Jump Over Left Menu
USA Visit March 1983
- 1. Overview
- 2. Office Automation Conference, Philadelphia
- 3. Ithaca Intersystems
- 4. IBM: Yorktown Heights
- 5. Celco
The reasons for the trip were varied. I have produced more detailed reports of each area for a distribution to the people particularly concerned. This short overview is for wider circulation.
2. OFFICE AUTOMATION CONFEREBCE
This was a large conference, 25,000 delegates, with an exhibition. On the hardware side, the major new product was the APPLE LISA system which is a bit-map system selling for $10K, 68000 based, with a great deal of office software on it. APPLE have clearly decided that they are getting into the Office Automation area in a big way.
Another exciting product was the GRID electro luminescent display. To my knowledge this is the first flat panel display really to hit the market. It is very stylish with an integral phone and is a major technological breakthrough.
Finally, there is an SL-l Displayphone which integrates a terminal into the SL-l management phone system and can be used as an IBM screen terminal. One organisation in Philadelphia is using it with PROFS.
More minor but interesting products included:
- A box to convert AES wordprocessor to a Disp1aywriter.
- Optical character reader which also recognised the format and turned it into a Disp1aywriter document.
- A number of laser printers.
The Conference itself showed us to be ahead of many companies in some ways but behind in others. Most companies seemed to have been through the word processor jungle. Many had not done any pilot studies on mail. Hardly any had an integrated Wide Area Network.
I list a number of observations relevant to RAL:
- The only people who really seem to have a total office automation system are the large companies with a large number of paper transactions - banks, insurance companies et a1.
- Everybody used VISICALC or similar products for such functions as financial planning, FYFL etc. It is an area where we are way behind and software is available for hierarchically breaking down expenditure etc without having to recalculate totals each time.
- The successful projects had tried to hit single areas at a time getting them right before moving on to another. You need a focus to the activities.
- Quite a few people used PROFS successfully - all seemed to use to Disp1aywriters for secretaries and typists. Most liked it.
- Everybody expected to have a workstation on every desk. Many had them now.
- People had produced successful systems using mainframes, minis and micros. Many had several systems linked together. Major emphasis was a consistent image to the user irrespective of how it got done in the background.
- Most organisations had created a new Information Systems Development organisation separate from both the DP Department and the conventional Administrative Division.
- Everybody said it would take a long time. Do not expect results in under two or three years - you probably need seven for a complete system. Management must be fully committed to it - financially and personally.
- It did change the organisational structure of the company sometimes in unexpected ways.
Finally, there was the largest percentage of women I had seen at a conference. About 30-40% were women and that included the speakers. I have some cassettes of about seven of the sessions if anybody is interested. I have produced two lengthier documents - one on the Conference sessions and the other on the industrial products in the Exhibition.
3. ITHACA INTERSYSTEMS
I visited II to look at their GRAPHOS colour terminal which is a 68000 based terminal with a novel architecture. It allows the user to define graphics segments to be sent to the display. These can be displayed on one of a number of virtual pages. Thus the user conceptually thinks he has a set of pages on which he has placed graphical information.
The screen of the GRAPHOS terminal can be split up into a number of windows and each window can display all or part of a page. Several windows could define different parts of the same page.
The terminal is a powerful graphics workstation, not a competitor to the PERQ as it is not stand-alone, but it could be used as a GKS workstation or as a PERQ add-on.
Conceptually, it is rather similar to the PERQ and nicely engineered. II turned out to be a typical small USA company - rather better organised and longer established than Three Rivers but no bigger. The cost of the basic system is $8000.
It would be a useful device. I have about 200 pages of information concerning it which needs studying.
4. YORKTOWN HEIGHTS
This was mainly to sort out the exchange for Graham Robinson. There is a person who may be able to come and it will either be September 1983 or March 1984 if it happens.
Yorktown are moving to IBM Personal Computers in a big way, expecting to have over 1,000 by next year. Dave Toll is working on a project involving them.
Their office automation is totally based on IBM screens and all secretaries and typists use it. They will move eventually to using IBM Personal Computers.
The major change envisaged by them in the CMS area was the move from EXEC2 to REX as the main facility. The power of REX is equivalent to a programming language and would be a powerful tool for the future. I have a copy of the manual.
The CELCO Colour System for film recording is an impressive piece of engineering. It is of high quality, very robust, easy to operate and may well be the major contender for replacing the FR80 colour film work.
Graham Robinson and I visited their factory in New Jersey. In every way, it is a better product than the FR80 for colour work.
