ACD Atlas Computing Division PERQ History: Part V

Jump To Main Content

Jump Over Banner

Home

Jump Over Left Menu

1982

25. PERQ-DAP

25.1 Introduction

Near the end of 1981, ICL decided to build a 32 × 32 DAP to go on the PERQ. ICL discussed this with Brian Oakley and RAL was asked to represent SERC on the project.

The initial ICL proposal was a SERC collaborative grant in conjunction with two UK universities. ICL believed the development costs would be £1.5M and hoped to get £750K from SERC principally in manpower at RAL and the universities.

The project was to be in four phases:

  1. Market Specification: identify a number of application areas and use them to specify the architecture of the DAP and its attachment to the PERQ.
  2. Hardware: produce a number of prototypes.
  3. Systems Software: develop the necessary systems software.
  4. Application Trials: establish a DAP bureau based on a 64 × 64 DAP and convert packages to run on it in preparation for the PERQ-DAP hardware.

The elapsed time for the product was at least 20 months so that the prototypes would be available in the autumn of 1983 if the project started immediately.

The response from SERC was that the project was interesting but it was unlikely that more than 25% of the funds could be found as money was tight.

25.2 Initial Discussions

An initial discussion took place between ICL and SERC in January 1982. Eric Thomas of RAL represented SERC. ICL had a vague idea of what they were proposing. It was clear that they had not decided whether the product was a parallel engine on a local area network (Cambridge Ring) or an integral part of the PERQ.

RAL were convinced that a DAP on the PERQ where DAP memory was also the bit map for a colour display was an interesting and unique product. A parallel processor on a ring was competing directly against many other floating point processors and it was unclear whether it had any market edge.

This was relayed back to ICL in a letter from Brian Oakley on 2 February 1982.

In March, ICL produced the initial cooperative grant proposal. A market survey had indicated a need for PERQ-DAP in the CAD area. Also, image processing with a colour display attached to the DAP memory had exciting possibilities.

ICL were also discussing with MOD the possibility of the PERQ-DAP being used in the signal processing area.

ICL anticipated total sales of 1,000 systems by the end of 1986. At a price of £25K per system, breakeven came at a sale of 150 units which could be reached by the end of 1984.

ICL could already produce chips with 4 processing elements per chip. There was also work going on to see if the FORTRAN development system on the 2900 DAP could be moved to the PERQ.

The proposal requested SERC to purchase a pool of PERQ-DAPs in a similar way to the purchasing of PERQs. The aim was to produce 8 prototype systems of which 4 would be paid for by SERC for university research while 4 would be used by ICL for industrially directed research.

25.3 Grant Proposals

ICL started talking directly to the university community and encouraged Prof Burge at Queen Elizabeth College and Dr Rhind at Birkbeck College to submit grant applications.

In April 1982, Dr Manning told Dr Hobbis of the Science Board that ICL were considering the production of a modern DAP on the PERQ. He indicated that the current ICL plans were in a rudimentary form and no decision had been made as to whether SERC would support it.

On 18 May 1982, Bob Hopgood wrote to the EB Secretariat telling them that the project was only vaguely specified and that efforts should be made to stop people putting in grant applications as it was far too early.

In May 1982, EB attempted to award, a rolling grant to QMC which included a PERQ-DAP. Heather Liddell wrote to Bob Hopgood at RAL requesting one of the PERQs that were available. He pointed out that no spare PERQs existed as they were needed by SERC and ICL for systems development. UNIX would not be available until later in the year so there was little point in having a PERQ yet even if the EB Secretariat agreed to one being allocated from the EB pool.

In the June grant round, two applications for PERQ-DAPs were submitted by universities for cooperative grants at the encouragement of ICL. One was referred and the other rejected. The PERQ-DAP had still not been specified yet ICL were encouraging universities to put in grant applications.

