Annual Report

July 5th 1982 - July 3rd 1983

Oxford University Computing Service


Contents

Introduction
Software and Service Changes
User Liaison
Computing Teaching Centre
Computing in the Arts
Performance and Reliability

Table 1. Machine Usage
Table 2. Usage by Disciplines
Table 3. Remote Site Usage
Table 4. VAX-2980 Link Usage
Table 5. Job Turnround on 2988
Table 6. Reliability and MTBF
Table 7. Response Time on 2988
Table 8. Items Input to KDEM or Oxford Archive


Introduction

At the time of the last annual report the Service’s main computing power was supplied by an ICL 2980. The unreliability of this machine was a cause of great frustration; the average reliability over the first 24 weeks of 1982 was 92.8% and the Mean Time Between Failures was 15.7 hours. During the last summer vacation the 2980 processor was replaced by an ICL 2988. The figures for this last quarter (given in Table 6) and the previous quarter show an average reliability of 98.4% and a Mean Time Between Failures of 42.8 hours. These figures amply justify the decision to undergo the trauma of change last summer. The Service feels that user confidence and satisfaction have noticeably increased.

The Service has taken advantage of the increased mainstore of the 2988 to reintroduce interactive running of most user programs and packages. This has caused no noticeable degradation of response time. Time required for both hardware and software maintenance has decreased, and an earlier start is made to the MAC service each morning.

As an economy measure the 2988 has been run unattended at night since the start of 1983. No unexpected problems have been caused by this.

The VAX service has continued to build up steadily and classes are now regularly run on VAX3. Until recently computer resources have been allocated by the Service to an individual project. The Service has been anxious to devolve responsibility for this work to the departments, both to reduce the large administrative overhead incurred and to enable some academic judgement to be applied to the priority of work. This process was started in December 1982, when 19 allocation groups were set up. This scheme appears to be working extremely well, and about 10 more allocation groups will be set up in the coming year.

Work on the Cambridge Ring has continued, albeit at a slower pace than had been hoped. File transfer between VAX machines is possible, and a print server is working in the user area. The 380Z floppy disc server has also been connected via the Ring.

One of the Service’s VAX machines is now a node on SERCNET, and file transfer is possible to and from SERCNET and PSS sites. This has been of particular benefit to Lasercomp users at sites unable to use routes through ULCC or UMRCC. A bid for funding for further wide-area communications facilities has been approved by the Computer Board.

There have been further discussions with the Users’ Group about the Advisory Service, and changes have been made to the manning rota to try to match more closely the users’ requirements.

[Top]

Software and Service Changes

General

A file archiving scheme was made available to users at Easter. The system is based on the 2988, where files are copied to a cycle of tapes, but is also available to VAX users. Archive commands on the VAX cause the files to be transferred via the local link to temporary file space on the 2988, from which they are copied to the same tape cycle used by files originating on the 2988. This scheme is intended to provide file security for users and to allow infrequently used files to be removed from disc space. It is not intended to provide a dynamic extension of file space.

The Cambridge Ring is now in use to connect the three Service VAX machines, the RML 380Z floppy disc server and a print server in the user area. Physical links exist to Nuclear Physics, Engineering Science and the Computing Teaching Centre. Work on the project to produce CR82 Transport Service software is essentially complete, although some work is being done to integrate this software with X25 Transport Service software, and modifications are being made to the VAX device drivers to enable them to work with equipment now becoming available commercially which conforms to the CR82 standard. Plans to make the user interface software more widely available have been delayed considerably by lack of robustness in the existing release; a later version containing many error corrections is awaited from the originating site.

A terminal packet assembler/disassembler (PAD) has been installed and is connected to SERCNET. This has absorbed much of the previous use of the link to the Gandalf at the Rutherford Appleton Laboratory, although some Tektronix users cannot yet be transferred. ULCC’s time scales for offering an X29 service, which will be accessed via a PAD, have been delayed.

