1983
45. CRITIQUE OF 1983
Any audit of 1983 must decide whether the project had a reasonable
return on the man years of effort agreed by Central Computing
Committee for the year 1983/4. Due to lack of staff the actual
manpower available to the project was 8 man years.
The achievements in 1983 were:
-
Hardware: another 40 systems were obtained and about 60
systems were installed in user environments.
-
PNX: a large proportion of the 139 systems were upgraded
from POS to PNX during the year.
-
Maintenance: first line maintenance was provided on all
systems during the year.
-
FORTRAN: about 3 man months of effort was put into the
FORTRAN software developments with ICL.
-
ACCENT/UNIX Assessments: about 3 man months were put into assessing the
bottleneck on ACCENT/UNIX.
-
Assessment: nearly one man year of effort went into the
assessment of alternative single user systems.
-
Utilities: porting of UNIX utilities continued in the first quarter
of 1983 at the level of about 2 people.
-
SPY: a major software development with two people on average
involved resulting in a high quality text editor for the PERQ which was
significantly superior to editors available from any competitors.
-
GKS: a significant amount of manpower and expertise was put into the
GKS implementation which would result in PERQ having the first GKS implementation
on a Single User System.
-
AI Languages: contracts were placed which resulted in three AI
languages becoming available albeit at reduced performance levels.
and functionality.
-
Communications: Wide Area Network communications were established.
The software was developed at York and tested at
RAL.
-
User Support: Three large user meetings took place in 1983, a
Newsletter was established with the Computer Board and a UNIX Support Desk
established.
-
Print Server: a letter quality print server was established and there
was significant progress in establishing further facilities allowing mixed text and
graphics with previewing on the PERQ.
It would be difficult to argue that SERC and the SERC Boards did not receive good
value for money for the effort put into the project in 1983.
Everybody worked extremely hard and there was considerable progress during the year.
Progress would have been better if more effort had been put in but that had been made
clear to the funding sources
including the Boards. It was their decision that progress should be limited by
cutting the manpower available to the project.
Assessing ICL's performance during the year:
-
Hardware: ICL designed a new PERQ which proved quieter, more reliable,
cheaper and higher performance during the year. The PERQ2 mid-life booster was a
reasonable advance over the PERQ1.
-
Future Developments: ICL proved slow in defining the successor to the
PERQ2. It was complicated due to the relationship with Three Rivers.
The decision to not go for a colour system early on was, in retrospect, a
grave mistake. The timescales for PERQ3 turned out to be about right looking at the
competition.
-
UNIX: the decision not to implement a true virtual memory system
early on was a mistake which haunted ICL throughout the year.
-
Compilers: neither FORTRAN nor PASCAL were good systems. The need to
keep compatibility with ICL's other ranges certainly caused some problems during the
year.
Assessing SERC's position during the year:
-
Confidence in Programme: all the Boards, but particularly the Science
Board, showed a lack of confidence in the programme which seriously disrupted its
progress. Criticism by the Boards meant that SERC management lost confidence in the
project.
Consequently, just at the time when real progress could have been made,
the lack of purchasing and manpower seriously disrupted progress.
A large number of users suffered unnecessarily.
-
Funding: for a further year, inadequate funding for the project was
provided and the little that was provided was given with a great deal of bad grace.
The year 1983 did show significant progress. By the end of the year, most users
could make significant use of the systems even if performance was less than
anticipated. The environment available by the end of 1983 was as good as or
superior to its competitors.
It is difficult to see what criticisms could be made of SERC staff concerning the
work done in 1983.