Do you see same type of variations in user mode IPCs? If user IPC are not much different then it seems that some kernel activity is affecting these IPC variations.
- Avadh On Tue, Nov 8, 2011 at 3:03 PM, [email protected] <[email protected]> wrote: > Sorry for the mistake, here is the output from the two. > > $cat marss_Ori_1/test.log |grep "insns/cyc" > Stopped after 28410444 cycles, 100000008 instructions and 415 seconds of > sim time (cycle/sec: 68458 Hz, insns/sec: 240963, insns/cyc: > 3.5198326361953374) > $ cat marss_Ori_2/test.log | grep "insns/cyc" > Stopped after 26290913 cycles, 100000006 instructions and 434 seconds of > sim time (cycle/sec: 60578 Hz, insns/sec: 230414, insns/cyc: > 3.8035957899217876) > > > [email protected] wrote: > >> Hello, >> >> I am confused by the different IPC obtained when I run exactly the same >> checkpoint. Could you please let me know if I made something wrong? >> >> >> Detials: >> 1. create checkpoint. >> script.sh >> #!/bin/bash >> # create checkpoint w1 >> # benchmarks: >> # 445.gobmk >> # 458.sjeng >> # 403.gcc >> # 444.namd >> >> #445.gobmk >> cd >> /root/spec2006/benchspec/**CPU2006/445.gobmk/run/run_**base_ref_x86_64-gcc43.0001 >> >> >> taskset -c 0 ./gobmk_base.x86_64-gcc43 --quiet --mode gtp < nngs.tst& >> >> #458.sjeng >> cd >> /root/spec2006/benchspec/**CPU2006/458.sjeng/run/run_**base_ref_x86_64-gcc43.0001 >> >> >> taskset -c 1 ./sjeng_base.x86_64-gcc43 ref.txt& >> >> #403.gcc >> cd >> /root/spec2006/benchspec/**CPU2006/403.gcc/run/run_base_**ref_x86_64-gcc43.0001 >> >> taskset -c 2 ./gcc_base.x86_64-gcc43 166.i -o 166.s& >> >> #444.namd >> cd >> /root/spec2006/benchspec/**CPU2006/444.namd/run/run_base_**ref_x86_64-gcc43.0000 >> >> >> taskset -c 3 ./namd_base.x86_64-gcc43 --input namd.input --iterations 38 >> --output namd.out& >> >> cd /root >> sleep 5m >> ./create_checkpoint w1 ; ./start_sim >> >> Then I just use the original marss (an old version) and run this >> checkpoint twice >> run command: >> qemu/qemu-system-x86_64 -m 8G -hda diskimage.qcow2 -loadvm w1 -simconfig >> running.cfg >> >> after the simulation done, the output from the terminal: >> Stopped after 58914145 cycles, 100000000 instructions and 550 seconds of >> sim time (cycle/sec: 107116 Hz, insns/sec: 181818, insns/cyc: >> 3.8035957899) >> >> I simply take IPC = insns/cycle = 3.8035... >> >> But in two times I run same simulation, the result is quite different. >> second is 3.5198. >> >> Is there anything wrong with my simulation? >> >> Thanks, >> >> Kuniors >> >> >> >> >> >> >> > > ______________________________**_________________ > http://www.marss86.org > Marss86-Devel mailing list > [email protected].**edu <[email protected]> > https://www.cs.binghamton.edu/**mailman/listinfo/marss86-devel<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
