|
Development Planning for Industrial
Computer Section
by
K. R. GEISER
Manager - Engineering
Industrial Computer Section
The development planning for the Industrial Computer Section can best
be covered by considering it in three separate parts. First, the
engineering program itself; secondly, the dollars involved in this
program; and thirdly, the people to san the program and their competence.
In outlining for you the engineering program, I will attempt to do as
Mr. Oldfield has done, And that is to limit my remarks, as much as
possible, to that portion of the program that is concerned primarily with
digital data processing equipment. It is somewhat difficult to partition
the program in a manner to isolate items that are of interest solely to
this area; so I beg your forgiveness if you detect some work, wherein it
will be readily apparent to You, that it also has a major effect on other
areas of computing activity.
Our planning for this engineering program dates back a little over a
year, at which time, a study team was organized and chartered to survey
the state of computer technology. This study team was active for some
three months and during this time they consulted with all of the
organizational components within the General Electric Company who were
concerned with the design and construction of either analog or digital
computers; and in addition, they consulted with some external groups
having similar interests. Also, they contacted approximately half of the
computer user components of the General Electric Company. The purpose of
these consultations was to determine the limitations of presently
available equipments, the limitations of the components or internal parts
that made up the computer; to the end of determining what both the
designers and the users of computers thought Deeded to be done next. The
results of this work were compiled under the heading "Computer
Technology Report", which was presented to Mr. Linder who in turn
presented it to the Advisory Council. It is available as TIS report number
R55GL167.
The heart of this "Computer Technology Report" had to do
with:
a. A survey of the technical computer problems.
b. Proposed solutions to these problems.
c. An analysis of the "areas" of benefit.
d. An analysis of categories of technical work required. e. An estimate
of the cost.
37.
|
| Briefly, some 29 technical problems were
pointed out, of which 14 appeared to have an
extremely high priority. Specifically, one problem had to do with input
equipment for digital computers or data processing equipment, two problems
were concerned with the arithmetic and control circuitry within this type
of computer, two problems were in the interest of lower cost and higher
capacity storage devices or "memories" for data processing
machines, one, of course, was concerned with a higher speed output
printer, six problems were concerned with the availability of complete
machines, the application of complete machines, and the improved
reliability of complete machines. The remaining two problems were in the
analog field.
There are, of course, numerous ways of classifying or categorizing the
se problems, such as those that are critical components, those that are
Advanced Techniques, and those that are product
development; but for your purpose, perhaps an even better way is to know
that six of the solutions would be in the nature of a General Electric
Company contribution to the computer art, two would result in lower cost
-of over-all computer and data processing systems, two are in the interest
of improved reliability, four would serve to extend the scope and
capabilities of data processing systems, and two were strictly advanced
engineering aimed directly at product development.
At that time, tentative solutions were proposed to most of these
problems and in the remaining cases approaches were outlined which would
lead to a major decision point concerning the future direction of the
problem.
Also, at the time of this study, an analysis was made to be sure that
this engineering program was not too heavily slanted toward any one group
or area of the Company.
This analysis, concerning the "areas that would benefit" from
this work, indicated that all areas of the Company benefited somewhat, but
the category entitled "User-Data Processing" benefited most
heavily.
Using as a backbone this "Computer Technology Report" our
present engineering program has largely constructed, and today we are
happy to report that of these 14 most important problems, we have work
under way that will directly influence 10 of them, and perhaps a few words
concerning some of these would be in order.
Work on the improved input device for data processing machines,
specifically character reading devices, is active at two locations. Work
on improved logical components and improved circuitry is active in the
Electronics Laboratory. Work on improved high density random access
storage device is active in the General Engineering Laboratory and in the
Microwave Laboratory. Work on a high-speed printer continues around
General Electric technique commonly referred to as "Ferromagnetography."
To gain field experience with this technique, a complete gray-scale
picture drawing facsimile recorder has been built and developed and is now
operating
38.
|
| in the U. S. Navy Department. At the present
time, two printers of the character printing type have Just been completed
which are being installed as part of prototype computers to be delivered
to the Signal Corps for field tests. In the analog field, we have work
active concerning two computers; first, one specifically designed for
production scheduling and called the "Productron" and secondly,
a computer for linear programming. Also, we bave two activities under way
at the Advanced Electronics Center in Ithaca which should result in a
transistorized operational amplifier and a transistorized multiplier for
use in analog computers.
