Robotics, and Age Ameliorization
October 11, 2002
For those who are becoming terminally bored with my gerontological teething exercises, I'm including a few other topics.
One is computer progress.
Intel has recently let it be known that the company expects to achieve one terops computing speeds, and perhaps, 1,000 gigahertz (1 terahertz) clock speeds in 2010, eight years from now. In the meantime, IBM, Sony, and Toshiba are presumably toiling away to produce "The Cell" by 2005. The Cell is their 1 terops chip that can be networked to form distributed supercomputers. And while that's happening, nVidia allegedly has graphics co-processor chips in their top-of-the-line video boards that will already churn out one trillion graphics calculations a second. nVidia has made noises about wagging the dog... about using their super-powerful chips for the main CPU, and relegating the Intel engine to operating the printer, the keyboard, the mouse, and other secondary functions. I haven't heard anymore about that, recently.
Actually, the normal, natural course of progress would be for Intel to reach 3 GHZ by December, which it's on track to do, and then work its way up to 10 GHz by 2005. (Actually, Intel has a portion of its 3 GHz Pentium 4 running at 6 GHz, so 10 GHz isn't that ambitious, if that's how we define 10 GHz.)
Another refinement that Intel is adopting is hyperthreading, which will boost throughput by anywhere from 30% to 100%, depending upon who's doing the talking.
Farther out, Intel expects to mount a billion transistors on a chip by 2010, and the Semiconductor Industry Association's (SIA's) technology roadmap calls for 8 billion transistors on a chip by 2016, with features that are 22 nanometers wide, compared to 130 nanometers today. Beyond that point, conventional chips may begin to use double- or triple-gated transistors. And beyond conventional silicon technology lies the hope of more exotic nano- and molecular techniques, using, perhaps, carbon nanotubes.
In other words, there's still no end in sight, at least through 2016.
The current plan calls for a shift from today's130-nanometer design features to 90-nanometer features in 2003, to 65 nanometers in 2005, and to 45 nanometers in 2007. Simple extrapolation would lead to 35 nanometers in 2009, and 22 nanometers in 2011--five years ahead of the SIA roadmap. That might happen. The industry is already well ahead of the 15-year SIA roadmap prepared in 1997.
Twenty-two nanometer design rules should get us to 16 gigabits of RAM on a chip, and maybe to 64 gigabits of RAM on a chip. And that should eventually enable a terabyte of RAM to be sold for several hundred dollars.
Hard drives store data in magnetic films.
IBM has just announced a technique, patterned
perpendicular magnetic films, that can lead to 10-terabyte disk
drives. This technique has been under development for a while, and might
be expected to happen in the 2008 - 2010 time frame. The important news
is that disk drives can become larger than they are now, and can be used
to store HDTV movies, and other, more-important data.
How far this can go is something I really can't guess. The theoretical limit of 62.5 gigabits per square inch for magnetic storage density, defined by IBM in 1991, is about to be exceeded, if it hasn't already been happened. My personal target is about 100 terabytes for robotics purposes, However, ten $100 disk drives would reach that storage level, and might be available by 2010 for $1,000.
We are certainly living through history in the making, as computer capabilities continue to increase exponentially. Sooner or later, it will have to taper off, since nothing can grow exponentially forever, However, the end is not yet in sight. There are probably many existing computer capabilities that we haven't yet exploited, since progress is occurring so rapidly. This will be a time to remember.
Next year, Fujifilm and Olympus will market an 8 GB xD-Picture Card developed by Toshiba. 8 GB is enough to handily replace laptop hard drives, and to beef up PDA's and tablet PC's. Price is not listed in this Popular Mechanics article (Technology: "Small Size, Big Brains", Tobey Grumet, Popular Mechanics, December, 2002, pg. 40).
This is an enormous amount of semi-random storage for next year.
The Robots Are Coming
Technology Review has an article praising iRobot's $199 "Roomba"
(Gadget Master- iRobot Roomba).
iRobot's CEO, Colin Angle says that its $199 price tag is what makes this
"an incredibly important robot.Ē
The author of the article, Simson Garfunkel, says, "If anything, Angle is being modest. Although Detroit has been using robots to build cars for four decades, Roomba is the first device to bring the labor-saving promise of robotics into the home. While Sonyís Aibo and similar toys proved that consumers want robots, Roomba may be the first that they actually need."
I think he's right. I think this robot is going to be to the field of robotics what the Tandy TRS-80 and the Commodore PET-2002 was to the personal computer industry. I think it will jump-start personal robotics.
"For all of my excitement about the Roomba, I did encounter some minor problems. Roomba shuts itself down when an object gets wrapped around its main roller, but it leaves it to you to guess what happened. I would like a better battery indicator. And in a few years time, I would like a machine that can automatically wake up when I am out of the house, clean the floors, and then plug itself in for a recharge.
"Despite these minor failings, Roomba is truly impressive. It really does clean your floors! But even more impressive is the robotís price. I showed my loaner Roomba to several friends and asked them how much they thought it should cost. One person said $800. Another said $600. If Roombas are in tight supply this fall, itís a sure bet that they will be showing up on eBay at those prices. But at its breakthrough price of $200, the Roomba really will take service robots out of the realm of science fiction and bring them into peopleís homes and offices. This little robot is going to have a huge impact."
Ten years from now, we'll look back on "Roomba" and say, "I've got one of those old Roomba's in the attic. I'm saving it until it becomes a collector's item."
A better time to review this might be in 2027. That will be as far in the future as we are now from the TRS-80 and the PET-2002. It was four years after the introduction of the TRS-80 and the PET-2002 that IBM brought out the first PC. If past is prologue...
Turtles Live Forever?
A quiet backwoods study opens a huge window on aging
Researchers have observed turtles at the E. S. George Reserve near Ann Arbor, Michigan, for five decades in order to build up a life-history database. Number 29 is the daughter of a Midland Painted Turtle that was first examined at age 8 in 1952.
Turtles are killed by infections, by raccoons, and these days, by people. But they don't senesce. They have been little noted because it takes so long to study them that researchers can't get publishable results quickly enough.
for article. Recent analysis of
commercial catches of rockfish showed that 16% were more than 50 years
old, and several had reached, if not surpassed, the 100-year-old mark.
But these fish are not only long-lived, they don't grow old! Instead, they manifest what is known as negligible senescence-chronological aging without increased mortality. In other words, they continue to grow and reproduce after reaching maturity but show no evidence of senescence. In fact, fecundity increases with increasing age.
The more Guerin studied the phenomenon of negligible senescence, the more passionate he became because he discovered the paucity of research that had been done in this area. Determined to fill this void, he began to assemble a network of scientists who could apply the principles of gerontological research to these issues.
These reports may put a new
complexion upon last night's article, "Reserve-capacity hypothesis (Weinstein & Ciszek 2002)".
One of the gerontological researchers whom Weinstein and Ciszek have taken to
task has written. Telomere
Technology: The End of Mortality" (by Bill Walker). The
author mentions that bowhead whales have a lifespan of 200 years, and that,
because of their size, they must have extraordinary defenses against cancer.
One idea behind a telomerase pill that one might take every two years to maintain, but not shorten the length of one's telomeres is that by merely retaining the length of existing telomeres, benign tumors, although they would be granted another doubling, wouldn't be given a license for unrestrained growth... or at least, that's my off-the-wall thought about it. But this will have to be tested carefully, possibly using animals that are so old that they're all but dead... animals that have nowhere to go but up.