Re: [Gems-users] MOESI state transitions

["Huang He" <huangh@xxxxxxxxx>]

If the directory infomation of replaced L2 line is stored in memory, it is OK.

Otherwise, say the directory isn't stored in memory, if the replaced line is refilled into L2 again,
another caching agent may acquire exclusive state without snooping the previous owner. This
violates cache coherence.


Huang He
Processor Architect

----- Original Message ----- 
From: "Edward Lee" <edwl202@xxxxxxxxx>
To: "Gems Users" <gems-users@xxxxxxxxxxx>
Sent: Monday, July 26, 2010 10:01 AM
Subject: [Gems-users] MOESI state transitions


> Hi,
> 
> In MOESI_SMP_directory protocol, after an L2 replacement E state
> issues a PUTX. As far as I know, E state has the clean data which
> means memory also has the most recent copy. So, why not just do like
> what S state does in this scenario which is, r(Deallocate L2 cache
> block. Sets the cache to not present, allowing a replacement in
> parallel with a fetch) and then go to I state?
> 
> Any ideas?
> 
> Regards,
> 
> Ed
> _______________________________________________
> 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.
> 
> 
> 
>