© 1978 by British Computer Society
| ||||||||||||||||||||||||||||||||||||||||||||||||||
Text compression with an Associative Parallel Processor
Department of Electrical Engineering and Electronics, Brunel University, Kingston Lane, Uxbridge, UK
Text compression, using a coding dictionary of 200+ n-grams, can halve file storage costs and double data transmission rates. However, software based text compression systems are slow and expensive in storage.
Two hardware systems (based on a fixed record length and a byte-organised variable record length Associative Parallel Processor), for the compression and decompression of textual information, are described. Algorithms are given and their execution illustrated with practical examples.
A feasibility study, comparing the performance and costs of the two systems with a conventional microprocessor (Digital LSI-11) implementation is also reported. The fixed record length system is 3,600 times faster for compression, 1,600 times faster for decompression and its production costs is nearly 6% cheaper. The variable record length system is 560 times faster for compression, 240 times faster for decompression and its production cost is 58% cheaper. Whereas the conventional micro-processor system cannot perform at typical disk transmission rates, the compression and decompression rates for the fixed record length hardware are 4.1 and 6.3M bytes-per sec. and for the variable record length hardware are 0.64 and 0.91M bytes-per-sec. respectively.
Received May 1976.
* Department of Electrical Engineering and Electronics, Brunel University, Kingston Lane, Uxbridge, Middlesex UB8 3PH