The major drawback is a complete lack of high level software and this is necessary as it is basically a raster device.
Management Graphics of Minneapolis (an off-shoot of Three Rivers!) does have software and Paul/Chris will need to see if it fits our needs when we come to evaluate the system.
Both Graham and I were very impressed by the quality of the hardware.
If anybody wants the more detailed reports, please contact Carol.
- Office Automation Conference, 21-23 February 1983.
- Office Automation Exhibition.
- Ithaca Intersystems, 25 February 1983.
- Yorktown Heights, 28 February - 1 March 1983.
- Visit to CELCO, 2 March 1983.
OFFICE AUTOMATION CONFERENCE Philadelphia, 21-23 February 1983
I had been in two minds as to whether to come to the Conference or not. I had to be in the USA on the week following and, eventually, I decided that it was probably worth attending. In fact, it turned out to be much more worthwhile than I had envisaged. The conference was the fourth in a line of annual conferences which had grown each year (exhibition space was 40% up on last year) and was well run by AFIPS. The size staggered me. There were approximately 25,000 delegates.
The slide facilities were not too impressive. On the other hand it was possible to get deaf and dumb language translation and also audio cassettes of each session about half an hour after they had finished. I will try and get a few of these before the end of the Conference (I am writing this at the end of the first day). Incidentally, the TV this morning was arguing the pros and cons of professional note-takers. Apparently, small companies are setting up near the larger universities who have professional note-takers sitting in on lectures and then selling the results to students!
The conference started with the usual keynote speech which proclaimed the brave new world. Companies now have Vice Presidents in charge of Information Systems. People employ Office Automation System Specialists etc. One or two interesting statistics were given. The average US farm worker uses $50K of equipment while the average blue collar worker uses $20K. The office worker often uses much less than this.
He made some points around the jumping too early on the word processor bandwagon. Many companies had installed Word Processor centres, in his view, mainly because the secretaries and typists liked them. As only one out of every 13 knowledge workers were staff of this type, it had made almost no impact on the quality of the decision-making in the companies concerned.
2. OFFICE ARCHITECTURE
This was after lunch on Monday. Three speakers from IBM, WANG and XEROX gave their view of why their architecture was the one for the future. All used the OSI 7-layer model as a starting point. The IBM man spent most of his time talking about SNA. The WANG man made some interesting points:
- WANG have implemented TELETEX.
- Did not believe it would ever handle real time speech.
- Had probably made a mistake in not going for a higher bandwidth.
John Shaw of Xerox was the most impressive of the three giving a good rationale for Distributed Interactive Computing with servers etc. He Dade a strong attack on the PBX fraternity, saying that with a maximum throughput of 64K bits/sec, they would never be effective data carriers. Emphasised the need for a document format standard. Made one or two attacks on the OSI 7-layer model (apparently ANSI have rejected it as a practical standard, keeping it only as a conceptual model). There is apparently a paper going the rounds by Danny Cohen about how important the Magic Number- 7 is. We should try and get a copy.
Xerox had one or two specialist workstations including STAR-like terminals for INTERLISP and Engineering Design.
The best quote of the session was to a question asking when data, voice and telephone system would be integrated. Reply was that just because I had pipes in the house for hot and cold water and sewage, it did not make any sense combining them down a single pipe.
3. ADVANCED WORKSTATION CONCEPTS
This was due to follow a session in another room (there were 6 parallel sessions) called Tomorrowland. It was still going on when I arrived and had obviously been very lively and rather futuristic. I will try and get a cassette of it.
The Advanced Workstation session had two people from The Wharton School at the University of Pennsylvania who have an enormous research contract from the US Navy, and a third person from the APPLE LISA project (most eyecatching system in the exhibition).
Again there was not a great deal of difference from our PERQ philosophy. University of Pennsylvania were using a variety of terminals interworking including Apollos, GRID-COMPASS portable terminals and some Burroughs fileserver (OFIS file).
He mentioned the speech devices in the exhibition. NEC have a $70K continuous speech recogniser and there were also some general speech recognisers.
They had implemented VISICALC using a voice input system and had done a controlled experiment which showed users preferred it even though they were slowed down by the recognition speed.
Barry Smith of the APPLE LISA project discussed the product. It had been designed as a Personal Office System. They estimated that there were 32 million office/knowledge workers in the USA and about the same number in Europe. All needed a personal terminal! LISA is a low cost 86000 based PERQ (more details in exhibition/glossies). He made a large point about how well the software was integrated. APPLE had invested 200 MY in the LISA Tool Kit alone. The major new software of next few years he estimated would be LISADRAW (text processing for pictures) and LISAPROJECT (VISICALC for project managers).
