Erlangen Classification System
The Erlangen Classification System (ECS) was developed by Wolfgang Händler in 1975 and is used to describe computer architectures with regard to their parallelism . For this purpose, triples used (in square brackets are not mandatory):
tRechnertyp = (k [* k'], d [* d'], w [* w'])
Here means
-
k is the number of concurrent control units (tail units)
- k 'is the number of specialized control units for program or processor pipelining
-
d the number of concurrent arithmetic units (per control unit)
- d ' the number of pipelining arithmetic units (per control unit)
-
w the number of parallel bit positions in the arithmetic unit (ALU)
- w 'is the number of elementary sub-works
There are also the following connection operators:
- + : inhomogeneous additional units
- v : modes of operation
- x : macro pipelining.
Examples
- A i860 32-bit computer with a floating point - coprocessor and three-stage pipeline:
- t i860 = (1 * 1, 1 * 1, 32 * 3) + (0 * 0, 1 * 1, 64 * 3)
- A distributed array processor (distributed field computer) with three operating modes:
- t DAP = (1, 64, 64) v (1, 128, 32) v (1, 4096, 1)
- A 60-bit central processor with nine- fold super scalarity and 15 upstream 12-bit preprocessing units:
- t CDC7600 = (15, 1, 12) x (1, 1 * 9, 60)