One of the VAX processors has been connected as a node on SERCNET, and a limited file transfer service is provided. It is usually necessary for Service staff to validate the means of connection to any new site and to give guidance to the user. The lack of robustness of the user-interface software (the same as that used for the Cambridge Ring) also makes it difficult to give general user access to it. The main use of this link at present is for incoming Lasercomp work, which is described in the section on Computing in the Arts.

A large amount of software has been produced to support the departmental budgeting schemes on both the VAX and the 2988.

2988

The ICL 2980 was closed down on September 17th 1982, and the service on the 2988 was fully operational by September 30th. In preparation for this, VME releases 7.01 and 7.02 had been installed in August and September, and release 535 of the front-end operating system (COS) was installed during the run-up of the 2988. Despite the wide range of problems encountered during the preparation and testing of the releases of VME, the actual change of machine showed up few problems with the operating system. However, the communications software had to be installed largely untested when it was found, apparently to ICL’s surprise, that COS530 would not support a 2988. This software caused a wide range of problems to most communications facilities, which were gradually cured by ICL or circumvented by the Service. However, most of these faults caused annoyance rather than loss of service.

The actual changeover was disrupted by severe problems with two of the four new FDS-640 disc drives; the service on the 2988 had to begin using some of the old EDS-200 discs and the last FDS-640 was not accepted by the Service until the middle of October.

Tidying the system and clearing residual problems took up a large amount of staff effort, which lasted until well into 1983. An underlying problem concerned with twin-processor operation became evident during March, and although it was circumvented it was only finally cured by the installation of release 7.05 of VME and consequent microcode amendments.

ICL have introduced a new release philosophy, intended to produce releases which are easier to install and which contain fewer errors. Release 7.05 was the first release to be made generally available under this new system. Although by no means perfect, it certainly proved better in both respects than any previous release. It was installed for user service in the last week of June.

Unattended night-time operation of the 2988 began at the beginning of January. Restructuring of the job queues has taken place to separate jobs using magnetic tapes, and therefore requiring operator intervention, from work that can be run at any time. It has been decided to offer no FREE queue service for magnetic tape jobs.

New versions of GIMMS, MDS(X), P-STAT, FACSIMILE, Ghost80, Algol68, BMDP81, QUILL, GENSTAT, the NAG Fortran Mark 9 library and the NAG graphical supplement Mark 1 were implemented.

VAX

During July 1982 the disc capacity on the VAX was increased by replacing the Digital 300 Mbyte drives by a mixture of 300 Mbyte drives and 675 Mbyte fixed discs supplied by Systems Industries. An extra disc controller was purchased for VAX3 in January. The systems had to be configured so that each of the two main user machines have available a 675 Mbyte disc for user filestore, and VAX3 has a 300 Mbyte disc. The other 300 Mbyte drive is used for backup of all of these discs, the backup being run automatically during the hours of unattended operation. Various switching arrangements can be used to reconfigure the systems in the event of disc or processor failure. The value of this was proved when a fault affected part of the 675 Mbyte disc on VAX1 in June, but the service was able to continue with a 300 Mbyte disc switched into VAX1 taking the place of the logical drive affected.

Version 3 of VMS was installed without major problem in October and two maintenance releases, 3.1 and 3.2, have been installed.

The Basic and Pascal compilers were purchased and put into service in October. New versions of GENSTAT, Spitbol, ECCE, Ghost, the NAG Fortran Mark 9 library and the first version of the NAG graphical supplement were implemented.

The card reader was withdrawn from service in October; a route is available to read cards into the VAX via the 2988.

[Top]

User Liaison

Advisory Service

After the experiments last year with the Advisory Service rota, a special Users’ Group meeting was held in February to discuss the Advisory Service. At this meeting it was made clear that the users would prefer to have more continuity of advice, offered by a smaller group of people. The Advisory Service rota has been modified so that one of two members of staff is on duty each morning from 1000 to 1300. Each occupies this slot for a week at a time, thus providing more continuity of advice for ongoing problems.

