By Noah | March 3, 2010
Greg Pfister at Perils of Parallel has a nice little tutorial up on the tradeoffs among superscalar, VLIW, and “Simple Multicore”. I guess you’ve got to care a bit about hardware for this to be worthwhile, but I’d encourage anyone with an interest in modern computer systems to take a look. It’s reasonably accessible even if you’re not a hardware expert, and it says a lot about the systems we’ll all be programming in the coming decades. Of course, these debates have been going on for at least 20 years, but it’s a very good posting (leaves out superpipeline, but I guess that’s not so popular anymore).
He also mentions that my friends in the IBM “fabs have managed to put DRAM on the same silicon as the POWER7 logic”. Getting DRAM densities and power consumption on on the processor chip is a big deal; I hadn’t noticed IBM had pulled that off.