The discussion session raised the question of whether the three operating systems CP/M, APPLES and XENIX could all run together on LISA or whether it was one or the other. Unfortunately, I never caught the reply but it sounded like it was either/or. I will try and get a cassette of this session.
4. LASER TECHNOLOGY
Tuesday was a less good day. The two talks on Laser technology from Philips and Xerox were spoilt by the first not talking English too well and the second describing an impressive system for the National Bureau of Standards but using products that the company do not intend to market.
The Philips speaker, Jan Kimble, promised a hand-out if you went to the Philips stand. I tried three times before giving up. They said that they would send one on. Main points of Philips system were:
- 1050 Mbytes per side.
- 2 Mbits/sec read.
- Access time 200 millisecs.
- Philips are putting together a complete system called MEGADOC which included an electronic scanner for capturing images and a laser printer.
- Uses CCITT standard GR3 HR for data scanning and CCITT 2D for data compression - whatever that means!
- Single disc system can store around 25,000 pages of captured data.
- Juke box system being produced with 64 discs, 2 selectors with 15 sec access time. Holds 2 x 64 x 25,000 pages.
Absolute disaster of a session. An innovative chairman decided to run it as a chat show with audience participation. They had broadened it out to include remote video conferencing and audio/graphics as well as electronic conferencing which I had assumed it would discuss. I gave up half way through and went back to the Exhibition.
6. MAINFRAME, MICRO OR MINI
The aim of the session was to explore which of these you should use for your office automation.
The chairman, Tom Billadeau, was quite a character, knocking all three I enclose a paper by him as an Appendix.
The first speaker from FORTUNE extolled the virtues of their micro system (68000 based, UNIX + display). His main arguments for the micro approach were:
- Micros use standard 68000 chip set and run UNIX. Therefore, much more compatible than the minis.
- Designed for non-programmers.
- Software costs from vendors low due to high quantity of sales. $600 on micro may cost you $30,000 on a mini.
- Many systems re IPL themselves and regenerate the system if you add more memory, while mini needs your local systems guru.
The mini speaker was from Data General and effectively said we shall still be around providing server-like functions and acting as PSE switches etc. Not convincing at all.
The mainframes were represented by IBM who said the others had their place but a total solution needed a mainframe to integrate the- office products with the other functions carried out centrally.
He said that IBM had two standards on top of SNA: DIA : Document Information Architecture (sounded like a mail protocol). DCA : Document Content Architecture. DCA recognised two kinds of documents: Final form text document. Revisable form text document. This could be of interest to the PERQ project. Do we have local knowledge?
They obviously correspond to the PRINT FILE format and DOCUMENT TRANSPORT format which ICL continue to muddle up in one protocol.
As IBM are behind them, they could become de facto standards. We should look at them.
I managed to get cassettes of about 6 sessions - mostly ones I did not attend but overheard people saying encouraging things about them. It will give others the opportunity to get a flavour of what was presented.
7. ORGANISATIONAL STRUCTURE
Three rather different talks all discussing impact of office automation on the structure of an organisation.
The first speaker said that office automation transformed raw data to facts to knowledge to wisdom and therefore power. People who provide OAS will therefore be a major source of power. She felt that current hierarchical or matrix management would change with the advent of OAS to a more network management with individuals being in charge of one function and subservient on another.
The second speaker discussed how it had changed a bank's overall organisation. This particular one pushed 1.4 million pieces of paper per night I They had managed to reduce this vastly by automatic teller systems in local shopping areas. Effect on the bank had been to devolve management decisions rather than centralise them.
8. USER EXPERIENCES
gain, three speakers - a government consultant, a railroad, and finally, life insurance. This was a good session and I will try and get a cassette of it. The railroad will be closed before the end of 1983 unless they make a profit. Consequently, OAS in their case was designed to save real money. They had saved $1.2M already in real money with a $2 return on every $ spent on OAS.
General comments agreed by all:
- You need an organisation whose sole function is OAS.
- It needs to have enough manpower and get a few projects done well rather than try and attack everything.
- It is a 5-year programme to really scratch the surface.
- Management needs to commit real money and resources. Typically $1M per year for an organisation like SERC and it will save $2M.
- Produce a complete plan for management to underwrite.
Particular interesting facts that came out:
- Consultant worked with small team consisting of a manager, 2 systems analysts and 5 programmers yet they had done significant tasks.
- Out of 17 proposals made to government, 15 had been done successfully.
