Hello all,

I am curious if there is a mode you can put m5 into where each core has its
own block of distributed memory instead of one large set of shared memory
for all X cores.

A simple example of the algorithm I am trying to implement would be where
each core has 1/Nth set of the data and each finds the max value and then
they send these max values to one "master" core and this core identifies the
max of these set.  I would like to do this with each core only being able to
see its set of data.

Any models similar to this in m5?

-- 
Aaron S. Williams
Graduate Student
Arizona State University
Department of Electrical Engineering
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to