Re: [Gems-users] Strange Ruby statistics


Date: Thu, 03 Aug 2006 10:26:00 -0500
From: Mike Marty <mikem@xxxxxxxxxxx>
Subject: Re: [Gems-users] Strange Ruby statistics

Well, I have another question: how can I understand the various cache
policies (like insertion, migration, search, L2 allocation and so on)
implementation?  Are they implemented at "low level" in network
components like controllers, directory and so on descripted in SLICC
or is there some more "high-level" code that implements them?

For most protocols, cache allocation and search is specified in SLICC.

For MOESI_CMP_NUCA, it is a combination of SLICC and a combination of C++ code in $GEMS/ruby/nuca

--Mike


[← Prev in Thread] Current Thread [Next in Thread→]