25.4 SERC/ICL Proposal

At the end of June 1982, David Worsnip wrote to Brian Oakley saying that he had recently had a meeting with Peter Lever and John Smallbone of ICL to discuss the rejected grant applications. As further applications were expected, he proposed a meeting between ICL and SERC where RAL, on behalf of SERC, would put forward a proposal on how the PERQ-DAP could be handled within SERC. This was agreed by Brian Oakley and the intention was to have John Taylor (Chairman of Computing and Communications Subcommittee), Laurence Clarke (Chairman of Information Engineering Committee) and representatives of other committees including the Mathematics Committee.

A meeting was held at Garrick House on 10 August 1982 where a paper produced by Eric Thomas of RAL was put forward for the first time giving a specification for the PERQ-DAP and proposing that:

  1. The high speed PERQ initiative of ICL, based on an attached DAP, should be supported;
  2. The project should be run by the Central Computing Committee's Single User System Steering Group (see Section 27.1);
  3. DoI should be approached for the funding of ten high powered PERQ/DAP systems under a pre-production order scheme;
  4. Funds should be provided to organise collaborative research grants with universities to the level of £200K per annum over the next 3 years to exploit the initiative.

The meeting was a disaster. John Taylor never turned up and Laurence Clarke pressed that the PERQ-DAP should not be organised as part of the Central Computing Committee's Single User System programme. Instead, he pressed that the mini-DAP be decoup1ed from the PERQ and should be handled by a new Specially Promoted Programme in Novel Architectures. The DAP could then be assessed in conjunction with other novel architectures.

As the major novelty of the PERQ-DAP was the coupling of the DAP memory to the display, this effectively removed one of the PERQ-DAP's main assets.

Laurence Clarke proposed that the PERQ-DAP be discussed at the IEC September meeting.

In general, the proposal received favourable comments but was delayed until at least September by no decision being made. It would not go to the Engineering Board until November. By not putting it to the Central Computing Committee, it effectively stopped support for the project coming from the Science Board or being discussed in a forum where the other Boards could give their comments.

The only Science Board representative at the meeting was Mr R F Hemmings, Secretary to the Mathematics Committee.

Nothing much happened in the Autumn of 1982 except that Brian Oakley in November 1982 recommended to David Leech that a note should be sent to all Committee Secretaries telling them what should be done when they received PERQ-DAP grant applications. It was unclear from his note what he suggested should be done!

25.5 Industrial Representation on SERC Committees

Nothing much happened concerning the PERQ-DAP at either the Autumn meeting of the Computing and Communications Subcommittee or Information Engineering Committee.

In ICL's view, SERC's support for the PERQ-DAP project was being deliberately blocked by GEC members of EB committees. In ICL's view, Laurence Clarke used his influence to ensure that responsibility for PERQ-DAP remained with IEC in order that GEC could filibuster on PERQ-DAP until their own product became available.

At CCSC, Roger Newey of GEC pressed very hard for the PERQ-DAP not to be supported (GEC had recently announced that an attached vector processor would be available on the GEC Series 63).

As a result of these activities, ICL refused to give a presentation in the autumn to the Single User System Steering Group on the basis that the information would most likely end up with their competitors.

25.6 Further Grant Applications

By December 1982, even though SERC had not agreed to support PERQ-DAP and had not agreed a collaborative project with ICL, nine applications for PERQ-DAP had been received by SERC. These all arose from direct marketing by ICL to the university community in an attempt to stimulate demand and indicate to SERC that there was an interest in the product. The applications were:

  1. Prof Burge (Queen Elizabeth College): Interactive Image Processing.
  2. Prof Rhind (Birkbeck College): Management and Processing of Spatial Data.
  3. Prof Evans (Loughborough): Computer Support for Parallel Algorithms Research.
  4. Dr Trowbridge (RAL/EB): Finite Element Solution of Large Problems.
  5. Dr Wallace (RAL/ASRB): Image Processing for Astronomy.
  6. Prof North (Leeds): Protein Crystallography.
  7. Prof Harrison (Sheffield): Protein,Crystallography.
  8. Dr Dodson (York): Protein Crystallography.
  9. Dr Sawyer (Napier): Protein Crystallography.

