November 23, 2002
What Does It Feel Like to Be 73?
Shockingly, it feels the same as it did to be 23! I certainly don't look 23, and I have to wear reading glasses, but internally, the changes are minor. I don't have any more aches or pains that plague me than I did when I was 23. I have a fallen left arch that hurt for years, but somehow, it healed itself. For months, my right shoulder hurt when I tried to bend my arm overhead and backward to turn on the overhead light in the car, but the problem went away. Somehow, I sustained quite a bit of damage in my neck, but I'm no longer aware of it. I don't suppose that I have the reserves of strength and energy that I had when I was 23, but it's hard to be sure. My ability to learn doesn't seem to have been compromised in a way that I can detect.
The flip side of this is that I'm no more eager for death than I was at 23. In thinking about old age, I had hoped that we would become desensitized to it, but unfortunately, that hasn't happened. Nor does 1990 or 1995 look very far back.
More on Blue Gene and ASCI Purple
A new article today sheds light upon IBM's plans for supercomputers. IBM will deliver the 100 teraflops ASCI Purple supercomputer and the 367 teraflops Blue Gene supercomputer in 2005. ASCI Purple will be a next-step in conventional supercomputer development, while Blue Gene will be a major departure from conventional approaches. If Blue Gene works, then a petaflops version of Blue Gene would follow. However, Cray, Inc., plans to deliver isX1 petaflops computer in 2010, and observes that a one petaflops supercomputer won't appear before 2010 because it requires a 20-terabyte/second memory bandwidth, and that won't be available before about 2010.
Intel's one-terops desktop chip
In the meantime, Intel has announced plans for a one-terops (1,000,000 MHz) desktop chip by 2010. This would require something pretty extraordinary in terms of memory arrangements. Memory speed currently maxes out at 433 MHz, interfacing with processors running at speeds above to 2,000 MHz, or at 1,066 MHz (Rambus) for Intel processors with speeds up to 3,000 MHz. It will be a while before we have memories that can operate at 333 times this speed to support a one-teraflops processor. One solution might be to mount additional memory upon the microprocessor chip. Intel is mentioning transistor counts of ten billion transistors by 2010. More than 128 MB of memory could reside on the chip. It might be possible to pump data in and out through very wide buses.
For 2005's 10-GHz microprocessors, it will be desirable to bump up memory speeds by a factor of 3 over what they are today. (That may be feasible: memory speeds have been doubling every two years.)
One-Hundred Gigaflops Desktops
Lacking this kind of quantum leap, I would anticipate clock speeds of about 50 GHz, and processing speeds of about 100 gigaflops by 2010.
Intel's fastest P4 will hit 3 GHz this year. If Intel keeps its promise of a 10 GHz microprocessor by 2005, I would expect to see
I personally expect to own
DDR memory speeds, which max out at
For 2003, desktop disk drives are projected to reach 400 gigabytes ( IBM Unveils "Pixie Dust" Hard Drives). One terabyte disk drives may be in our computers by 2005. One terabyte could store about 1,333 full-length movies, or about 1,000-feet-by-1,000-feet of 72-dpi scenery.
I expect to own a
The National Storage Industry Consortium has set a goal
of disk drives that can store one terabit per square inch by 2006. That would
enable 3-to-4 terabyte, 3.5" disk drives by 2006, in keeping with the
doubling-every-year capacity increases that have characterized disk drives for
the past decade. NIST has awarded Seagate a $21,000,000, 5-year research
contract to develop disk storage techniques that will increase disk storage
capacities by a factor of 100 or more, eventually allowing, perhaps, 20-or-more
terabytes to be stored on a 3.5" disk drive. Impressive as are these
numbers, it is perhaps worth mentioning that the human brain, with 1015
synapses, might store something like 1,000 terabytes of information. Of course,
the brain is constructed of unreliable components, and may require redundancy to
a degree that isn't necessary in computers. In that case, 200 terabytes might be
a more realistic estimate of the brain's storage capacity. (My estimates are
By 2010, 200 terabytes might be achievable with 10 20-terabyte disk drives. And by 2014, that number might shrink to one or two disk drives.