Browse Prior Art Database

Parallel Computing Systems

IP.com Disclosure Number: IPCOM000094924D
Original Publication Date: 1965-Jul-01
Included in the Prior Art Database: 2005-Mar-06
Document File: 2 page(s) / 51K

Publishing Venue

IBM

Related People

Goldstine, HH: AUTHOR [+5]

Abstract

Instructions are executed concurrently in indexing unit 5, data storage unit 6, and a pair of arithmetic units 7 and 8. Under the control of indexing instruction ring 10, unit 5 generates addresses for fetch and store operations. Memory reference instruction ring 11 controls the execution of fetch and store operations in unit 6. A pair of arithmetic instruction rings 12 and 13 cause associated units 7 and 8, respectively. to perform add, subtract, multiply and divide operations.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 83% of the total text.

Page 1 of 2

Parallel Computing Systems

Instructions are executed concurrently in indexing unit 5, data storage unit 6, and a pair of arithmetic units 7 and 8. Under the control of indexing instruction ring 10, unit 5 generates addresses for fetch and store operations. Memory reference instruction ring 11 controls the execution of fetch and store operations in unit 6. A pair of arithmetic instruction rings 12 and 13 cause associated units 7 and 8, respectively. to perform add, subtract, multiply and divide operations.

Rings 10...13 are cyclic in operation causing their associated units 5...8 to repeat the same sequence of operations. Rings 10...13 can be simultaneously, synchronously advanced in response to pulses from clock 15. As an alternative mode of operation, rings 10...13 can be advanced asynchronously under control of interlock 16. This allows each ring to advance when the previous operation is completed by the associated unit. Interlock 16 prevents the premature execution of an instruction in one ring before data has been made available by the execution of instructions in the other rings.

Data is transferred along the paths between unit 6 and a pair of buffers 18 and 19. The latter temporarily store the data for use in units 7 and 8. The arithmetic operations are carried on in units 7 and 8. The data can be returned to unit 6 or transferred to another one of buffers 18 or 19.

Rings 10...13 store instructions for solving any repetitive problems, such as Gaussian elimi...