Quote Originally Posted by Luka_Aveiro View Post
As far as we know, no L2 cache.
huh? but dont they need at least a tiny piece of l2 to NOT break cuda support and all?