If I recall correctly, the different MessageBuffers are virtual
networks. The same response virtual response network is used for
L1->L1 responses as they are L2->L1 responses.
--Mike
2009/10/9 shanshuchang <shanshuchang@xxxxxxxxx>:
> Hi all,
> I am puzzled with data and state transition in such a situation in
> MOESI_CMP_directory:
> Assuming a cache miss in L1, so ask for data to L2 which has state
> ILS. According to the protocol,
> it just forwards the request to L1' who has a shared copy. When the L1'
> receives the Fwd_GETS request,
> it sends the L1 with the shared data by action "e_sendData".
> However, after looking into the codes of e_sendData, I found that the L1'
> would enqueue the ResponseLocal Data
> into MessageBuffer responseNetwork_out,which as far as I know is used to
> communicate between L1' and L2'.
> So, I am really confused how can the requestor L1 get data from the
> sharer L1' in such a situation?
>
> _______________________________________________
> 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.
>
>
>
|