Re: [Gems-users] question on silence drop

[Mike Marty <mikem@xxxxxxxxxxx>]

The data isn't written back from E.  Look at 
qq_sendDataFromTBEToMemory.  If the data is clean, then the message is 
changed to a control message.
Thus in this protocol, E replacement is not totally silent as it sends a 
small control message.  In doing so, it eliminates a potential race 
where the Directory forwards a request to a E that might have silently 
replaced.
--Mike

Liqun Cheng wrote:
> Hi,
>
> I am confused about the transition in MOESI_SMP_directory-cache.sm 
> <http://cache.sm>, line 835. Why a clean line (E state) needs to be 
> written back?
> transition({E, M}, L2_Replacement, MI) {
>     i_allocateTBE;
>     d_issuePUTX;
>     rr_deallocateL2CacheBlock;
>   }
>
> BTW, what's the difference between M and MM (locally modified)?
>
> thanks
> Liqun
> ------------------------------------------------------------------------
>
> _______________________________________________
> 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.
>
>