Hardware Progress: $272/GFlop/s

From: Dani Eder (danielravennest@yahoo.com)
Date: Fri Aug 23 2002 - 16:37:13 MDT


The cost of processing power continues to drop at
a Moore's Law pace. Cost has dropped 15%
in the past the months and by over half
in the past 9 months.
 
I've been tracking the cost of a node in
a Beowulf-type cluster of commodity computers, where
the memory is kept at a ratio of 1 byte/flop/sec,
and storage is 100 bytes/flop/sec. The best
cost/performance nodes I have found are as follows:
 
Date $/Gflop/s
Nov 01 600
Jan 02 437
Mar 02 397
May 02 319
Aug 02 272

 
Component prices are obtained from pricewatch.com,
including shipping but not sales tax. Current
configuration is as follows:
 
Case: $23 Generic Mid-ATX w/300W power supply
Motherboard: 36 Compaq reseller
CPU: 48 AMD Duron 1.3GHz
Memory: 114 3x512MB PC133 SDRAM
Storage: 166 2x80GB EIDE internal Hard Disk
Networking: 100 4 NICs/node, wiring, hubs

Total $487/node

AMD processors when used efficiently perform 1.375
Flop/Hz (3 FPUs @ 2 cycles per calculation -
overhead),
which makes the calculation:
 
$487 / (1.3 GHz * 1.375 Flop/Hz) = $272/Gflop/s

If optimistic estimates of the required computer
power for human-level AI are correct at 100 TFlop/s,
it presently costs $27.2M to buy a human's worth
of computers.

I have estimated an 'economic
crossover' of $3M for 100 TFlop/s when computers
become generally cheaper than humans. This is
based on a computer being able to put in 5x as
many productive hours as an average human, a 5 year
payback time on the hardware, and $120K as the total
cost per year of a technical professional. We are
therefore about 3.2 doublings in performance/$
away from economic crossover. At current rates of
cost reduction, that would be reached in 44 months,
or early 2006.

Economic crossover is not implying we would have
human-level AI at that point, but rather that
skilled human tasks which require lots of processing
power would be cost-feasible. In particular, high
level factory automation should be possible. This has
implications in the social and economic realms.

The processing power to run a human-level AI is
unknown, but is estimated to be in the range of
100 - 100,000 TFlop/s. Assume the most powerful
machine available for AI work is in the $3 M price
range. Then the expected date of a human-level
machine is 2006-2019. The expected date for
superintelligent machines is 6.5 years later. This
is calculated from a human taking about 20 years
to train. A machine 8 times faster than a human
is expected 4 years after a human-equivalent, and
it would be expected to train in 20/8 = 2.5 years.

Once superintelligent machines are available, the
time constant for doubling performance may drop to
much shorter periods as superintelligent machines
design their successors ever more rapidly. This
singularity may occur in the 2012-2025 time period.

Trends in labor productivity also point to a
singularity
in the ~2020 time period. The trend indicates labor
input will fall to zero about that time, i.e. fully
automated production. The cost of any given product
should then be much lower than today.

 
Daniel

__________________________________________________
Do You Yahoo!?
Yahoo! Finance - Get real-time stock quotes
http://finance.yahoo.com



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT