Had Fujio Masuoka not invented flash memory for Toshiba in the 80s, this would be a very different blog. Forget your massive SSDs, your infinite digicam memory cards, and yes, forget even your Virgin Mary beating-heart thumbdrives. Masuoka-san isn't finished with us yet, though, and the next-level tech he has his sights on next could keep us busy posting for many years to come. Moore's law states that the amount of transistors able to be jam-packed into processors will double every two years or so. But if chip makers are going to keep this up, they're going to have to move beyond the quaint world of two dimensions and into the realm of 3D processors. Masuoka has made a deal with Singapore's Institute of Microelectronics to develop new 3D chip designs, which he will then licence through the Singaporean government. What 3D processors make possible, aside from simply jamming more transistors in to a smaller space, are huge speed gains due to a greater number of transistors being in closer proximity to one another. On a 2D chip, if a transistor on one end has to make a linkage with one on the other end, it naturally takes a bit longer than if the two were directly adjacent. A 3D model allows for much more efficient interconnection, as well as the potential for much faster on-chip RAM.
Stacking layers of silicon to form 3-dimensional circuits is not a new concept - Masuoka is certainly quite familiar with it already, as it's exactly what enables flash memory sizes to continue to grow so quickly. But as of now, 3D stacks have yet to become a production reality in the processor world, although big players like Intel and IBM have made several announcements and demonstrations this year indicating progress. [Institute of Microelectronics (Singapore) via Digital World Tokyo]