You need to put the checkpoints on the same image. You are essentially running two simulations back to back.
On Mon, Jan 25, 2016 at 7:31 AM, 조민신 <[email protected]> wrote: > Hello, all. > > > > I am trying to run marss.dramsim for multicore simulation. First, I made > ooo_2_th core configuration with shared L2 cache, and checkpoints of two > PARSEC benchmarks (bodytrack and blackscholes) on two parsecROI.qcow2 > provided from MARSSx86 official website. after that, I also modified > util.cfg file. > > > > The following is my util.cfg file. > > > > ----------------------------------------------------------------- > > [DEFAULT] > > marss_dir = /home/msc/marss.dramsim > > util_dir = %(marss_dir)s/util > > img_dir = %(marss_dir)s > > > > qemu_bin = %(marss_dir)s/qemu/qemu-system-x86_64 > > > > default_simconfig = -kill-after-run -quiet > > > > [suite parsec_bench] > > checkpoints = bodytrack, blackscholes > > > > [run parsec-roi] > > suite = parsec_bench > > images = %(img_dir)s/parsecROI_body.qcow2 > > %(img_dir)s/parsecROI_black.qcow2 > > memory = 4G > > simconfig = -logfile %(out_dir)s/%(bench)s.log > > -stats %(out_dir)s/%(bench)s.yml > > -machine ooo_2_th > > -corefreq 3G > > -dramsim-device-ini-file ini/DDR3_micron_32M_8B_x4_sg125.ini > > -stopinsns 1000000 > > %(default_simconfig)s > > --------------------------------------------------------------------- > > > > Next, I run the simulation using run_bench.py. > > > > ./util/run_bench.py parsec-roi -d ben_out > > > > It runs sequentially on my linux terminal, and I saw those simulation > results which is in the files named bodytrack.log and blackscholes.log. > However, I can't be sure whether It is multicore simulation or not because > ipc of the results are not same. > > Here is my simulation results > > > > [blackscholes.log] > > > > user.base_machine.ooo_0_0.thread0.commit.ipc = 0.185967 > > user.base_machine.ooo_1_1.thread0.commit.ipc = 0.582412 > > kernel.base_machine.ooo_0_0.thread0.commit.ipc = 0.119389 > > kernel.base_machine.ooo_1_1.thread0.commit.ipc = 0.241613 > > total.base_machine.ooo_0_0.thread0.commit.ipc = 0.121494 > > total.base_machine.ooo_1_1.thread0.commit.ipc = 0.485627 > > Received simulation kill signal, stopped the simulation and killing the VM > > > > [bodytrack.log] > > > > user.base_machine.ooo_0_0.thread0.commit.ipc = 0.694209 > > user.base_machine.ooo_1_1.thread0.commit.ipc = 0.687764 > > kernel.base_machine.ooo_0_0.thread0.commit.ipc = 0.0408346 > > kernel.base_machine.ooo_1_1.thread0.commit.ipc = 0.0127316 > > total.base_machine.ooo_0_0.thread0.commit.ipc = 0.291307 > > total.base_machine.ooo_1_1.thread0.commit.ipc = 0.361208 > > Received simulation kill signal, stopped the simulation and killing the VM > > > > > Also, I made my checkpoints with the following way on QEMU terminal of > parsecROI.qcow2 (I downloaded it from official MARSS website) > > > > 1. cd parsec-2.1 ; source env.sh ; cd .. > > 2. ./create_checkpoint bodytrack; (bench command "parsecmgmt ~~" > ;./stop_sim > > 3. finish checkpoint creation. > > > > is it correct multicore simulation? (I think if it is correct multicore > simulation, the ipc results of both should be same) If not, please let me > know how I can run correct multicore simulation. > > > > > > Thanks in advance > > > > > > > > > > _______________________________________________ > http://www.marss86.org > Marss86-Devel mailing list > [email protected] > https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel > >
_______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
