This is the first issue of a newsletter produced for actual and potential users of the SRC Interactive Computing Facility. We hope to publish an Issue at about six weekly intervals. If you wish to receive future copies please write to the address below. Where a number of people at the same institution are interested in obtaining copies it would be appreciated if they could all be sent to a single address. Spare copies for display may be requested. Please state clearly the number of copies you need and the address to which they should be sent. A form has been provided at the end of this issue which you may like to use. All communications about the newsletter should be sent to The Editor, Rapid Response, Atlas Computing Division, Rutherford Laboratory, Chilton.
Interactive computing is a way of using the computer where the user can talk directly to his program. This allows him much greater freedom. It is possible for the program to prompt the user and to notify him immediately when errors occur. Because it does this he can often correct errors on the fly preventing the particular run of the program from being a waste of computer time. The user's time is often used more efficiently since he is given the results of his computations at once.
The most obvious difference between a program written to run interactively and one that is not shows up as soon as the program is started. The user does not have to specify off the cuff all the information the program needs - instead he can wait until the program asks him. A good program will only ask for the information it needs for that particular run and will indicate the set of possible replies that he can make to it.
For some disciplines the use of interactive computers is almost a necessity; this is the case when the degree of complexity of the program is great enough to prevent the user from being able to predict in advance the inputs it will demand. Interactive working can also be very valuable for checking and correcting large amounts of data before submitting it for processing to a large batch facility or for rapidly examining the results of batch runs. In both these cases a good interactive editor (used to examine as well as to edit) is needed.
The price paid for this is extra computing facilities required. The system being used must be powerful enough to respond quickly to its users even though for much of the time it will be doing very little - the user will tend to spend much of his time thinking, ie leaving his terminal waiting for him to type in a response to a previous message.
There are two ways of looking at an interactive computer. How many people can it support? How many of these can there be using the machine at the same time? - In the jargon, a channel is a pathway in to the computer - a user terminal on line to the computer will occupy one channel. If we assume that our computer is available for 50 hours in a week (e.g. Monday to Friday, 8 am to 6 pm) then each channel can be used for 50 hours also. Let us assume that a single user needs to be in contact with the computer for 5 hours a week it follows that each channel will support up to 10 users. This ratio of channels to users is assumed adequate for the interactive facility.
In practice the facilities will be available for most of a 24-hour day. Operator coverage will probably only be available during the prime shift.
In the past few years there has been a growing number of requests to the SRC from engineering departments for the provision of interactive computing facilities for individual research projects. This growth led the Engineering Board to appoint a Working Group to establish the needs of engineers in the universities and polytechnics and to see if a rational framework might be constituted for the provision of facilities to meet these needs.
The Working Group recommended that the SRC should set up computer facilities at an SRC establishment to provide facilities for interactive computing to engineers in universities and polytechnics.
As a result, the Engineering Board set up a Technical Group to look into future of interactive computing. This Group came up with a number of recommendations. Copies of the Report can be obtained from the address below.
The facilities proposed are specifically intended to help meet the needs of engineers for those programmes or projects which are supported by the SRC using the normal criteria for research grant support. They will not, however, be reserved exclusively for the use of engineers but will be available to people funded by other SRC boards. Access to the facilities will be approved through the normal subject committee structure of the SRC.
The Rutherford Laboratory will be responsible for the implementation and management of the programme within the SRC.
Enquiries concerning the facilities or use of them, and requests for copies of the Report, should be addressed to Interactive Facility support group, Atlas Computing Division, Rutherford Laboratory, Chilton.
(This has been modified - see below)
Before introducing the recommendations of the Technical Group it is worth pointing out that their Report contains a number of interesting paragraphs illustrating the relevance of interactive computing techniques in various engineering disciplines.
The Report expressed concern about the duplication of software effort that can occur when many sites are involved. It recommends that support for major pieces of application software should be provided by one site only, and that users should be able to access it from other locations via a communications network. Even when a popular piece of software is mounted at a number of sites it will be supported from a single location.
The Interactive Computing Facility will provide users with access to medium/large computers and to multi-user minis. The larger machines will be capable of supporting about 20 simultaneous users at anyone time from a user population of about 200. The multi-user minis will permit up to six users simultaneously, and will cater for a user population of 50-60 people.
In the long term the Report expects the bulk of the power available to users to be provided by minis. In the short term, however, the main load will be taken up by the larger machines; minis will be provided only where there is a sufficiently large local community of users to justify doing so.
The Report recommends the enhancement of 5 minis currently in use. A choice of sites is not made.
At the moment satisfactory multi user system software is not provided by mini computer manufacturers (although some university and other software has been written). The Report recommended the purchase of two minis for development purposes - the aim is to encourage development of suitable software (hopefully on British manufactured machines) and to assess minis in a multi user university environment.
The Report envisages that both large and small machines will be needed for their separate characteristics and a good communication network will be needed to connect the different machines.
The Report recommends the setting up of a newsletter - this is IT - to help keep close contact between the users and the staff of the facility.
The main hardware recommendations of the report in the short term are:
(This has been greatly modified - see below)
A singe site support policy is to be implemented. No staff were allocated by the Report for software production. University or commercially available software should be obtained where relevant.
It has been agreed that staff will be needed in the short term for software development.
