Hello, I am trying to understand some of the timing behaviour of a multi-threaded application. I am running gem5 in FS mode and I can obtain two useful files "stats.txt" and "system.tasks.txt". Throughout the lines of the two files, I could find three ways of referring to a cpu/core: (i) "cpu" suffixed with a natural number (cpuN), (ii) "cpus" suffixed with a natural number (cpusN), and (iii) "cpu_id" equals some natural number (cpu_id=N).
The following are (non-contiguous) lines from the "stats.txt" output file. system.cpu0.dtb.hits 6 # DTB hits system.cpu1.pwrStateResidencyTicks::OFF 203472000 # Cumulative time (in ticks) in various power states system.cpu2.itb.inst_hits 0 # ITB inst hits system.cpu3.itb.flush_tlb 2 # Number of times complete TLB was flushed system.switch_cpus0.fetch.rateDist::samples 312140 # Number of instructions fetched each cycle (Total) system.switch_cpus1.rob.rob_reads 0 # The number of ROB reads system.switch_cpus2.iew.iewIdleCycles 0 # Number of cycles IEW is idle system.switch_cpus3.itb.write_accesses 0 # DTB write accesses The following are lines from the "systems.tasks.txt" output file. tick=3203692196500 5 cpu_id=0 next_pid=1107 next_tgid=1107 next_task=runscript tick=3203956858500 6 cpu_id=0 next_pid=1 next_tgid=1 next_task=init tick=3204033514000 3 cpu_id=0 next_pid=-1 next_tgid=-1 next_task=kernel My understanding is that "cpuN" in the "stats.txt" file refers to [0..N] processor cpus/cores specified by the flag "-n N" that is passed to the simulation script "fs.py". My guess is that "cpusN" refers to the same entity but what is the "s" doing there? To mean "switch"? Finally, from my reading around, it looks like "cpu_id=N" is not related to the cpus/cores used to simulate a program but rather to the cpu model used in simulation e.g. "--cpu-type=O3_ARM_v7a_3" has "cpu_id=0". My question finally is: "is my understanding of the meaning of the three cpu entities above correct"? Thanks, Amine. -- marref.org
_______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
