Contact us Heritage collections Image license terms
HOME ACL Associates Technology Literature Applications Society Software revisited
Further reading □ Preface □ Contents □ Members □ 1 Welcome □ 2 Introduction □ 3 EDSAC □ 4 EDSAC Demo □ 5 Relay Computers □ 6 Discussion □ 7 CRT Storage □ 8 Coding □ 9 Library □ 10 Sign Correction □ 11 Nozzle Flow □ 12 Magnitude □ 13 France □ 14 Checking □ 15 Large Integers □ 16 Discussion Storage □ 17 Magnetic Storage □ 18 Magnetic Recording □ 19 Photographic Store □ 20 EDSAC Auxillary Store □ 21 Circuit Checking □ 22 Circuit Checking □ 23 Addition Circuit □ 24 Trigger Circuits □ 25 Checking □ 26 Discussion □ 27 USA □ 28 Comment □ 29 Holland □ 30 Ficticious Traffic □ 31 Sweden □ 32 Manchester □ 33 Discussion □ 34 Bibliography
ACD C&A INF CCD CISD Archives Contact us Heritage archives Image license terms

Search

   
⟶ ACL ⟶ Literature ⟶ Other manuals ⟶ Cambridge Conference 1949 :: High Speed Automatic Calculating-Machines 22-25 June 1949
⟶ ACL ⟶ Literature ⟶ Other manuals ⟶ Cambridge Conference 1949 :: High Speed Automatic Calculating-Machines 22-25 June 1949
ACL ACD C&A INF CCD CISD Archives
Further reading ⇓

□ Preface
□ Contents
□ Members
□ 1 Welcome
□ 2 Introduction
□ 3 EDSAC
□ 4 EDSAC Demo
□ 5 Relay Computers
□ 6 Discussion
□ 7 CRT Storage
□ 8 Coding
□ 9 Library
□ 10 Sign Correction
□ 11 Nozzle Flow
□ 12 Magnitude
□ 13 France
□ 14 Checking
□ 15 Large Integers
□ 16 Discussion Storage
□ 17 Magnetic Storage
□ 18 Magnetic Recording
□ 19 Photographic Store
□ 20 EDSAC Auxillary Store
□ 21 Circuit Checking
□ 22 Circuit Checking
□ 23 Addition Circuit
□ 24 Trigger Circuits
□ 25 Checking
□ 26 Discussion
□ 27 USA
□ 28 Comment
□ 29 Holland
□ 30 Ficticious Traffic
□ 31 Sweden
□ 32 Manchester
□ 33 Discussion
□ 34 Bibliography

26 General Discussion: Prof D R Hartree and others

Prof. HARTREE gave a summary of checking facilities provided by American designers, saying that:

  1. The Bell Telephone Laboratories relay machines were originally designed to be used in duplicate with cross checking at each step. They were never, in fact, used in that way, mathematical checking being used instead.
  2. The arithmetic unit of the EDVAC was built in duplicate and each step cross checked.
  3. The UNIVAC is to be a decimal machine using a binary coded form for decimal digits with an additional digit having the value 0 or 1 in such a way as to give an odd number of 1's in the representation of any decimal digit. A check is to be based on the parity of the number of 1's in the group of five binary digits representing each decimal digit
  4. The RAYTHEON machine will use weighted counts to check transfer and arithmetical processes. This will amount to doing the calculations once in full and once modulo 15 or 31.
  5. A feature of the design of the EDVAC and UNIVAC machines is the provision of break points. The machine runs automatically up to a break point at which the results can be examined before running to the next break point. When the operator is satisfied with the functioning of the groups of operations, the break points can be cut out and the calculation run straight through.

Prof. HARTREE also said that Prof. AIKEN was fond of pointing out that unless the reliability of a machine was very high its overall speed depended primarily on the frequency of faults and the ease and speed with which they could be diagnosed and detected, rather than on the speed of arithmetical operations. Reliability, therefore, was the most important thing to aim at.

Dr. UTTLEY mentioned that the Bell Telephone Laboratories machine had a lot of built in checking equipment.

Prof. WILLIAMS said that at Manchester they had tried to avoid the incorporation of any non-essential components of their machine. They felt that the less apparatus there was, the less there was to go wrong and the sooner they would have a machine. He thought faults were usually reasonably obvious without red lamps and he would always back the man who had lived with the machine against other systems of fault finding. Dr. KILBURN referred to the use of test programmes for localizing faults. Experience was that once a fault had been traced to an 8-valve chassis it could be located in a few minutes.

Prof. HARTREE suggested that if Mr. NOBLE could make a circuit with 10 stable states it might sound the death knell of the binary system. Dr. UTTLEY said that so far they had only obtained 5 stable states. In reply to Dr. KOSTEN he said that at present their storage system had only 2 stable states. In his view a 50% increase in complication would be well worth while if an indication of where things were going wrong could be obtained. At T.R.E. they were fortunate in having time available for investigation before they started to build their machine.

Dr. GOODWIN said that although compensating errors did not occur very often in computation they occurred occasionally. With a high-speed machine it might possibly be necessary to guard against their occurrence.

⇑ Top of page
© Chilton Computing and UKRI Science and Technology Facilities Council webmaster@chilton-computing.org.uk
Our thanks to UKRI Science and Technology Facilities Council for hosting this site