MooresLaw

MeatballWiki | RecentChanges | Random Page | Indices | Categories

Originally a 1965 observation by Gordon Moore that memory storage tends to double every 12 months. The doubling time was adjusted in the late 1970s to 18 months, which has been very accurate through at least 1999.

Since its introduction, Moore's Law has been applied to many other areas of computing capacity, like the number of transistors and the speed of processors. It has also been used (abused) in a more general sense suggesting that "technology will double in capability" every 12 months. In a few cases, such as hard disk storage and some measures of network bandwidth, growth rates have recently greatly exceeded Moore's Law. (Does anyone remember when a gigabyte was a huge amount of disk storage, and a T1 line was only for large companies?)

The main problem with Moore's Law is that it assumes unbounded growth, much like capitalistic economy. However, at a certain point there is no further to go. At least that's the theory. So-called pundits have been claiming the end to Moore's Law for ages, mostly as a scare tactic to raise their own profile. Always they have been wrong. Another late-1990s trend was to predict faster growth than Moores Law would give. For each pundit pointing about an insurmountable problem, there seemed to be a new technology to demolish the limits (such as new optical technologies or material improvements).

One of the limiting factors of Moore's Law is the enormous investments that have been made to make the new products quickly, like overlapping research/development generations and huge factory investments. In some cases, products may become largely obsolete before they even reach the market (Arguably Intel's Pentium Pro, Xenon, and Merced products fit this category). Rock's Law (a small addendum at [1]) says that "the cost of capital equipment to build semiconductors will double every four years".

MooresLaw

Discussion