- The railroad had 19 departments using 21 different wordprocessors, when they started. His best quote was 'to build a house you first have to dig yourself a big hole! '.
- The insurance company used PRIME OAS with 850 terminals for 1500 users on 5 PRIME systems in 2 separate buildings. They charged people's projects for access to the system. Also had it connected to their two IBM 3081s . Never managed to get the calendar/appointment scheduling to work. Only 10% used it in small pockets. Estimated savings $3M. Savings of managerial to executive time estimated between 5 and 12%.
8.OFFICE OF THE FUTURE - 1985
This was interesting in giving a view of where USA felt we would be 3 years hence. Disappointing in the sense that it was not much further forward. Major points made were:
- Every worker would have a terminal.
- CSX railway were a good example of progress towards complete OAS. The company was formed out of 3 railways and instead of reallocating 500 people a la Swindon, it was decided to put in strong OAS communication and have a central office of 47 people. It uses PRIME 750!
- Ellenby of GRID made general point of understanding user requirements and in particular insisted that space in office had to be minimised thus the decision not to manufacture CRT devices. Also, flicker on CRTs made terminals obtrusive which was main reason for non-acceptance.
As always with conferences of this type, the Exhibition and discussion with individuals is of equal value to the paper sessions. My overall impression is that, apart from a number of large companies who are inherently paper pushers (banks, insurances etc), nobody has really got into office automation in a total sense. Instead they have either worked down from the DP department or up from the typing pool. The companies that had got complete systems installed were quoting real savings in money and manpower together with many ancillary benefits such as better management decisions and faster ones.
If anything, word processing was not regarded as an issue. Everybody had word processors and were trying to get rid of their obsolete equipment.
A major difference from RAL was that almost everybody that I spoke to did management accounting, resource allocation etc automatically. They could not believe that we had effectively a manual method of doing the FYFL and yet had got round to considering mail facilities etc. Everybody used VISICALC or some other system for financial planning.
In the fringe areas, there was much more going on than in the UK. Many people had systems for storing voice messages and relaying them. Teleconferencing using voice was common and several had video systems. One or two people had video conferencing installed on local sites to stop people having to move about to attend meetings.
ITHACA INTERSYSTEMS - GRAPHOS TERMINAL 25 February 1983
I visited Ithaca Intersystems primarily to see if their GRAPHOS terminal was a feasible contender for a 68000 based PERQ alternative. We had very little information but it did seem to be based on a window management system and was 68000 based.
Other possible uses might be as a colour terminal to add-on to a PERQ, as an intelligent terminal to replace the INMOS 68000 based systems, and finally as a GKS workstation.
To avoid unnecessary reading, it is clearly not a single user system, but could be of interest as a terminal.
2. COMPANY BACKGROUND
Ithaca is a small American college town with almost no industries apart from supporting Cornell University, a typical large Ivy League establishment similar to Harvard and Princetown but situated on a hill.
Ithaca Intersystems was started by a Cornell engineer in his apartment, buying chips in bulk, adding a small mark-up and reselling. This led to producing individual boards aimed at the micro market but looking specifically at the high performance end. Following this, the Company developed small computer systems based initially on the Z-80 and later the Z8000. The systems are based on the S-100 bus. and typically sell to OEM customers. They have shipped in bulk to NASA, Lockheed, Shell Canada etc. The areas that the systems are used in range from scientific to industrial with a strong market in the process control area. Their systems are typified as being in the high price bracket compared with other similar systems but providing higher performance.
They have recently repackaged the systems into a single box and both are now called ENCORE systems. DPS-8000, the Z8000 based system uses Xenix as an operating system (all the literature says COHERENT but they have recently changed). You can purchase it with a variety of floppy or hard disc drives. It is designed as a multi user or multi tasking system. II have developed their own PASCAL compiler which is very efficient and which they sell as a product. It compiles down to machine code rather than P-codes. Associated with it is InterPEST, a symbolic debugger. They claim it runs at about 4 times the execution speed of competitive products and generates code about half the size. I have a full set of brochures on the product.
