Foresight Nanotech Institute Logo
Image of nano

Moore’s Law: good to go…

I would note of interest an article (just look at the pictures) from the recent Intel Developers Forum (IDF) to see where Intel intends to go. They are planning for chips based on a 22nm process by 2011. On a more subtle note they appear to be seriously considering 'die-stacking' to allow the 'chips' to grow in 3D. This has the potential of significantly reducing interconnect delays between processors and memory and/or stacked multi-processors [see Note 1]. That would imply significant performance improvements without having to figure out the messy process of trying to wire chips with buckytubes or DNA.

The only thing one can say is that the next decade is definitely going to be the "nanodecade" [see Note 2]. Note 1. I can see stacking the processors on top of the memory or perhaps interleaving the processors between the memory but stacking multi-core processors (think 4×4 or 8×8) suggests that one is going to have to go out and find a fire-hydrant for the water supply to keep the PDA or cell-phone cool if you expect to be doing conversions of entire movies from low-definition to high-definition in real time (which requires all of the processors be active rather than just sitting there…). I can see the coming complaints along the lines of "Your last email just melted my cell phone! You owe me $5 to purchase a new one."

Note 2. I would almost have expected "nanodecade" to be a new word but Google finds several references. (The first of which involves one strange fictional usage I wouldn't have quite anticipated…)

4 Responses to “Moore’s Law: good to go…”

  1. MikeMcCauley Says:

    A sign of the times

    Soon we wonít be asking how many transistors are on a chip. Weíll be asking how many processors are on a chip.

  2. rumi54 Says:

    Re:A sign of the times

    Can someone explain whether Intel actually has a way to build multi-level processors? I looked at the presentation, but didn't know if any advances had been made in multi-levels? What would some of the problems be?

  3. RobertBradbury Says:

    Re:A sign of the times

    It isn't clear what technology Intel may have. The two biggest problems (IMO) are (a) communication between stacked chips and (b) heat removal.

    One can avoid the headaches of the heat removal problem by stacking the heat producing devices on top of those that don't produce heat (e.g. DRAM).

    With regard to interchip communication even if Intel doesn't know how to do it, Sun may (see here and here). Its known as "proximity communication" or "capacitive coupling". It would appear (at least to me) what makes the approach great is that allows you to stick the chips together in a stack without having to heat them up to get the solder to melt (which of course tends to fry the chips…).

  4. jayakar Says:

    Molecular quantum-computing device is far away

    Impacts on the Mooreís law by the Molecular quantum-computing device developments may be the reality of the future, but not of the near-future.

Leave a Reply