A new system of record keeping has also been introduced. All problems coming to the Advisory Service are logged and a weekly analysis of these is carried out.

Questionnaire

A questionnaire was circulated last summer to all users and heads of departments. After analysis of the replies a 54-page report was produced in April. The Service has to point out that the weight of this report is lessened by the fact that only about 25% of the recipients of the questionnaire responded.

Documentation

The following is a complete list of all the currently available User Guides. Those marked with an asterisk (*) were produced during the last quarter.

A1.1/3 Introduction to the Computing Service

A2.1/4 Rules of the Computing Service

A3.1/5 Service Facilities

A4.1/2 Glossary of Terms (2988)

A4.2/2 Glossary of VAX Terms*

A5.1/1 2988 Budgetting, Accounting and Scheduling

A5.2/1 VAX Budgetting, Accounting and Scheduling

B1.1/7 Oxford Macro Specifications

B2.1/10 Getting Started on the 2988

B2.2/1 Using the 2980

B3.1/2 Editing Files

B4.1/3 The Job Journal

C2.1/1 Running Algol68 Jobs

C3.1/1 Running Fortran Jobs

D1.1/2 Files and Filestore

D2.1/2 Magnetic Tape Usage on the 2980

E2.1/3 BMDP Biomedical Computer Programs - P series

E2.2/2 SPSS- Statistical Package for Social Sciences

E2.5/2 MDS(X)- Multidimensional Scaling Programs

E2.6/1 GLIM- Generalised Linear Interactive Modelling

E2.7/2 TSP- Time Series Processor

E3.1/3 GHOST Graphical Output Library

E3.2/2 SYMAP

E3.3/1 SYMVU

E3.4/1 GIMMS- General Purpose Geographical Processing System

E4.1/1 Getting Started on the Lasercomp*

E4.2/1 Famulus

F1.1/1 Using Remote Computers from the 2988

G1.1/1 Creating and Editing Files using ECCE

G1.2/1 Getting Started with ECCE*

G2.1/1 VAX/2980 File and Job Transfer

G2.2/1 Listing Files to Film

G2.3/1 The Oxford File Archive

G3.1/1 Spitbol

G4.1/1 OCP- Oxford Concordance Program

G4.2/1 QUILL

G4.3/1 P-STAT- Princeton Statistical Program

G4.4/1 GENSTAT- A General Statistical Program*

G5.1/1 NAG Library

M1.1/1 Algol68 Library

M2.1/1 ACSL*

M2.2/1 STRUCTURES*

M2.3/1 GPCP*

M2.4/1 FACSIMILE

M3.1/1 Calcomp-on-Ghost on the 2988

M3.2/1 Ghost80 Extensions for Dacoll M249 Extended Features Package*

M4.2/1 Standards for Magnetic Tape Transfer*

M4.3/1 Standard VME Tape Handling*

M5.1/1 Using the PAD*

V1.1/4 Getting Started on the VAX

V3.2/1 MINITAB- An Interactive Statistics Package

V3.3/1 TSP- Time Series Processor

V3.5/l GLIM- Generalised Linear Interactive Modelling

Courses

Introductory courses for new users were given regularly throughout the year. ‘Getting Started on the VAX’ was run 23 times during the year and 228 people attended. ‘Getting Started on the 2988’ was run 19 times and 130 people attended.

The Computing Service gave two open presentations during the year. The first, in October, introduced the computing facilities to new users. Despite widespread publicity the attendance was disappointing. The second presentation, held in March, described the use of graphics in local area and wide- area networks, and also the ICL PERQ system. Attendance was much better than at the previous presentation, and more than 60 people attended the practical demonstration on the PERQ.

A number of statistical courses and seminars were given throughout the year. A series of presentations was given in Michaelmas Term to users planning to use the computers for statistical analysis. They covered the packages SPSS, MINITAB, SIR and P-STAT. A similar set was given in Trinity Term, as well as a series of lectures on the facilities and use of the package GENSTAT. All these courses were well attended.

