Dear all, I am attempting a multi-threaded simulation in x86 ISA in SE mode after generating the binary by using m5threads library. When doing simulation with o3 cpu (i.e., detailed/Derivo3CPU), i am confused as to how the instructions are allocated to different cpu's (the statistics of which can be viewed in the stat file as system.cpu0......, system.cpu1........ and so on). I have two specific questions in this regard: a) Even if i have only 2 threads (cpu's) in my binary, i get statistics for 3 cpu's (system.cpu0......, system.cpu1........ and system.cpu2.....). My understanding is that cpu0 must be doing some system-related tasks while the execution is done on the other 2 cores(cpu1 and cpu2). Is this understanding correct? The only thing compulsory to my work is to use shared memory by all threads(cpu's). b) Furthermore, if i have same number of instructions in each of thread in the binary, after simulation, in the statistics file, number of instructions simulated in each cpu (system.cpu1.... and system.cpu2.....) are significantly different. For my work, i need to have a correspondence between instructions present in the binary and the order of occurrence of these instructions in the simulator. I have read the facts on http://gem5.org/Frequently_Asked_Questions#How_do_I_run_multiprogram_workloads_on_M5.3F , however, i am not getting clarity.
Any assistance would be greatly helpful. Thank you so much.... *With Best Regards,* *Binod Kumar* *IIT Bombay*
_______________________________________________ gem5-users mailing list gem5-users@gem5.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users