Estimations of Solutions of Extended Similarity Search Problem of DNA Base Sequences on Computers of Different Architecture

Results of investigations on estimations of different computer architectures' efficiency for solving of problem of similarity search in genetic texts sequences are adduced in the paper. The analysis of DNA texts is conducted with the aim to study the functional purpose and evolutional origin of different genom parts. An estimation of efficiency is conducted on typical fragment of job for mutual information definition in sequences DNA under comparison. The adaptation of industrial system for genetic text analysis for powerful computer complex of matrix processors IZOT is described. Algorithms and programs are designed for Universal and Specialized computer architectures. The former are presented by computer families ES (models 22, 33, 35, 36, 45, 46, 60, 66) and IBM PC (models 8088, 286, 386). The latter are presented by following variants: 1) Microprogram 'accelerators' the - implementation of special microprograms for ES1035 (ES1036) and IZOT (ES1037+ES2706) complex; 2) Hardware 'accelerators' - the implementation of specialized chip for IBM PC; 3) Vector computer system PS-3000; 4) Super computer SMS-14 with transputer intersections organization. For the record of problem following languages were used: Fortran; C; Micro-Assembler of CPU ES-1035/36; Microprogram language APAL; Assemblers of ES, IBM PC and PC3000; OKKAM; Hardvare of different series. The optimization is proposed for array elements access on displaced indexes.