The Company has its own building about five miles out of Ithaca on Hanshaw Road. This is owned by them, not rented. Until recently, everything was done there but in the last few weeks they have rented two floors of a large block in downtown Ithaca for sales, program development etc, leaving manufacturing at Hanshaw Road. The International Saleswoman is Jan Potts, a person with a Business degree whose husband is a biochemist at Cornell University doing a PhD. She has been with the company about 18 months. Any future contact should be through her. The Chief Executive of the Company is Bob Kleine. Most of the staff seem to be shareholders and it is reminiscent of early PRIME or 3RCC days. The size of the company is about 100 people with about 50 on each site. The company's main financial backers are Olivetti and OAK. The overall impression is a slightly haphazard company but not as slippery as 3RCC. They really do not do any major production. They design their own boards which are made outside. The boards are populated by hand and sent out for flow soldering. They use their own systems for diagnostic testing with a soak test of about 72 hours at the end. The production area is about 100 feet by 40 feet wide. They are certainly not cramped at the moment. However, their volume capacity for GRAPHOS cannot be much greater than 3RCC at Gross Street. One advantage they have is that there is almost no high technology work in Ithaca apart from them. Consequently, they only lose people to places like California rather than round the corner.
The current situation is that they have shipped out 10 systems to Beta test sites (one is a company in Vancouver) and they are just updating the software on those sites to the full release. They estimate that they can deliver systems from now within 2-3 weeks. For overseas, this could be 4-6 weeks due to having to get Export Licences etc (apparently Washington are quite strict at looking at 68000 based systems whereas the Z80 and Z8000 equipment is almost a formality).
hey have sold quite a lot of their computers in Europe and have agents in Switzerland, Italy, Germany etc. They had an agent in UK, Euromicro, but have dropped them. Olivetti England have an option to service the systems, although purchasing would be done direct from Ithaca.
The Design Aim was to produce a high quality colour workstation aimed at working with international standards (GSPC and GKS) when they arrived but using the terminal to do nearly all the graphics processing in a multi window environment. Unfortunately, as they were doing most of the design work about 3 years ago, it does not fit too easily into either CORE or GKS philosophy but it is possible that may not be true if we take a close look at it. Effectively, it is an RS232 terminal to which you can send graphics information. The exciting features are that the screen can be split into a number of overlapping windows which can display graphical pages. A graphical page may appear with different scaling etc in more than one window. The graphic information can be stored in segments or output direct to pages. Segments can have a priority associated with them so that it is very easy to move items in a picture in front or behind other items while at the same time having the pictures appearing in windows which are also being moved or positioned relative to each other. The main application area envisaged is business graphics. It works very well as a desk top with bar charts and coloured pictures on the pages. Scrolling of text is as fast as the PERQ, as is moving of windows.
They are showing it at NCGA this year and possibly SIGGRAPH. The demo showed the usual PERQ-like SIGGRAPH demo but in colour - different colour tables for each window so that the 16 colours per window still allow quite a rich colour scene on the display. The phosphor is a longer persistence than the PERQ which gives a slight ghosting if you do the animation type of demo. For example, they had a dice bouncing around with the orientation changing which had a slight ghosting but not too bad. It is very restful on the eyes.
The only peripherals attached at the moment are the keyboard and a tablet (either Summagraphics Bit Pad or Hitachi TIGER Tablet). The tablet input looks like a LOCATOR and returns NDC coordinates relevant to a particular window. The software is being updated so that it can also act as a PICK device. Effectively, the software will allow you to define the tablet as a different device in each window. At the moment, it is possible for the LOCATOR to have different echoes in different windows. They have an S-100 bus interface so any S-100 device can be attached. They are looking at HP Plotter and Hitachi. They are also going to interface either a Matrix or DUNN colour printer. Coordinates are actually sent to the device in world coordinates as 32bit floating point numbers. Consequently, it is possible to get a much better output on a pen plotter or the colour printer than appears on the 480 x 640 display. The GRAPHOS terminal can therefore be used as a quick view terminal for slide production with the actual picture going to the DUNN or MATRIX being a higher quality image. They are looking at interfacing the PRISM printer. The general view was that the last few months had been spent getting the field release software up to first release status. They would now freeze it for a while and concentrate on getting the set of peripherals supported up. They have tested the system at 50 cycles without any problems.
Rather than attempt to give a full description here, I will attach the general fact sheet to this report and just note that I have a complete set of manuals (200 pages) for anybody interested.
Some bits of information picked up which may or may not be of use:
- They started modelling the CORE but have moved to GKS as the releases came out. They do not have GKS 7.2 document yet - I promised to send one. Person in charge of software development is Jeff Moskow.
- Ideas for the terminal came from an ex-Cornell engineer, Kells Elmquist who was also interested in animation.
- Jeff Moskow wrote the PASCAL compiler. He is probably around 30. Had his arm in a sling resulting from a pulled muscle when he was moving a desk to the new block. Its that kind of company. Everybody mucks in I They have a weekly Friday morning meeting between engineers and software staff.
