_1066_x32 \
     --caches --l2cache --cacheline_size=32 \
     --cmd="$spec_dir/401.bzip2/build/arm/bin/gc/bzip2.arm" \
     --options="$spec_dir/401.bzip2/data/ref/input/text.html 1" \
     --checkpoint-restore=11 \
     --restore-with-cpu=arm_detailed \
     --max-checkpoints=1000 \
     --take-checkpoints=9439923598750,943991943750

I would like to know if I am doing something wrong or is it a gem5 issue?

best regards,

Marcelo
_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to