Hello,

Probably the gem5 version I'm familiar with is different from yours, but
stats in system.cpu.iew represent the total number of executed
instructions, including speculatively executed and subsequently squashed
ones. In system.cpu.commit, only to correctely executed instructions are
counted (those whose state is made visible to the software), so they do not
take into account the speculatively executed and squashed instructions.

Hope it helps,

--
Fernando A. Endo, PhD student and researcher

Université de Grenoble, UJF
France



2014-03-14 0:42 GMT+01:00 Aditya Deshpande <[email protected]>:

> Hi,
>
> I am running X86 architecture in SE mode. I ran various applications for a
> single core system using TimingSimpleCPU cpu_type and observed following in
> the stats file:
>
> 1. Total number of committed instruction are given by
> system.cpu.committedInsts or sim_insts. Both these values are same
> 2. The integer and FP operation counts are given by
> system.cpu.num_int_insts, system.cpu.num_fp_insts respectively
> 3. The memory references are given by system.cpu.num_mem_refs
> 4. Branch control instruction count is given by
> system.cpu.num_conditional_control_insts
>
> From my understanding total number of instruction executed should be sum
> of integer, FP Mem references and branch-control instructions.
> But in the stats file,I observe that the
> system.cpu.committedInsts is always less than (cpu.num_int_insts +
> system.cpu.num_fp_insts + system.cpu.num_mem_refs +
> system.cpu.num_conditional_control_insts)
>
> Is that normal? What can be the possible explanation for the same?
>
> Regards,
> Aditya
>
> _______________________________________________
> gem5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
>
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to