The Report suggests that a good interactive FORTRAN library should be given priority. Although the main emphasis in the Report is on FORTRAN - because of its ubiquitous nature - it recognised that other languages may be required in particular subject areas. This will be discussed with the user community in the Software Meetings described below.
The Report suggests that a Facilities Committee should be set up by the Engineering Board to be responsible for the general supervision of the facility. This body has now been set up under the chairmanship of Professor Rosenbrock. It has met 6 times during the year.
The Director of the Rutherford Laboratory is responsible for implementing the Report and managing the facilities.
A staff of 16 people are working, within the Rutherford Laboratory. on implementing the Report. Many have been working for at least a year on this work. A large meeting of interested parties including about 400 university staff - attended a meeting organised by the Rutherford Laboratory to discuss the proposals in the Report. After that, over 20 institutions were visited. Attention was paid to successful interactive computing centres as well as to possible users of the facility.
At the same time a lot of effort was put into preparing benchmarks and specifications for both the minicomputers and the Chilton central computer. A large number of manufacturers have been approached to determine what was available - particular attention was paid to British manufacturers.
Negotiations about the operation of the enhanced facilities at UMIST and Edinburgh are almost complete. There are always little problems e.g. do we work under Scots law or English?
University groups have been approached to see which have programs or projects suitable for the assessment and development of the two multi user minis and to determine who should collaborate with the Rutherford Laboratory.
Grant applications have been scrutinised to find out the sort of facilities that users are requesting. We need to see what proportion of the non-batch work could be done on the facility and the sort of software that would be needed. Some non-batch work is not suitable for the facility, control of real-time systems being an example.
The enhancement of the Edinburgh machine is complete and the acceptance tests are being carried out. The UMIST computer will be enhanced early in the new year.
The most serious change to the recommendations of the Technical Group Report is the decision not to buy a large central machine for the Rutherford Laboratory. This is because of lack of funds generally within the SRC. Instead. a large mini computer will be purchased. This will serve as a large multi user mini - allowing the Rutherford Laboratory first hand interaction with users in order to understand their problems.
A Prime 400 mini computer, capable of supporting a user population of 100, has been purchased for this task.
The various machines of the interactive facility will be linked into a communications network by small dedicated mini computers. These are often called gateway computers. As well as linking the processors of the facility together they are used to connect user terminals to the computer.
Two such machines will be used to connect the central machines at UMIST and Edinburgh into the network.
At Chilton a slightly larger gateway machine will be installed. This will serve three main purposes. It will link the Prime 400 Chilton into the network, connect users in the South onto a fast private line to UMIST or Edinburgh, and route work from the network onto the batch facilities at Rutherford. In this way machines within the Laboratory will give us experience with both local and remote users.
One of the two multi user minis for development work has been ordered. It is a GEC 4070 computer. It is hoped to purchase a second one in 1977-78. No choice of second machine has as yet been made. Some development work will be done on the Prime 400 also - at least until the 4070 arrives.
Because of cancellation of the Chilton machine less channel capacity will be needed in the short term. Only 50 channels will be required compared with the 100 channels in the Report.
It will not be possible to purchase any more communications or terminal equipment before the next financial year. Because of this only limited access over currently installed lines can be given to the Edinburgh computer up to April 1977.
Private telephone lines have already been installed from Chilton to UMIST and Edinburgh. Some application groups within the Rutherford Laboratory have already tried using the private lines as a way in to the Edinburgh machine. This is very valuable experience at this early stage.
It is now intended to upgrade 2 university minis in the next year. Investigations to select these machines are being carried out now.
AVAILABILITY IN mid-1978 SITE USERS CHANNELS E'BURGH 250 25 UMIST 200 20 CHILTON 100 10 upgrades 100 10 TOTAL = 650 65
The number of existing users of the Edinburgh and UMIST machines equals half the projected total number of users. We can conclude that:
*400 NEW USERS CAN BE SUPPORTED BY mid-1978*
It is very difficult for the Rutherford Laboratory to see what software will be needed. Because of this a number of Software Meetings are being held (by invitation) in a number of different subject areas. Currently meetings have been planned in the following areas
These meetings will be reported on as they occur.
The meeting on Finite Elements has already taken place - it will be reported on in the next issue.
The Report states that beside their needs for interactive computing some engineers will continue to need access to the largest available batch processing machines and some engineers will require access to large processing power through the interactive system.
The facilities at the Rutherford Laboratory are being enhanced by the addition of a second 360/195 processor and an extra megabyte of core. In due course the Rutherford Laboratory will have available two 360/195 computers each with 2 megabytes of core. The first stages of the enhancement, viz the running up of the second processor as a minimal stand alone system, is being done at the present time. The exact way in which users of the interactive facility will submit jobs to the 195 is under consideration.
A small gateway mini computer will be purchased to allow access to the 195 computer from the nodes of interactive facility network.
Here is a list of the more important topics we hope to cover in the next issue.
Details of the UMIST and Edinburgh enhanced configurations, and of the user channels into them.
A brief description of the software available at the two sites.
Network plans for inter-connecting all the nodes of the facility.
This issue has been, in a way, more history than news. Its purpose has been to set the scene for further issues. In future the newsletter will present the changing scene in as up to date manner as possible.
Don't forget to tell us if you want further issues!
By all means write to the Editor giving him your views about style and content. While we may not be able to reply individually we will certainly try to take account of any comments you make. Suitable letters (or extracts from them), or even short articles, may be published.