- GKS Cell Array not implemented but the rest of the primitives are.
- Only CHAR precision text in four sizes WxH, 2WxH, Wx2H, 2Wx2H. If you define characters in graphical mode, they are defined in world coordinates and you can have LEFT, RIGHT, UP and DOWN.
- They can emulate both VT100 and 4010 order codes. You can have 4010 going to one window, VT100 to another and the native order code to a third. For those native mode orders which are not incompatible with T4010, you can output in 4010 mode. Effectively, you can extend the T4010 order code to include circle generation etc.
- They have two modes of drawing. One checks for segment priority, window overlaps etc whereas the other does not. May use second for sketching.
- They have 16 hatch styles and 10 markers. They had one demo with 16 circles each slightly larger than the other filled with a different hatch style. Defined them initially with next largest on top of old and then changed the priority to the other way round to get a set of concentric rings each with a different pattern. This is quite slow - possibly two secs to do a 2" diameter circle but it is doing quite a bit of work.
- All the colour table work can be done in RGB or HLS.
- Each window can have a border and a title added to it.
- They are aiming to build 20-25 a month in the first few months.
- The code is broken down into a number of processes with a simple scheduler. May put in separate processors at some stage for particular processes. Examples of process are Refresh Screen and Draw Line. Thinking of adding a user process that he can downline load.
- They would commit to producing a GKS workstation if they had orders for 100 systems depending on it.
- They may do a GPIB interface. However, once you get 9600 baud connection, system is pretty well flat out so that improvements in performance would not be great.
- They have a 3COM-S100 Ethernet connection.
- Their UK agent EUROMICRO had been talking to ICL as an OEM customer. Nothing came of it.
- Avery in the UK at Liverpool Univ (or Poly) had been enquiring about Graphos.
- They seem to have a lot of interest in USA and abroad. Jan Potts had had 5 international visitors this week.
- The terminal uses a Mitsubishi tube.
It is an impressive device which could be of use as a terminal. There are sufficient differences from GKS that I think it would be difficult to act as a GKS workstation as it stands. Either Dale, Julian or Chris should look at it in more detail. If we wanted to add it on to the PERQ as a colour terminal, it is clearly feasible. It has enough similarities to the PERQ that the same software might be able to drive both. The indirect operations on the GRAPHOS correspond in many cases to a set of RASTEROPs. You would need to work at a higher level. If we do need to get a Colour PERQ early, this could be the cheapest solution in both software developments and hardware terms. The trip to Ithaca turned out to be more useful than I expected.
YORKTOWN HEIGHTS 28 February - 1 March 1983
The major motivation for the trip was to sort out Graham Robinson's exchange with Yorktown and to see how Dave Toll was getting on. Ancillary reasons were to talk to the Computer Service side particularly in the areas of office automation and microfilm recording.
2. HARDWARE AND PERFORMANCE
Since I last visited, the Laboratory had had a complete reorganisation of hardware and now had three 3081's and were expecting a fourth in April. The three systems had 401, 423 and 274 users logged on while we were there. Having said that, most people log in first thing in the morning and stay logged in all day. The official view is that about one third of the people are active at anyone time. The Dave Toll view is that the systems often have only 25 users logged in! Needless to say, performance is very good. Their aim is to provide a service where most trivial commands get turned round in 1/3 sec and a stretch factor of 3 for non-trivial commands. I spent some time talking to Walter Doherty concerning performance aspects (he is the Jed Brown equivalent + Tony Lobley). Some main points of interest were:
Nobody uses EXEC2 any more. All use REX and this will be a software product by June. The view was that this was a real programming language in its own right rather than a macro system. It allows EXECs to be debugged interactively and you can also stop an EXEC and do symbolic debugging in the middle of execution. General view was that it was a major advance over EXEC2.
The only way to get interaction of quality was to use screen terminals. You could reach 400 interactions per hour using screens but only 150 over telephone lines running at 4.8 Kbaud. Yorktown have 1700 screens.
Yorktown can access ARPAnet if you have management permission. Dave did not seem to be aware of this.
I had about an hour talking to Dick Kelisky. He has recently been promoted so that he is now responsible for all services at Yorktown including computing (library, buildings, restaurant etc). AI Weiss, who was responsible for the CMU single user project, is taking over from Kelisky. There seems to be a major thrust towards the IBM PC with 500 already committed. Each would have the ANR card to allow it to run as an IBM screen terminal. With the ANR card, the PC outperforms a screen terminal due to buffering available. They now have a single operator console for their whole system. A 4341 receives all operator commands, analyses them, and produces a single display defining the status of all systems.
