Yep, it's pretty confusing. What is actually even more confusing is
that this variable (L2_CACHE_NUM_SETS_BITS) changes, once I initialize
ruby, to its "per core" value! In other words, i set it as 16 and
right after initialization i find out that it's 10
[10=16-log(num_cores)]
simics> ruby0.setparam L2_CACHE_NUM_SETS_BITS 16
simics> ruby0.init
Ruby Timing Mode
Creating event queue...
Creating event queue done
Creating system...
Processors: 64
Creating system done
Ruby initialization complete
simics> ruby0.listparam
L2_CACHE_NUM_SETS_BITS : 10
anyway, i think it's clear to me now. Thanks Mike.
-Kostas
On Mon, Apr 21, 2008 at 11:01 AM, Mike Marty <mike.marty@xxxxxxxxx> wrote:
> RubyConfig.C will take the specified L2 parameters and divide it by the
> number of banks per chip. So right now, the L2 size is specified for the
> entire chip (spanning several shared banks). If you prefer specifying the
> L2 size per bank, comment out the single line of code in RubyConfig.C. In
> retrospect, doing it this way was probably a bad decision and leads to
> confusion.
>
> --mike
>
>
>
>
> On Sun, Apr 20, 2008 at 9:22 PM, Konstantinos Aisopos <kaisopos@xxxxxxxxx>
> wrote:
> > thanks Mike. does the same happen with L2? If yes, isn't the default
> > value huge?? 16MB L2 per core?
> >
> > L2_CACHE_ASSOC 4
> > L2_CACHE_NUM_SETS_BITS 16
> > => L2 size = 2^16 * 4 * 64 = 16MB
> >
> > thanks,
> > Kostas
> >
> >
> >
> >
> > On Sun, Apr 20, 2008 at 10:10 PM, Mike Marty <mike.marty@xxxxxxxxx> wrote:
> > > If L1_CACHE_ASSOC is 4 and L1_CACHE_NUM_SETS_BITS is 8, then each L1
> cache
> > > instance is 2^8 * 4 * 64 bytes
> > >
> > >
> > >
> > >
> > > On Sun, Apr 20, 2008 at 1:26 PM, Konstantinos Aisopos
> <kaisopos@xxxxxxxxx>
> > > wrote:
> > > >
> > > >
> > > >
> > > > hi all,
> > > >
> > > > I have a quick question about some configuration parameters. I
> > > > simulate a MESI_SCMP protocol with 64 cores on one single chip.
> > > >
> > > > my question is: if LX is L1 or L2 and we have these parameters:
> > > > LX_CACHE_ASSOC 4
> > > > LX_CACHE_NUM_SETS_BITS 8
> > > > do these parameters represent the "total LX cache capacity" or the
> > > > "per core LX cache capacity"?
> > > >
> > > > in other words is the size of LX *per core*..
> > > > 4 * 2^16 = 256K entries => 256K * 64bytes/line = 16MB???
> > > > or.. (4 * 2^16) / 64 = 4K entries => 4K * 64bytes/line = 256KB???
> > > >
> > > > thanks,
> > > > -Kostas
> > > > _______________________________________________
> > > > Gems-users mailing list
> > > > Gems-users@xxxxxxxxxxx
> > > > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > > > Use Google to search the GEMS Users mailing list by adding
> > > "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
> > > >
> > > >
> > >
> > >
> > > _______________________________________________
> > > Gems-users mailing list
> > > Gems-users@xxxxxxxxxxx
> > > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > > Use Google to search the GEMS Users mailing list by adding
> > > "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
> > >
> > >
> > >
> > _______________________________________________
> > Gems-users mailing list
> > Gems-users@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > Use Google to search the GEMS Users mailing list by adding
> "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
> >
> >
>
>
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding
> "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.
>
>
>
|