I've been trying to get some results running bbench, using the development repository of gem5 and following the instructions over at http://www.m5sim.org/Bbench-gem5
I've been running them with atomic cpus to get a checkpoint right before the browser launches using the following: ./build/ARM/gem5.fast configs/example/fs.py -b bbench-ics --kernel=/home/gyessin/bbench_experiment_switch/dist_68/m5/system/binaries/vmlinux.smp.mouse.arm --num-cpus=2 --frame-capture --checkpoint-dir=bbench_capture --clock=0.75GHz --mem-size=512MB And that runs through everything without a problem. But when I try restoring from the checkpoint and switching over with this: ./build/ARM/gem5.fast configs/example/fs.py -b bbench-ics --kernel=/home/gyessin/bbench_experiment_switch_dual/dist_68/m5/system/binaries/vmlinux.smp.mouse.arm --num-cpus=2 --frame-capture --checkpoint-dir=bbench_capture --caches -s 5000000000 -r 1 --l1d_size=32kB --l1i_size=32kB --l2cache --l2_size=2048kB --clock=0.75GHz --mem-size=512MB The memory usage on my compute nodes will grow without bounds, eventually eating up all 8GB and then crashing. It's currently using about 40% of that memory after 17 hours. Is this a known problem with dual core configurations? Any thoughts on how to remedy this? Any help would be greatly appreciated.
_______________________________________________ gem5-users mailing list gem5-users@gem5.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