A number of one-day seminars were given during the year. Dr. Tindell of Physical Chemistry talked about the ICL Distributed Array Processor (DAP). A conversion course on Algol68 for Algol60 programmers was given by the Service. A one-day workshop on ‘The Effective Use of the Cray 1S at ULCC’ was given by a member of ULCC’s staff in December. A presentation about the new Cyber 205 computer being installed at UMRCC was given by a member of UMRCC’s staffing January.

Remote Services

This year has seen the start of conversion to the new system at UMRCC. The Amdahl 470/V7A front-end system was delivered in April and the CDC Cyber 205 rear-end computer was delivered in June. However, acceptance of both has been delayed. Meetings have been held with UMRCC staff to plan the communications routes, using modern protocols, which will be used. As yet implementation of these protocols at UMRCC is some way off.

Methods were set up to submit work to the Cray computer, firstly at Daresbury and then at ULCC.

The direct connection to Cambridge was discontinued in January and all work is now submitted via Metronet. This is available throughout the day.

Budgetting

The 19 departments or groups selected to be the first to be allocated on a group basis applied for their year’s allocations of computing resources at the beginning of November. The Allocations Committee approved these, with some modifications, and the scheme was implemented on December 6th. A further 10 groups have been approached to join the group allocation scheme for 1983-84.

[Top]

Computing Teaching Centre

Courses

This year regular courses have been given on Algol68, Interactive Basic, Fortran, Pascal, Graphics (microcomputer and mainframe), Micro processor Assembly Language Programming and Microprocessor Systems. Specialist courses have been given for Arts students, Engineering, Inorganic Chemistry, Geology, Geography, Graduates, Physics and Experimental Psychology. In addition DIY courses are available on a number of subjects on video cassette.

External Courses

A computer appreciation course was given to Local Authority Housing Managers. The Centre was hired by a local company to give Prolog courses on seven occasions, and by J. Walter Thompson, an advertising company.

Equipment

New colour CCTV equipment has been installed in the lecture room, replacing the 24 inch monochrome televisions. This was wired using the ‘single-wire PAL’ transmission system, but this has proved unsatisfactory for graphical work, and is currently being changed to the ‘four-wire RGBS’ system.

The new colour videographics system using U-MATIC/BETAMAX format has been installed in the TV studio. The production of colour videotapes is underway for use with the DIY workstations and lectures. New videotape players have been delivered to use the new tapes being produced.

The first RML network now has eight workstations, and a second RML network arrived in January. Both networks have, in addition to eight 480Z workstations, a 380Z network server, twin 8 inch floppy disc drives and a dot-matrix printer.

Other new equipment includes eight 4-colour Tandy printer/plotters for use with the RML networks, a bitpad digitiser and a joystick graphics control panel.

[Top]

Computing in the Arts

Courses

The Spitbol programming course has been given six times, and the courses on Text Analysis and Computer Methods in Historical Research once each.

The external advisor gave courses at a number of other universities, including a one-week course at Newcastle, and at the British Library. Mrs. Hockey lectured on a course on Computing and the Humanities at the University of Rome.

Oxford Concordance Program

During the year OCP has been released to a further 30 sites, of which four were in the United Kingdom, ten in the United States, three in West Germany, two in France, Belgium, Canada and New Zealand, and one in Israel, Singapore, Australia, Finland and the Netherlands. The users’ manual was reprinted in September; over 200 copies have been sold this year.

KDEM

The reliability of the KDEM has been very good. New releases of the software have improved the scanning capability. A variety of work, mostly for external users, has been done.

The five staff places provided under the Youth Opportunities Scheme have been filled for most of the year, and the trainees, as well as operating the KDEM, have been able to type in some texts that the KDEM could not handle.

Lasercomp

The workload on the Lasercomp has built up considerably during the year. Five books and several articles, pamphlets etc. have been completed and published. Copies of these are retained at OUCS.

