The article implies we could hit 48-cores in 5-8 years before problems (memory contention) set in. After that I guess something new needs to come along. Also t

The article implies we could hit 48-cores in 5-8 years before problems (memory contention) set in. After that I guess something new needs to come along. Also that might get us close to when Moore's Law itself ends, when something beyond integrated circuits needs to step in. I wrote about multi-core and Moore's law recently: http://www.kmeme.com/2010/09/clock-speed-wall.html http://www.kmeme.com/2010/09/clock-speed-wall.html