Now I have left until the end the most important one of all, as far as
data processing machines are concerned, and that has to
do with the ERMA work for the Bank of America. As Mr. Trotter told you,
ERMA is a form of data processing machine, and our activity here will do
more than any other one thing to provide us with new components,
techniques and experience that will directly benefit your interest in data
processing equipments. We are not simply copying ERMA as it was designed
by Stanford Research Institute; we are starting right from scratch and are
thoroughly analyzing both the basic philosophy and the logic design
underlying this computer. We are not sure that the cascaded logic that
they have used is the best for us in the long run. Following this, we have
told the customer that we will transistorize this machine in
so far as possible. My guess is this may be approximately 75 per cent.
Within this 75 per cent will come the development and availability of new
transistorized control and arithmetic components that we will be able to
use in machines for you. The transistorized components should provide
better reliability which means less service cost to you. Also, they should
reduce the power drain, which means less power supply, which means less
first cost to you. ERMA has in it many of the elements that you will have
in your data processing machines. That is large memory, tape storage,
arithmetic units, input and output devices, etc. There is no doubt in my
mind that the obtaining of this contract is one of the best things that
could have happened to us, as far as you are concerned.
Now the second part that bears on our development planning is dollars.
I will cover this very briefly. In light of our market predictions, we are
estimating our advanced development requirements as follows:
1956 $750,000
1957 1,150,000
1958 1,200,000
1959 1,400,000
1960 1,600,000
39·
|
| An analysis of our 1956 expenditures
indicate that they in the correct order of
magnitude:
Self-sponsored expenditures in our own advance
development
organization
$300,000
Directly sponsored Labs. Dept.
Programs
127,000
Supporting Labs Dept. Programs from assessed
funds
130,000
Advance developments sponsored through ERMA 150,000 Military sponsored
advance developments 160,000
which totals to be $867,000.
In 1957, we plan to increase our self-sponsored expenditures by at
least $100,000 and beginning in 1958 we will increase our self-sponsored
expenditures by at least an additional $50,000.
The important point here is that we are establishing the magnitude of
our advanced development work at a level that should achieve us a
respectable place in the industry. You remember that the Electronics
Business Study showed that to achieve a certain standing in any particular
segment of the Electronics Industry, the continuing expenditure for
advanced development could be related to over-all
industry sales. To achieve third or fourth position in the industry, it is
necessary to invest .15 per cent of over-all industry sales of three years
hence.
The third and last part in this development planning has to do with
people, and here there a couple of points. First, I would like to make it
clear that the ERMA contract has not or will not require an exorbitant
amount of manpower to the point that it will dilute our other activities.
The ERMA project has been set up in Palo Alto as a separate subsection and
is recruiting much of its own manpower in open competition with the other
subsections. It, of course, has the West Coast market to draw on and
therefore is not serious competition to the point of hurting the
Schenectady and Syracuse recruiting. Some real quick figures will tell you
the magnitude of the ERMA manpower requirement. The engineering bill on
the ERMA project is roughly $ 2 1/2 million, of which we will "farm
out" approximately $1 million worth. The Stanford Research Institute
are easily capable of taking care of this much of the job. This leaves $1
1/2 million worth of work to be done in the General Electric Company. Our
time schedule tells us that this effort must be applied within the next
two and one-half years; so $1 1/2 million divided by two and one-half
years is some $600,000 a year. Figuring that the average engineer costs
the job approximately $35,000 per year, this of course includes
engineering assistants, draftsmen, etc., it says that the average number
of engineers on the ERMA project will be about 17. The second point is
that we are not yet experiencing too much difficulty in recruiting
engineers. We are undoubtedly riding on the crest of at least two waves.
First, the announcement of the establishment of a new section always
brings a wave of applicants and we are seeing some
of this; secondly, the obtaining of the ERMA project indicated to those
who were skeptical that we were really serious about getting into this
business, and this has brought other
40.
|
| people who have been associated with other
computer projects. These two influences, plus the fact that there were in
the General Electric Company a few computer people who were holding their
breath waiting for the establishment of this section, has resulted in the
availability of a sufficient number of the applicants to enable us to get
a fair start in building up our organization. These people are not all
inexperienced. We have been able to bring into our new activity all of
those desirable, experienced people that were in the General Engineering
Laboratory, with the single exception of one man.
Finally, I would like to repeat what Mr. Oldfield said; that is, that
we have no axe to grind. We do not have a standard product on the shelf
that we will try to apply to - your problem. I have
always contended that the data processing requirements of the General
Electric Company are of sufficient magnitude that you can write your own
specifications and get what you want. You certainly don't have to take
something that was designed to fit a whole gamut of possible applications.
We are interested in working with you to the end of providing a proper
impedance match between your application and the data processing machine.
That is, I think, that our computer design engineers
working closely with your procedures people can, by each gaining a proper
understanding of the other's interests, arrive at that optimum point where
we have an economical data processing machine that adequately handles the
important parameters of your business.
41.
|
| BACK
TO INDEX OF SPEECHES |
|
