Talk:ICL Distributed Array Processor

Mil-DAP

I wasn't connected to this side of DAP development but I have a hazy recollection the Mil-DAP was used in aircraft for side-looking radar - this can be used to spot submarines by analysing the waves they cause on the surface. Dmcq 18:23, 7 May 2007 (UTC)

Just noticed this. I believe Mil-DAP was part of a pitch for the UKADGE network updates, which were won by IBM. It seems a little large for aircraft use. Maury Markowitz (talk) 18:56, 27 March 2020 (UTC)

Sites

Listing the numbers of systems made and possible site locations could be historically useful. Terming a site as "classified" can be mentioned for site of sensitivity. Hardware is preserved at the Computer History Museum. 143.232.210.38 (talk) 18:32, 24 August 2010 (UTC)

Production figures

The number of systems made in comparison to competitors would be interesting. 198.123.57.167 (talk) 19:57, 29 September 2011 (UTC)

How is this the first MPP?

According to the dates in this article, it appears design and implementation of the DAP followed the ILLIAC IV by two or more years. On what basis does this article claim precedence? Maury Markowitz (talk) 19:56, 22 December 2011 (UTC)

It says commercial as in salesmen selling it from their portfolio rather than built by and for a university. Dmcq (talk) 20:36, 22 December 2011 (UTC)

That is not clearly true in the article as it stands.

The LEAD claims that the first machine delivered to a university, and it appears from the context that it is referring to the prototype.

The delivery date is referenced to a cite that doesn't mention anything about this. I did find this date in a FAQ document, it does not clarify.

The original cite does imply, however, that ICL did not commercialize the DAP, and that this did not occur until it was with CPP.

A more likely document found here states the date was in 1980.

If either of these are correct, it still post-dates delivery of the Goodyear MPP, which was moved to Goddard in 1977 according to the same FAQ, and is mentioned as "late 1970s" in this book. However, other references state it is some time in 1983, 1984 and 1982.

The confusion may lie in the fact that the STARAN was being sold throughout this period. It used a 4 x 256 1-bit PE, which makes it "massive" if the same can be applied to the DAP's 64x64 1-bit array. By this measure I believe it is safe to say then that the STARAN predates the DAP.

Maury Markowitz (talk) 16:42, 23 December 2011 (UTC)

The pilot DAP was built sometime between 1974 and 1976 I'm not certain of the date. There was at least one full 64x64 DAP built by ICL for internal programming and testing work before the initial commercial sale of a DAP to Queen Mary. I can't see where the date of delivery of the Queen Mary University one in the article came from so if you want to stick in that date fine by me.

The pilot DAP was a 32x32 model. The version delivered to Queen Mary was a commercial 64x64 model. I don't get where you get that it wasn't commercialized. It was sold to a number of places but wasn't a great commercial success if that's what you mean.

The Goodyear MPP article says 'Development of the MPP began circa 1979; it was delivered in May 1983, and was in general use from 1985 until 1991.' so that is definitely after the DAP. If you want to show otherwise I suggest you talk at that article rather than this one. Dmcq (talk) 17:35, 23 December 2011 (UTC)

The first non-university DAP sale was, as I recall, to the British National Oil Corporation. This paper about the software development suggests that DAP was operational some time between May 1980 and October 1982 - preceding Goodyear MPP it seems. — Preceding unsigned comment added by Rwendland (talk • contribs) 18:15, 23 December 2011‎ (UTC)

I think the ICL DAP may have been delivered to Queen Mary College in December 1979 so people wouldn't have had use till 1980 so probably 1980 is a better figure for 'installed' even if 1979 is delivered. Dmcq (talk) 18:25, 23 December 2011 (UTC)

Sorry, just returning after a bout of "real life". So can someone present a list of the best available dates for:

First prototype in operation
First full version in prototype operation
First operation in a production setting

STARAN was build in some form in 1971, started full prototype operation in 1972, and was delivered operationally some time in 1974, and fully operational and used in production by 1975. It was also demonstrated publicly that year.

http://books.google.ca/books?id=lNcDAAAAMBAJ&pg=PA46-IA6
http://groups.csail.mit.edu/cag/6.893-f2000/lectures/l05-parallel.pdf
http://dl.acm.org/citation.cfm?id=724259
http://books.google.ca/books?id=uDDvVoE7sfMC&pg=PA617

MPP was a larger and faster version of STARAN. According to what I can find, Pilot DAP did not start construction until 1975, and was not operational in a production setting until the 1980s:

http://books.google.ca/books?id=m8S4bXj3dcMC&pg=PA627

If anyone has any other references, please post them! Maury Markowitz (talk) 20:20, 17 December 2012 (UTC)

Your figures are probably as good as any. I think calling STARAN an MPP computer is rather stretching the point. The list at the end of the SIMD article is what people normally mean by MPP for that era. Staran was a specialized adjunct to a computer helping with search and capable of a bit more programming, it would be like calling the cpu of an ICL CAFS unit an MPP. Dmcq (talk) 21:20, 17 December 2012 (UTC)

I just found a good cite about the STARAN hardware and have added it to that article. It seems more capable than I thought, it had quite an advanced and fast interconnect system for moving data around which seems to have compensated for the very limited memory per element. Dmcq (talk) 22:09, 17 December 2012 (UTC)

Ok, so does the STARAN meet the requirements for "earlier" in this case or not? If not, why not? Maury Markowitz (talk) 17:01, 10 June 2013 (UTC)

Sorry I didn't check what it said in this article properly. It said massively parallel and I don't believe anyone claims that for the STARAN. It is the Goodyear MPP that was massively parallel. The STARAN is claimed to be the first commercial computer designed around associative memory. I think some of those links should be added to the STARAN article though. Dmcq (talk) 17:57, 10 June 2013 (UTC)