4 ms·

What is the next big thing in cpu design? Graphene based cpus? Light based optical cpus Thz? 3d stackable cpus?

• agumonkey 13y ago
  pick all three
• Tuna-Fish 13y ago
  Of those, 3d integration is the one closest to market. It's already used in some cellphone chips, and will likely hit mass market within 3-5 years. Fully light-based CPUs won't happen for a while, but we are getting to the point where having some optical components to manage long-distance communication (within the chip, and off-chip to memory) begins to get closer to feasible. Think less than 10 years. Graphene based cpus are not near at all, as we are currently at the point where we are merely studying the qualities of the material. The time from that to mass production is measured in decades.
  • politician 13y ago
    What's your take on Crossbar RRAM? It sounds fantastic, but does it work?
    • Tuna-Fish 13y ago
      Maybe? All we have now are glorified press releases. They paint a very rosy picture, however, as they are press releases, we only get to see the rosy side of the picture. The principle of Crossbar RRAM is very fantastic indeed -- take the ideas of RRAM worked on by other companies and develop an implementation of them that works on current fabs, tools and materials. However, a lot can go wrong at nano scale manufacturing, and a lot of the things they are doing have not been characterized before by anyone else. Frankly, we won't know how manufacturable it really is before it's in mass production. Of the next-gen memory options, RRAM is the least disruptive of current systems. This is because it's properties are sufficiently similar to flash (specifically, it needs wear leveling, preventing use as universal ram) that if it wins, it will be used exactly as flash is used, only being denser and faster.