25.7 PERQ-DAP Progress

Due to no firm commitment from SERC despite a year of negotiation, delivery dates of the PERQ-DAP had slipped from mid-1983 to January-June 1984. This would be a special pre-production run with 9 systems available to SERC if a decision was made. The cost was estimated at £40K for the prototype systems with a price of £25K for the first production systems.

ICL also told SERC in December that they had agreed to develop a military version of the DAP in conjunction with RSRE.

ICL again pointed out to Brian Oakley that information concerning the PERQ-DAP was confidential and any leak to competitors would damage ICL's sales. ICL urged that all PERQ-DAP grants be considered by a single committee, preferably the Single User System Steering Group.

25.8 Science Board Grants

On 20 December 1982, Prof North resubmitted a grant application to the Biological Sciences Secretariat in conjunction with Prof Harrison and Dr Dodson.

Earlier in the year, a proposal had been submitted requesting support for three PERQ computers with a single Evans and Sutherland display to be shared between the laboratories. As a result, Evans and Sutherland had approached RAL to see if their colour display could be regarded as an add-on system to the PERQ. Dr Manning indicated to the company that it was unlikely to get SERC support due to ICL's own plans.

The new proposal from Prof North requested PERQ-DAP systems with colour displays. It saw the PERQ-DAP as a development of the greatest importance for the application of the molecular graphics approach to the study of biologically important structures and of intermolecular interactions.

Prof North asked the Science Board secretariat to bring his interest to the appropriate bodies in SERC. Dr Findlay of the Science Division wrote to me in January 1983 stating that the Biological Sciences Committee declined to fund the initial grant proposal as the equipment did not exist and hence could not be costed. He requested information on the current state of the PERQ-DAP and whether it was possible to acquire PERQ-DAPs with colour displays on favourable terms.

Bob Hopgood replied on 13 January 1983 stating that Dr Manning was producing a list of possible homes for the 9 pre-production systems being developed by ICL and which might be funded by Dol. Prof North and his associates were already on that list.

Bob Hopgood told Dr Findlay that no agreement had been reached with DoI but that the pre-production price was £40K. He said that deliveries were scheduled between January and June 1984 and the current plan was to have a colour video channel on it. He also told him the long term price of £25K and the need for commercial confidentiality.

25.9 Continuing Problems

By March 1983, ICL were still trying to get SERC to agree to take pre-production systems funded by DoI and to agree a collaborative research proposal with ICL.

Problems arose as, if DoI funded the PERQ-DAPs, ICL would need to put other contributions into the collaborative project, whereas if DoI did not fund, the cost of the PERQ-DAPs could be ICL's contribution.

Furthermore, if DoI did fund the PERQ-DAPs, they would be owned by DoI and would be sold to the customer (SERC) at the end of the period. This cost would need to be identified on a grant proposal but ICL's request for confidentiality did not allow this to happen.

25.10 Abandonment of Colour Display

ICL struggled throughout 1983 without getting any support for the product from either DoI or SERC. As a result, progress was slow due to lack of funds. By the end of November 1983, the date for the first prototype had slipped to January 1985 with a possibility that it would slip even further to April 1985.

ICL had found problems with the fast I/O connection between the PERQ-DAP memory and the display. They had little technical experience in this area. As a result, ICL were proposing to produce an initial system without a video channel hoping that funds to develop the video channel would be available later.

They were planning to go to DoI yet again in an attempt to get funding for pre-production orders.

Two years had elapsed with SERC's position in terms of supporting the product being approximately the same as when it had started. Almost no progress had been made by the SERC committee structure as to how grants would be awarded, who would coordinate the programme and who would fund it.

By October 1985. SERC had still made no decision as to whether they would make a positive commitment to the PERQ-DAP.