Billion-Transistor Architectures: There and Back Again. This is a short paper that reviews the predictions made seven years ago in an IEEE Computer special issue on future CPU architectures, and compares them to what has actually happened. It is a nice general overview of the current state of the art in CPU design. As an aside, I think that CPU architecture research is dead and it is unlikely that any of the architectural changes described in this paper will ever become mainstream. The reason of my belief is that the market won't support anything that is not x86, but I would love to be proven wrong.