[forwarded message originally from Shae Matijs Erisson on 2006-10-18.] Kragen Javier Sitaker <[EMAIL PROTECTED]> writes:
>> On Wed, 11 Oct 2006, Kragen Javier Sitaker wrote: >> > I think that in general non-shared-memory multi-CPU computing is going to >> > be really important [...] I still haven't gotten around to subscribing to the various Kragen lists so I haven't read the original post, but I had an idea on how to make this work... Why not etch dies with either tiles of octagons and squares or just hexagons? In the first case, the large octagon is RAM and the small square is CPU, and each edge is a commucations link to the chunk next to it, giving you a highly parallel computer where each CPU has links to four banks of RAM, and each bank of RAM has links to four CPUs and four other banks of RAM. http://www.missiontilewest.com/styles/nantucket/n32.jpg (octagon & square) For the second, you can pack more hexagons than squares into a circle: http://mathworld.wolfram.com/CirclePacking.html So switching to hexagonal CPUs or just circles with hexagonal packing would increase the yield simply because fewer of the tiles would be hanging off the edge of the die. I got this idea from reading about the Cell cpu and what drops the yield on silicon dies. One of the loss factors is diamond saws cutting into the chips accidentally when they saw the die apart. So, why saw it apart, and why not sell an entire die that mostly works? If kerf ('saw off space'[1]) is currently left around the edges of dies, taking that out of the design would also save a bit more space. So, mount the entire die on a socket and price it according to how much works. That way nearly every die could be sold for some money, and nearly everyone could afford a computer, as low yield dies would be cheap, and high yield dies expensive. [1] http://en.wikipedia.org/wiki/Kerf -- I've tried to teach people autodidactism, | ScannedInAvian.com but it seems they always have to learn it for themselves.| Shae Matijs Erisson
