Greg thanks for the reply,
It seems that you are right, but I have (already :-) made a few
small changes
(in DirectoryMemory.C, StateMachine.C, SymbolTable.C and RubyConfig.h)
in addition to map_Address_to_DirectoryNode function in
RubySlicc_ComponentMapping.h and now, I think, I can support
FILE_SPECIFIED
topologies with corresponding changes in RubyConfig.h
Thanks anyway. I'll keep your approach in mind in the future.
Stamatis
How about just redefining the mapping of addresses to directories, and
leaving everything else the same? You'd still create one directory per
processors, but the directories that don't get mapped will go unused.
That would be easier than trying to redo all the configuration stuff.
...Greg
Stamatis Kavadias wrote:
Hi,
I was wondering if it is possible to use MOESI_SMP_directory
protocol with fewer directory controllers than the number of
chips. The reason is that I am using the multi-chip configuration
to model large CMPs (>> 16 processors) and don't want to model
the same number of memory/directory controllers as processors.
I assume that I will have to re-define the mapping of addresses to
directory controllers - is that correct ? Anything else ?
By the way, I am using MOESI_SMP_directory because I need a
private L2 cache which other protocols do not have.
Thanks...
Stamatis
_______________________________________________
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.
|
|