Talk:English Electric DEUCE

Drum heads description
I was a user of DEUCE back in the 1960s. The description of the read and write head assemblies is a little misleading. The 16 read heads were on one movable stick and the 16 write heads were on another diametrically opposite. Each had its own servo system to move them over the requested track. Perhaps someone would care to edit the main page so as to word this description better than I have! Wojja62 (talk) 11:00, 29 March 2012 (UTC)


 * So each of the 32 heads weren't each on a separate servo, instead it was only two servos, one for each of the set of read heads and the set of write heads? - Denimadept (talk) 22:39, 29 March 2012 (UTC)


 * That's correct. Only two servos - each able to bring its own stick of 16 heads to 16 different positions, thus giving 256 tracks, each of 32 words - 8192 words in total. The read and write assemblies were completely independant. Wojja62 (talk) 00:24, 3 April 2012 (UTC)


 * Thanks. Updated.  Note, I don't know this machine at all.  I encountered the article last week via the agency of "random article".  I know computers, though. - Denimadept (talk) 16:38, 3 April 2012 (UTC)

First commercially produced computer in GB
"Davies discusses a much larger, second ACE, and the decision to contract with English Electric Company to build the DEUCE -- which he calls the first commercially produced computer in Great Britain".

Although this is quoted as a personal opinion, surely some qualification would not go amiss. I would think the Ferranti Mark 1* would be a challenger, or would that be ruled out because only 7 were manufactured? Flying Stag (talk) 12:17, 14 April 2012 (UTC)

Hardware and Instructions
I programmed a DEUCE computer in 1962 for 8 months before going to college and I remember it reasonably well. The third paragraph under "Programming" says that there were twelve delay lines for instructions. In fact there were eight and the other four were used for data and other functions. It was a 3-address instruction computer and the field for the next instruction source was 3-bits long, hence delay lines 1-8 were instruction delay lines. Source and destination fields were 5 bits long and there were 5-bit wait and timing field5. Wait defined how many minor cycles to wait before connecting the source and destinations after 2 minor cycles (a minor cycle was 32 microseconds) and the timing was the similar delay but this time to the next instruction. In this way instructions were best to be sequential in time but could be scattered between the 8 instruction delay lines. It was possible to transfer a whole 32-word delay line in one instruction with suitable choice of next instruction timing and an additional bit set. Equally, one could have self-counting loop instructions, a one-instruction loop, by setting another bit and, if necessary, filling in the 4 bits between the wait and timing fields. in this way the wait counting would eventually spill into the timing field and by then incrementing it, choose the following instruction in the delay line to leave the loop. Another interesting feature was that source zero was the card reader brushes and destination zero was control. The 31st bit was a go bit. If this was not set, then a one-shot was needed to cause the instruction to be executed. This could come from the card reader and thus the first row on a card would be executed. In this way, the first card would include the bootstrap to load a delay line by having the bootstrap instructions punched first. It is amazing what one remembers after all these years. The computer had a dedicated engineer to maintain it. I wrote card reader and card punch routines in support of the senior programmers. We developed the programs on paper and then punched the instructions as 32-bit words. Peterclout (talk) 02:24, 4 October 2013 (UTC)

SODA (Symbolic Optimum Deuce Assembly Program)
A Translation Routine for the DEUCE Computer

Abstract

Most computer in operation today have supplementary programs which do automatic coding or program assembling. These programs either translate, automatically code, or interpret pseudo instructions which in themselves may cause the enaction of hundreds of actual machine instructions. The outstanding feature of such routines is that programming time and effort is cut to a minimum.

This paper deals generally with translation and interpretive schemes, and specifically with a suitable translation routing for use with the DEUCE computer. The translation program is called SODA, or Symbolic Optimum Deuce Assembly Program. Two examples of SODA use are included in the appendix. --89.25.210.104 (talk) 15:13, 14 June 2018 (UTC)

Schematic
Does anyone know if there are any schematics available an excerpt from them would be a good addition?--Kitchen Knife (talk) 20:35, 22 July 2020 (UTC)