4. JOE WELLS (SLAC)
Joe Wells from SLAC is visiting Yorktown for a year and I spent some time talking to him. Some points that came up in conversation:
The 3380 discs have performed very well. For sequential I/O they are faster than 6250 bpi tapes. Consequently, SLAC now stage tapes to disc. A set of MAXI DISCS have been defined. User groups can mount specific tapes on them. as a result they have been able to reduce operations staff at weekends by insisting that groups stage required magnetic tapes to 3380 before the weekend. SLAC have REX and are very pleased with it. They have developed a number of Virtual Machines to control functions. A cheduler turns off batch if stretch factor exceeds 7 and starts it if it gets better than 3. To extend 3270 screen-like terminals across campus, SLAC have installed LEE-DATA terminals because they can be used to emulate large or small screen terminals. It is also possible to multi drop a set of screen terminals off a single cluster controller connection. It is also possible to run screen terminals remotely with a single twisted pair wire between box attached to controller and box attached to a set of screens. SLAC have been impressed by the STC solid state drum. They can get a throughput of 3 Mbytes/sec.
5. PERSONAL COMPUTER
Dave Toll, for security reasons, could not say anything about his work on the IBM PC. Information gleaned from undefined sources:
A New PC Mark 2 is soon to be announced. It can have 10 Mbyte fixed discs and an optional extension crate. PCs run faster than screen terminals using ANR card. CMS runs on the PC. Current software for PC is slow and bug ridden.
Yorktown do all their secretarial work using screen terminals and home-grown software. They are just about to move to PROFS. They will change from screens to PCs with possibly Displaywriters as an interim stage. Absolutely no problem with secretaries using screen terminals as long as you teach them properly. They plan to do down-line loading of new versions of PC operating system and also archiving from PC to mainframe automatically. They give each new Secretary on arrival a 5 x 1.5 hour course on XEDIT which covers the keyboard, log-on, filing and editing. A separate 5 x 1.5 hour course is taught later on document formatting. Courses are given about 6 times a year.
The major motivation for going to displaywriters at the moment is that their screen terminals are obsolete and the printers no longer supported. They were quite emphatic that they were not going to displaywriters for functionality. In fact, secretaries would still do document formatting using VM and the displaywriter as a screen. They emphasised the need for secretaries and typists to be taught by one of them. They have a person, K C Keane, an ex-secretary, who is responsible for all training and advice. I discussed with Kelisky the possibility of her giving her Editor course at the Laboratory, or one of our staff to attend the Yorktown course to get some idea of it.
Sadagopan had arranged for me to talk with Don Rozenberg (Room 24/234) who was interested in a stay at the Laboratory. He is involved with their 68000 based programme and has purchased both PERQs and APOLLOs. He may only be able to come for 6 months. I agreed to send him a RAL report and other relevant information concerning the Single User Programme. His background is also in CAD design work. If he comes for a year it would be from next September, but, if only for 6 months, it would be from around March 1984. He has a daughter of 16 that he was slightly worried about schooling.
VISIT TO CELCO New Jersey, 3 March 1983
Graham Robinson and I visited Celco in Mahwah, New Jersey to see and discuss their colour film recorder as a possible FR80 replacement. The Celco plant is just outside Mahwah on Route 17. You can only reach it by travelling North on Route 17. The access to the building is by turning off into its driveway. . If you are coming to it from Route 287, you will have to go past it to the Ramsey exit on Route 17, turn round and come back on the northern carriageway.
We spent all the time at Celco with John Constantine and Carl Ludwig, who I had met before at SIGGRAPH. The company was started in 1950 by John and Michael Constantine's father from a disused chicken house. The company has always been involved in precision optics and cathode ray tube yokes particularly in the flight and military areas. They still produce a great deal of aircraft instrumentation and large radar tubes as well as supplying precision optics and yokes to the rest of the film recorder industry. The company is about 100 strong at all sites and is still very much a family concern.
3. HARDWAREThe main film recorder is the CFR-4000 which is a high quality colour raster system. The major work done by Celco is the optical work, including all the associated metal work and putting the complete system together. The aim has been to produce a high quality colour raster device which produces output frames very quickly. In their view, line drawing systems are unlikely to be viable in the next few years except for very specialised applications. Any system will need to produce filled area images in the future and the line-drawing systems such as the DICOMED do not have the speed in fill area mode or accuracy to compensate for their line drawing advantages.