A new version of LASERCHECK was produced; 37 sites now use this. An assessment of the feasibility of offering the pagination program LASERSET on the service is being made.

A system for submitting Lasercomp jobs via UMRCC and ULCC was set up early in the year, and in June a service was made available via PSS and SERCNET.

All the existing Latin fonts have been bought as software-sized characters (which enables OUCS to generate many different sizes from each character), and in addition Ehrhardt and Helvetica Medium and bold italic for all the Latin fonts have been bought.

A program for typesetting Hebrew is now available.

Two meetings of external users and three of internal users were held, and two issues of Lasercomp News were produced.

Archive

The number of texts in the Archive has continued to increase. A catalogue of entries was published in July and widely circulated .

[Top]

Performance and Reliability

2980 Service

The ICL 2980 service finished on September 17th 1982, having run since February 2nd 1978.

2988 Service

This service started on Thursday September 30th. After the initial problems with the FDS-640 fixed discs the hardware has performed well. Following the Christmas shutdown, processor problems caused some loss of time, and in March time was lost due to a problem concerned with twin-processor operation. The system was run with one processor during the week ending March 20th. This caused a large build-up of batch work, but the scheduling algorithms were able to maintain the interactive service without noticeable degradation. The problem was circumvented and with a recent release of the microcode has now been completely cured.

The new tape decks (GTS470) are a considerable improvement on the old, and this has resulted in a reduction of media failures affecting user jobs. All new tapes being issued are for use at 6250 bpi.

Unattended running was introduced at the start of January, and since then the system has run between midnight and 0800 without operator attention. Initially this resulted in some lost time for user service in the evening due to security copying of the filestore. This has now been avoided and copying is done overnight automatically. There have only been four occasions on which the system was unavailable in the morning and two of these incidents happened when all user work had been completed.

Overall the changeover to the ICL 2988 system has meant a great increase in system availability for users as well as improvements in other areas. The workload on the system has increased gradually since installation and there is still spare capacity available (particularly for batch work overnight).

VAX Service

The reliability of the VAX systems has been satisfactory although memory faults continue to affect the machines on occasion, resulting in periods of running on reduced store.

The introduction of unattended running on the ICL equipment, which has increased the number of operators available in the evening, has enabled batch jobs requiring tapes to be run for two shifts instead of only one.

As with the 2988, security copying of the filestore is now carried out automatically overnight.

The workload on the VAX systems continues to increase (particularly on VAX2) and a slight worsening of response time has been noticed. This has been largely as expected, and has resulted in some changes to the tuning of the systems.

VAX3 is now increasingly used for teaching purposes, both by the CTC and for courses run by OUCS.

Graphics Service

The Calcomp 1051 drum plotters continue to perform reliably. The change from biro to Pentel pens as the default has resulted in a reduction in the cost of providing this service.

The Calcomp 1670 microfilm system continues to suffer from repeated breakdowns as does the associated film processor. The 35 mm service has been particularly affected.

All these systems were converted to use nine-track tape decks in September; after a few problems these now perform reliably.

Lasercomp Service

The Lasercomp service has expanded over the past year but few serious problems have arisen. The laser was replaced in August and there remains a problem with setting to the extreme right-hand edge of the paper.

Microcomputer Service

During the year the RML 380Z provided for file transfer was connected to the Cambridge Ring. This has improved the facilities for transfer. An Apple II and a Cifer 2684 have also been installed and provide facilities for file transfer in a variety of floppy disc formats.

Remote Services

The MOD1 system has suffered a number of breakdowns during the year, resulting in interruptions to the transfer of jobs to and from the regional centres at London and Manchester. Because of the age of the equipment, these problems must be expected to continue until the advent of new protocols at the remote sites enables more modern hardware to be used.

