37 Years of Moore's Law
By Sally Adee
First Published May 2008
The data
From 2300 transistors on an Intel 4004 chip to the
forest of 2 billion transistors residing on the latest
generation of Intel microprocessors, Gordon E. Moore’s
famous law has guided the steady shrinking of
transistors and their consequent density on
microchips. That doubling about every two years is how
we went from Pong in 1972 to the astonishing real-time
rendering of hair that moves according to the laws of
real-world physics in the video game Heavenly Sword in 2007.
Sources: Intel Corp., VLSI Research
|
Enabling such technological marvels has been the
sharp growth of available processing power and the
commensurate decline of the cost of DRAM over the past
37 years [see timeline]. Though Moore made
his prediction in 1965—which was 45 years ago—it was
the concurrent invention of the microprocessor and
DRAM in 1971 that sparked the complementary arcs shown
below. Because the semiconductor industry advances in
lockstep, this curve, though based on Intel chips, is
likely representative of the general trend.
Technological advances like personal computers, the
Internet, and video games drove these complementary
slopes of cost decline and processing-power growth. But
just as in the chicken‑and-the-egg scenario, those
slopes were also the driving force behind the consequent
technological advances. Some might balk at the US $399
price tag of an iPhone, but in 1971, its then strictly
hypothetical 128 MB of DRAM would have set the
consumer back about $50 688 in 2008 dollars. In recent
years, processing power has hit a plateau: to continue
ramping up performance according to the expectations set
by Moore’s Law, companies like Intel and AMD have turned
to multicore processing. The future path of this graph,
therefore, is not necessarily predictable.