Re: [Gems-users] Question on MOESI_CMP_token


Date: Thu, 27 Mar 2008 11:23:16 -0500
From: "Mike Marty" <mike.marty@xxxxxxxxx>
Subject: Re: [Gems-users] Question on MOESI_CMP_token
by default, the L2 is banked to the number of processors per chip.  It
is still logically shared

--mike


On Thu, Mar 27, 2008 at 5:38 AM, KS Chow <g0706256@xxxxxxxxxx> wrote:
>
>
>
>
> Hi guys, I have a question on L2 caches on MOESI_CMP_token.
>
> I assumed that under this protocol the L2 would be shared among processors
> in the same socket/chip – but when I took a look at the ruby dump file for a
> 2 processor 1 socket system (a dual-core):
>
>
>
> L2Cache_L2cacheMemory numberPerChip: 2
>
> cache_size_Mbytes: 8
>
>
>
> Compared to a 1P-1S system
>
>
>
> L2Cache_L2cacheMemory numberPerChip: 1
>
> cache_size_Mbytes: 16
>
>
>
> Seems like I have 2 separate L2's for my dual core. I have specified
> g_NUM_PROCESSORS as 2 and g_PROCS_PER_CHIP as 2 for the dual-core system,
> other settings on default.
>
> The same goes for quad-cores – seems like my L2 is split and each gets
> 16MB/4.
>
> Am I setting my system right or am I just misinterpreting the dump file?
>
>
>
> Thanks/Regards.
>
>
>
> Chow Kian Sim
>
> National University of Singapore
>
>
> _______________________________________________
>  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.
>
>
>
[← Prev in Thread] Current Thread [Next in Thread→]