Air Conditioning

During the week beginning March 28th both the 2988 and the VAX services were shut down for two days while the old chillers (installed in 1972) were replaced. These units had become very unreliable, particularly during hot weather. The new units should provide sufficient reserve capacity to ensure that service can continue during very hot weather. This year (for the first time in many years) there was no interruption or breakdown. A standby humidifier has been installed, which will enable the equipment to be kept running while maintenance is carried out on the main humidifier.

[Top]


Table 1. Machine Usage

Week Ending

2988

VAX

 

Interactive

Batch

Interactive

Batch

 

No. of Jobs

CPU Hours

No. of Jobs

CPU Hours

No. of Jobs

CPU Hours

No. of Jobs

CPU Hours

17/4/83

3459

15.31

3022

124.39

3005

27.75

1608

65.85

24/4/83

3780

15.99

3280

115.27

3981

39.49

2024

71.39

1/5/83

3396

15.05

3377

101.94

3901

39.29

2180

98.36

8/5/83

2659

12.14

2680

84.06

2962

34.89

1540

73.77

15/5/83

3202

14.00

2850

77.79

3603

33.74

1868

79.92

22/5/83

3529

18.62

2656

124.32

3643

44.22

1622

113.72

29/5/83

3933

18.27

3167

138.03

3796

47.51

1774

133.63

5/6/83

2594

12.24

2583

94.36

2620

31.05

1739

86.98

12/6/83

3345

13.34

3060

92.04

2625

37.98

1674

109.99

19/6/83

3257

12.05

3014

132.09

3149

24.17

1583

116.28

26/6/83

3235

13.05

3172

143.07

3140

39.94

1770

125.59

3/7/83

3637

13.78

3491

131.61

3732

45.97

2001

114.87

On the 2988 CPU hours are measured relative to one of the twin processors, each of which is roughly half the power of the single processor of the 2980.

[Top]

Table 2a. Usage by Subject Grouping on 2988 for 12-Week Period Ending 3/7/83

Group

No. of MAC Jobs

Cost of MAC Jobs

No. of

Batch Jobs

Cost of Batch Jobs

Total Cost

Four-Weekly Income

Filestore Usage (Mbyte)

Filestore Alloc-ation (Mbyte)

Paper Usage

Four-Weekly Paper Alloc

Chemistry

3303

21431

3272

9916

31462

34100

160

172

11694

7720

Physics

3785

32545

3105

9262

41929

27560

68

72

6949

10800

Oth Phys Sci

933

7668

777

2427

10208

44600

87

78

4507

7540

Mathematics

663

4152

491

2006

6196

9400

27

34

1114

2460

Biol &Ag Sci

4980

44987

5449

21060

67161

49200

297

312

23885

12200

Medical Sci

6377

49148

7731

20920

70354

63800

135

163

20505

21680

Social Sci

3478

31257

2263

7939

40341

32360

105

112

16155

9740

Humanities

2102

18908

1369

2320

21604

50000

102

139

8887

15460

Ext. Univ

181

3016

201

657

3694

5400

12

26

519

3380

Non-Univ

1811

11255

2519

4698

16164

23200

125

160

10341

5140

Service

5829

63725

3422

12464

72252

37660

278

296

19939

11920

Overheads

1237

10778

38220

44355

57286

219

26759

Totals

34679

298870

68819

138024

438651

377280

1615

1564

151254

108040

Total Cost = Cost of MAC Jobs + Cost of Batch Jobs + Cost of Spooling

Table 2b. Usage by Subject Grouping on VAX for 12-Week Period Ending 3/7/83

Group

No. of On-line Jobs

Cost of On-line Jobs

No. of

Batch Jobs

Cost of Batch Jobs

Total Cost

Four-Weekly Income

Filestore Usage (Mbyte)

Filestore Alloc-ation (Mbyte)

