"That point was INTC was scoffing at IBM and SOI. Then Poof they have a better SOI mousetrap when they said a while back SOI doesn't bring anything to the table.
My Point was IBM was right all along.
"
My sense from reading your stuff over the past several months is that you have a naive and unsophisticated view of what chip R & D is all about. Further, you seem to think the views expressed here, e.g., those scoffing at AMD's claimed imminent leap ahead of Intel because of SOI licensed from IBM, represent some kind of "official verdict" of Intel.
Fact: Technology is advancing on many fronts, with many false starts, dead ends, side tracks, minor breakthroughs, and even major breakthroughs.
Fact: What has historically been important for the chip industry is the ability to manufacture a technology in high volume, with reasonably standard equipment, and reproducibly. Less important has been whether a technology is at the absolute bleeding edge.
Digression: I've spent my years since retirement from Intel dealing with a lot of wild and woolly thinkers in areas like cryptography, nanotechnology, AI, etc. Many of them lack experience in actual corporate projects. I like to remind them of these "promising technologies: Josephson Junctions, laser pantography, e-beam addressed memories, wafer-scale integration, integrated injection logic (I squared L), bubble memories, Al Tasch's radical new memory cells every couple of months, ferroelectric memories, GaAs, molecular beam epitaxy, SOS (ha!), and on and on. These are the technologies gracing the covers of so many issues of "Electronics" (a once-famous magazine), now sort of superceded by rag sheets like The Register (double ha).
The point is that what gets talked about is not necessarily what will be used in products. The list of technologies above, all of which were going to revolutionize electronics, supposedly, needs to be looked at with a jaundiced eye.
Most radical technologies will never make it out of the lab. This should not surprise anyone.
Fact: SOS and SOI have been researched (and developed) by many companies. RCA did moderately well with an SOS version of the 1802 uP. Some memories were made with SOS. (I did an analysis of them for a 1977 internal Intel report on alpha particle and cosmic ray effects, and I believe I mentioned SOS sensitivity in a 1978 public paper...so SOS and SOI were definitely being researched back then. Some SOS devices were designed into satellites and space probes lofted in the mid-80s.)
Fact: Companies use the hype of the new technologies, such as those listed above, to gin up interest in their products and their overall future. This has been going on for many decades...tours of the research labs, interviews with scientists, etc. The Nobel prizes collected by Bell Labs didn't exactly help ATT, did it?
Don't believe the hype. Look to the actual products, the actual revenues. Look to _some_ extent at roadmaps, because high tech investing is more than just a Benjamin Graham type of analysis of instantaneous rates of return. But don't the "headlines" about "world's smallest transistor" or "world's fastest chip" without looking much more deeply.
I'm not interested in some sophomoric deconstruction of whether Mark Bohr "dissed" SOI or not. Whether he did or not, even as Portland TD was working on SOI, is neither here nor there. He and Marcyk and all the others can't talk about things they're not released to talk about, so most interviews will be "noncomittal" about what other companies are doing. (I haven't seen the interviews being alluded to, but from snippets thrown around here they sound just like the typical "what do you think of this technology?" questions that journalists pose. Much more interesting would be seeing the discussions about SOI which have been going on inside Intel for the past several years. As I said, many of today's technologists at Intel were well aware of SOS and SOI work, and even led projects. It's naive to think Intel was asleep. The upcoming paper at IEDM indicates they weren't, that they had a large effort going on for years. Which doesn't surprise me.
Much of science, and this includes the chip industry, is now operating on the basis of press releases. The latest "cloning" hoopla is a good example. Other cloning groups are pointing out that the "first example of human cloning" is not nearly as important or impressive as the breathless reports on CNN and Oprah would have us believe. "World's fastest transistor!" has been a staple of the headline squad for as long as I've been in the chip business.
(I remember the impressive reports in Electron Devices and in the electronics press about how fast the ring oscillators were running at Bell Northern, or at Mostek, and so on. It was sometimes scary to hear how fast these things were running...and the press reported new breakthroughs from El Masry's group, or Tasch's group, on a monthly basis. And yet did these things end up making a difference? The ability to design, manufacture, and market a device turned out to be a lot more important that "world's fastest transistor" bragging rights.)
Finally, you, Milo, really need to take a more nuanced and sophisticated look at technology roadmaps. It is not as simple as "AMD says SOI is good, Intel doesn't have it, now Intel is announcing they _will_ have it in 2005. Intel is copying AMD, again."
This is a naive view of a complex subject. Get up to speed.
--Tim May |
