Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: J_W who wrote (26080)	7/31/1999 5:14:00 AM
From: John Stichnoth	Read Replies (1) \| Respond to of 93625

Related to the issue of latency and cost is the following, from The Economist. Questions for our engineer contributors--How much of the latency issue is mandated by the way chips are attached to their packages? And, will something like this take away some of the cost issues surrounding pin count? Sharp claws S A N F R A N C I S C O Computer Technology Search archive MICROPROCESSORS get faster every year, at a rate that has become boringly predictable. But processing speed is not everything. As anyone who has twiddled his thumbs waiting for a web page to download knows, bandwidth—the rate at which information can be transferred—is just as important. And although the bandwidth of connections between machines has improved steadily over the years, inside them it has lagged behind. Indeed, a chip's speed has become almost irrelevant for some data-intensive applications, because the chip spends more time waiting to receive (or send) bits of data than it does actually processing them. The problem is that every bit of data to be worked on by a microprocessor has to travel through one of just a few hundred wire connections. And these connections, which are usually made of gold, have proved hard to miniaturise. However, Donald Smith, David Fork and Andrew Alimonda, all at Xerox's Palo Alto Research Centre, now think they have managed the trick. Their invention, which they dub the “microclaw”, is a tiny finger of silicon that curves up out of the package housing a chip in order to connect with special pads on the chip's surface. A microclaw's size—a mere six-millionths of a metre across—means that each chip can have thousands rather than hundreds of connections with the outside world. Microclaws are made by coating the silicon of the package with a thin film made from an alloy of molybdenum and chromium. As it is laid down, this film builds up stress in the underlying silicon. The pattern of stress can be changed by varying the pressure at which the alloy is deposited. The silicon is then cut into fingers by microlithography. (This is also the process by which a chip's components are created; it uses a mixture of light and chemicals to etch away parts of the silicon's surface.) After that is done the fingers, or microclaws, curl back from the surface in response to the stress imposed by the alloy layer. By making cuts in the right places, the microclaws can be aligned with their respective bonding-pads on the surface of the chip. Applying a little heat then produces a firm electronic and mechanical connection. This has obvious advantages, such as high connection-density, which makes it useful, and compatibility with conventional microlithography, which makes it cheap. But the microclaw has something else in its favour too: low capacitance. Capacitance is a kind of electrical inertia. If a system has high capacitance, this means either that a signal must be sent through it at high power, or that it will move at a lower speed. In conventional chips, the wires are attached by balls of solder. Such connections have relatively high capacitance because this property is related to surface area—and the balls of solder are large. The microclaw, however, has low capacitance because it requires such a tiny pad to fix it to the chip. It therefore breaks the information bottleneck in two ways: by providing more connections and by speeding up the signals. Whether a busier processor makes for a more productive user remains to be seen.

To: J_W who wrote (26080)	7/31/1999 11:20:00 AM
From: richard surckla	Respond to of 93625

Jim, I also thought the Dell article was excellent. Much of it I didn't understand but the parts I did understand I knew were good. And with todays solid INTEL "GO!" News should take the questions out of "further delay FUD" that we've been reading about. Looking for a better monday than this past friday.