biswabandan panda <biswa.uce <at> gmail.com> writes: > > > Hi, It's really difficult to change the coherence protocol but you can try some older versions of m5 where coherence protocol was a separate module but you may end up with some old bugs. > > On Wed, Apr 6, 2011 at 11:46 AM, prasanth_iitd <mcs092094 <at> cse.iitd.ernet.in> wrote: > > I have read m5's new documentation about Classic Memory System > and Ruby and found that Classic has an abstract cache coherence > protocol implemented which will be difficult to change. > Ruby's documentation has many examples but all were directory > protocols or token-based. > I was expecting there would be atleast one example of a snoopy > cache-coherence protocol, but wasn't able to find it. I don't > know whether i missed it or it isn't there? > > > > _______________________________________________ > m5-users mailing listm5-users <at> m5sim.orghttp://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > > > -- thanks®ardsBISWABANDAN PANDAM.S.(RESEARCH SCHOLAR)RISE LABIIT MADRAShttp://www.cse.iitm.ac.in/~biswa/ > > > > > _______________________________________________ > m5-users mailing list > m5-users <at> m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
thanks for replying. So, taking an old version of m5 and modifying the protocol defined there into a snoopy protocol with cache-to-cache transfer might turn out to be easy. Do u have any suggestions about the old stable versions which could suit the needs? How about writing a snoopy coherence protocol using ruby? Is that possible? and how difficult it would be? (i think this might involve making the state diagrams for different controllers as given in Ruby which seems difficult) _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