The large amount of NCR equipment for production of the necessary mountings etc indicates where Celco's strength lies. It is a high precision product with excellent optics which means that they are able to run the Litton tube at a fraction of our current FR80 settings (only use 1.5 microamps). As a result, they have not had a tube failure in the 25 systems out in the field even though several have been in use for several years. This compares with the 6-month life on our FR80 and 1-year life on the DICOMED).
The system can run with either a FOROX or Marron Carrel 16/35/Vistavision format camera. The FOROX is a development of the animation stand camera. It has a single sprocket wheel which the film goes over on both the input and output film paths. It is the more rugged design of the two but requires the film movement to be changed from 16 to 35. The Marron Carrel has a long straight film path which should improve accuracy and has a combined sprocket system that allows both 16 and 35 mm film to be mounted without changing the movement. It is the more precise camera of the two and the one I prefer. Having said that, most of TRON was shot using the FOROX camera.
The Marron Carrel camera allows you to back up the film and re-expose which is a facility unavailable on the FR80. Both cameras have alignment grids allowing good and simple registration of the camera relative to the tube.
The colour wheel has a transparent filter and the three primaries in a wheel. It is very well made and designed so that it is impossible to get a wrong filter in place.
Celco are producing their own camera and this should be available within the next 6 months.
On the electronic side, they do their own multi layer board design which is manufactured externally. They populate boards by hand and handle solder. They have just tried getting boards flow soldered outside. Due to the high light output, they are able to use standard Kodak Ektachrome film.
I found the hardware very impressive. It is fast, well engineered, of high resolution and looks completely reliable. The Celco engineering is very professional. The CFR-4000 has a more flexible optical system which in our environment is probably not worth having. The CFR-4000B allows 16/35 and one other media. The combined 16/35 camera moves backwards to expose the other camera mounted above it. We could purchase a CFR-4000B, for example, with an 8 x 10 viewgraph camera on top of the combined 16/35. The adjustments on the system are very straightforward and should not produce any major problems. Maintenance in the UK could be a problem. Celco felt that our best solution would be to train a technician from RAL at Celco's and do it ourselves. It requires so little regular maintenance and most faults are so easy to diagnose that a maintenance facility in the UK was unlikely to be viable. Most of the current owners do their own maintenance.
The Celco system accepts pixel data or run-length encoded data. The most popular approach is to interface it to a small DEC system such as a PDP11/34. This would be their recommended solution. However, the Yorktown Heights system has been connected to an IBM mainframe using the AUSCOM box. It sits on the UNIBUS and could equally well run off a small VAX.
Now the bad news. Celco's software is non-existent. They have some very basic software running on a PDP1l but it really is of little value. However, Jim Teter, ex Three Rivers, at management Graphics does have software. Jim Teter and Dave Glass of Three Rivers left the company with the rights to market the Three Rivers colour system when Three Rivers decided to just develop the PERQ. Teter had also been largely responsible for the Mellon Bank Teller System work using colour terminals (low resolution). His company, Management Graphics, has added a PDP1l/34 to the Celco system together with the Three Rivers colour system to form an integrated workstation and colour recorder mainly aimed at the business graphics area but also looking at the animation market. The software runs under RSX-11M and will be converted to run under VAX VMS.
On the colour recorder side, it allows you to spool a number of jobs for the recorder giving the operator the ability to influence the scheduling. The system automatically adds identification information to each frame outside the visible area. There is the ability to produce accounting information for the jobs run even to the extent of calculating the running cost for a job.
The software accepts high level metafile commands and performs the rasterisation operation. It has stored fonts in run-length encoded form and translates user text requests into raster information. Effectively the output to the Celco consists of one graphical image and 3 sets of intensity information. The Celco system processes the information with the relevant intensity information for the first primary, then the second and then the third. The colour wheel rotates only once for each frame. Only the 8-bit colour table is supported by the software. The software allows other tasks to run in parallel with the film processing. The two major ones are font design a la PERQ and Palette definition using RGB rather than HSI.
Whether this software would be of value to us is unclear. Somebody needs to talk to Management Graphics. If this is not of use, I think that we shall have to produce our own software. The metafile format used by Management Graphics is BGL, the format used for the Mellon Bank terminals. Teter has been considering moving to the ANSI Metafile standard when it is defined. The current software has good background and windowing facilities to reduce the information that needs to be redefined. Consequently, he will require the ANSI standard to be at least as flexible.
The only new product likely to come out from Celco is a lower quality product that they have under development aimed at the complete workstation for producing business graphics at a much lower price than the current system ($50K was mentioned).