You can have a dual-core configuration. If you look in config/default.conf, you can see that we provide ooo_2_th and shared_l2. ooo_2_th is a configuration that uses SMT. shared_l2 is a multicore design that has as many cores that you compile with.
It sounds like you'll want to use either shared_l2 or private_L2. For run_bench.py, the -n option means how many instances of Marss you want to run at once. However you set the simconfig options when you run marss, that's what will go in util.cfg or whatever you choose to call it. I have one setup for spec2006, so I do something like: ./util/run_bench.py -c specref.cfg -n 16 -d output/ which runs 16 different benchmarks at once, which are specified in my specref.cfg I hope this answers your questions. On Tue, Feb 17, 2015 at 6:41 PM, Fabio Ghiozzi <[email protected]> wrote: > Dear all, > First of all i would like to thank you for the incredible support I had > from this mailing list. I'm writing to you because I'm getting some trouble > with the marss.dramsim2 simulator and I'd like to understand how it is > possible to perform the simulation as I want to. > > I want to perform ONE simulation with two single_thread cores running two > different benchmarks (checkpoints) on a disk image. This is my objective. > > So, my first question is about the difference between contexts and cores. > If I want to achieve that simulation environment, I would need 2 > single_cores (from the default machine configurations file) that have 1 > context. But, as soon as I try to compile with c=2, the single_core > configuration file is avoided. So what I cannot understand is the > relationship between cores and contexts. Why I can not have TWO single_core > cores, having one context each? > > The second question is about the run_bench,py script. In order to achieve > my objective, how can I set my util.cfg file and the run_bench parameters? > I've tried to use -n 2 but this is different from what I want because it > launches two SIMULATIONS in parallel, outputs TWO dramsim.log and the TWO > simulations I think are independent. What I think I should do is to have > ONE simulation running two benchmarks in parallel on two distinct cores. > > I hope I've been able to explain my concerns and I hope to get some help. > > Thank you very much. > > Fabio > > _______________________________________________ > 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