Paper Usage

Four-Weekly Paper Alloc

Chemistry

6032

114802

5261

25928

140714

106400

63

83

17309

7824

Physics

4613

73732

5518

18436

92169

65000

28

47

11212

6720

Oth Phys Sci

2285

33585

2852

16019

49607

73000

17

30

14271

5136

Mathematics

2940

68350

965

4831

73181

78500

25

36

7014

8304

Biol &Ag Sci

3514

54564

1185

7066

61631

88900

34

65

9903

11112

Medical Sci

2692

37969

1199

2033

40004

80660

31

56

11495

7104

Social Sci

4304

76928

2002

5289

82231

93000

50

78

35038

9648

Humanities

2758

52581

736

2181

54766

113000

31

66

16233

13176

Ext. Univ

11

104

1

44

148

3000

2

3

6

576

Non-Univ

2591

66192

629

3803

70003

76800

69

164

4596

2040

Service

5075

52330

413

940

53287

66000

87

86

12626

11064

Overheads

3799

45099

6118

18299

63408

82

9550

Totals

40614

676236

26897

104869

781149

844260

519

714

149253

82704

[Top]

Table 3. Remote Site Usage

For the year August 1982 to August 1983 Oxford has been allocated 7.0% of ULCC and 10% of UMRCC.

The %Units and %Jobs columns are respectively Oxford’s percentages of the total units used and of the jobs submitted by all users in the specified periods.

University of London Computer Centre

Period

Units

Jobs

%Units

%Jobs

11/4/83-8/5/83

7446

584

3.82

1.97

9/5/83-5/6/83

15262

530

7.51

1.76

6/6/83-3/7/83

23706

413

13.05

1.74

The CDC 7600 at ULCC produces approximately 500 units per hour.

University of Manchester Regional Computer Centre

Period

Units

Jobs

%Units

%Jobs

11/4/83-8/5/83

92709

1167

14.53

8.66

9/5/83-5/6/83

147960

1027

23.33

7.37

6/6/83-3/7/83

119327

915

19.66

7.46

The CDC 7600 at UMRCC produces approximately 1500 units per hour.

University of Cambridge Computing Service

Period

Units

Jobs

%Units

%Jobs

4/4/83- 1/5/83

4744

451

0.199

0.247

2/5/83-29/5/83

4229

425

0.165

0.215

30/5/83- 3/7/83

4997

318

0.153

0.141

The IBM 3081 at Cambridge produces approximately 9000 units per hour.

Queen Mary College

Period

Units

Jobs

%Units

%Jobs

11/4/83-8/5/83

95.25

192

0.225

1.266

9/5/83-5/6/83

87.23

109

0.241

1.362

6/6/83-3/7/83

110.14

59

0.286

0.686

The ICL 2980 and the Distributed Array Processor together produce approximately 100 units per hour.

[Top]

Table 4. VAX-2988 Link Usage

The link supports file transfer to and from the 2988 job submission from the VAX to the 2988, and transfer of files to be spooled via the 2988 to graphical devices and remote lineprinters.

Month

Into Vax

Into 2988

 

File Transfer

File Transfer

Spooler

Job Transfer

April

2888

1282

2826

625

May

3145

1334

2825

1305

June

3203

1703

2877

1068

[Top]

Table 5. Job Turnround on 2988

Production of these figures has been temporarily discontinued.

[Top]

Table 6. Reliability and Mean Time Between Failures

Reliability is the percentage of scheduled productive time (i.e. switched-on time not allocated to maintenance, development or backup) during which the machine is fault-free. Weighting factors are applied to equipment failures that allow the service to continue with impaired performance. Mean Time Between Failures (MTBF) is the mean time between incidents requiring a system reload to take place.

Week Ending

2988

VAX1

VAX2

 

Reliability (%)

MTBF (hrs)

Reliability (%)

MTBF (hrs)

Reliability (%)

MTBF (hrs)

17/4/83

99.15

110.90

100.00

¥

100.00

¥

24/4/83

100.00

¥

100.00

¥

100.00

¥

1/5/83

99.26

41.33

99.96

168.00

99.96

168.00

8/5/83

98.69

50.19

99.81

167.88

99.70

55.83

15/5/83

97.38

30.40

97.95

33.60

100.00

¥

22/5/83

98.73

40.50

99.64

56.00

99.70

84.00

29/5/83

99.72

124.30

100.00

¥

100.00

¥

5/6/83

99.33

33.25

99.75

84.00

99.78

168.00

12/6/83

98.51

61.39

84.88

24.00

100.00

¥

19/6/83

98.83

61.93

93.62

27.63

99.95

167.57

26/6/83

99.33

41.41

100.00

¥

99.09

82.94

3/7/83

99.93

¥

100.00

¥

100.00

¥

[Top]

Table 7. Response Time on 2988

This table shows the mean terminal concurrency at 15-minute intervals throughout the day, averaged over the 12-week period. The concurrency (the number of terminals logged in together) is given in figures and plotted. The right-hand column gives the average MAC response in the same time intervals. Again the figures are averaged over the 12-week period, and 5% of samples on either extreme are omitted to avoid periods when response was adversely affected by major hardware malfunction or was unusually good through low usage. Thus the figures given may be taken to represent the average response in ‘normal’ circumstances .

Time

Concurrency

MAC Response (seconds)

09:15

7.7

       *

2.0

                   *

09:30

14.6

              *

1.5

              *

09:45

18.8

                  *

1.4

             *

10:00

21.5

                     *

1.4

             *

10:15

22.2

                     *

1.6

               *

10:30

21.8

                     *

1.4

             *

10:45

20.2

                   *

1.4

             *

11:00

18.9

                  *

1.5

              *

11:15

20.7

                    *

1.4

             *

11:30

24.3

                       *

1.5

              *

11:45

26.2

                         *

1.5

              *

12:00

26.6

                          *

1.5

              *

12:15

24.9

                        *

1.4

             *

12:30

21.5

                     *

1.4

             *

12:45

17.7

                 *

1.3

            *

13:00

13.4

            *

1.2

           *

13:15

12.5

            *

1.1

          *

13:30

13.5

             *

1.0

         *

13:45

15.8

               *

1.0

         *

14:00

18.9

                  *

1.0

         *

14:15

22.4

                     *

1.3

            *

14:30

25.3

                        *

1.2

           *

14:45

27.4

                          *

1.3

            *

15:00

27.7

                           *

1.4

             *

15:15

28.1

                           *

1.4

             *

15:30

27.1

                          *

1.4

             *

15:45

25.0

                        *

1.4

             *

16:00

22.1

                     *

1.3

            *

16:15

21.3

                    *

1.2

           *

16:30

22.2

                     *

1.2

           *

16:45

22.0

                     *

1.2

           *

17:00

21.3

                    *

1.2

           *

17:15

20.4

                   *

1.1

          *

17:30

18.2

                 *

1.1

          *

17:45

15.7

               *

1.3

            *

18:00

13.5

             *

1.1

          *

18:15

11.2

          *

1.0

         *

18:30

9.1

        *

1.1

          *

18:45

7.2

      *

1.0

         *

19:00

6.0

     *

0.9

        *

19:15

5.4

    *

0.9

        *

19:30

5.0

    *

0.9

        *

19:45

4.9

    *

0.9

        *

20:00

4.7

    *

0.8

       *

20:15

4.9

    *

0.8

       *

20:30

5.3

    *

0.8

       *

20:45

5.6

     *

0.8

       *

21:00

5.6

     *

0.8

       *

21:15

5.5

     *

0.8

       *

21:30

5.0

    *

0.6

     *

21:45

5.0

    *

0.6

     *

22:00

4.9

    *

0.7

      *

[Top]

Table 8. Items Input to KDEM or Oxford Archive

Review of the Foreign Press 1939-45 (France) (Part of)

Park, D., Elements and Problems of Perception

Lent Sermons from Filius Matris Sermon Cycle

Rossetti, W.M., The Works of Dante Gabriel Rossetti

Chaucer, Canterbury Tales (Part of)

Dictionary of Geology

[Top]