original in French, wrtten by Claude LEBRAS and Jacqueline VIDAL
adapted to English by Jean BELLEC
1- Gamma 140 Project History
Initiation (end of 1964-early 1965)
The 1964 agreements for the take-over of Compagnie des Machine Bull by
General Electric lead to the creation of two GE subsidiaries:
- BGE, Bull General Electric, that controls the sales network and distribute the GE computer product lines
- SIBGE, Société Industrielle Bull General Electric, where French interests have 51% of shares, that regroups the ex-Bull activities in R&D and manufacturing.
Around the same time, General Electric has bought the Computer division of Olivetti in Italy and has created the GEIS company to sell GE computer product lines and develop new products.
GE coordinates all computer R&D activities in USA and Europe through a Central Product Planning that decide the projects to initiate and the allocation of budgets to the technical operations. However, as the money is charged to the local operations (US, Italy, France), those operations continue to have some autonomy.
It should be noted that the long maturation of the GE-Bull merger has affected the personnel morale specially after the first lay-offs occurring in Bull in 1964.
Bull-GE Internal organization
Instead of an ad-hoc organization for R&D, a centralized Software division, under Jean-Paul BOSS, regroups the development of all current projects, including GE-400 and Gamma M-40.
GE Central Product Planning demanded to its European subsidiaries to launch the development of two compatible systems able to upgrade the Gamma 10 (card computer) and to take the place of Gamma 30 (alias RCA 301).
A franco-italian work group was constituted of Gérard COTTET (architect), Max de FERRAN, Michel KOPP (marketing), André MAITRE and Jacqueline VIDAL , for Bull; and of Simone FUBINI (architect), Marisa BELLISARIO and Mario MORPURGO for Olivetti-General Electric.
The Italians propose to use the ELEA 4004 as a base, proposal caught
reluctantly by the French. The workgroup produce the specifications of the two systems,
code-named Alpha and Beta that were presented to the management (Mr. Fancher for Bull-GE)
at the end of 1Q 1965. The Alpha project allocated to Olivetti-GE will eventually become
the successful GE-115 (sometimes called Gamma 115 in France).
Bull-GE got , as expected, the responsibility of Beta that will take the development name of Gamma 140.
Gérard COTTET was not satisfied by the specification of the Beta project
and the responsibility for the system architecture was transferred to Georges LEPICARD
(who had similar functions for the Gamma M40.
François SALLE and his team join the Software division from Sales support. François SALLE took in charge the development of the Operating System.
During 1966, was developed a Tape Operating System and a strategy to carry on Gamma 30 applications. A Disc Operating System (with Claude CARRE, Claude CHEMLA, Paul NICOURD, Dave SLOSBERG and Henri VERDIER) , using still to be developed GE discs, was specified in 4Q 1966.
Hardware used a "mini-module" technology (discrete components assembled in functional building blocks) and prototypes were available by 3Q 1966.
In September 1966, at SICOB trade fair, Bull-GE made a public introduction of Gamma 140. Reception of this announcement was pretty good and many orders were booked in three months, essentially in France, but also in other European countries.
However, the development was not without its lot of problems. Tapes were not very reliable, and software was late.
In December 1966, GE Central Planning face that Alpha and Beta have
somewhat diverged from the initial objectives and were only remotely compatibles. In
addition the two projects had derived upwards: the Gamma 140 was , including its first
batch of orders, cannibalizing the GE-400 market segment.
The technology, although somewhat more advanced as the contemporary GE-400 and GE-600 was still menaced of fast obsolescence by the dawn of integrated circuits.
So, GE terminates the 140 project
Fate of the G140 team (1Q 1967)
Late in 1966, the Compagnie Industrielle pour l'Informatique, née from the French Government Plan calcul started a project , the Iris 50, targeting approximately the same market segment as the Bull-GE G140. It started recruiting en masse for that project, specially software engineers. Many G140 software engineers will follow François SALLE, to become the core of SIRIS3 developers.
Tesla revival (March 1967-3Q 1968)
The development files of the G140 had been closed with care . In March 1967, contacts were taken with the Czechoslovak company TESLA for selling them the manufacturing license of the G140.
After the contract was signed , the development of software started again in Paris to fulfill the contract. In September 1967 the TESLA 200 was presented at the Brno fair and the first production models (15) of the TESLA 200 were produced by the Bull-GE Angers plant
The transfer of technology, including the DOS, was initiated between Angers and Prague. In summer 1968, the Russian tanks disturbed somewhat that cooperation. But Tesla produced around one hundred T200 systems between Sep 1968 and March 1969, and had some exported in USSR. The Tesla 200 faced the competition of Comecon sponsored Ryad computer product line, an imitation of the IBM S/360.
2- G140 Brief Technical Description
The G140 is a single processor system with a core memory, byte-addressable, having a maximum of 64KBytes.
The processor uses an asynchronous logic (without a centralized clock). That leads to prototype trouble-shooting difficulties, component failures being difficult to distinguish from design errors. Contrarily to other Bull projects, the processor is not micro-programmed.
The logic was implemented with discrete transistors, resistors and condensers packaged in "mini-modules" used as manufacturing units and spare parts.
Instructions are fixed length referencing two addresses. Indexation and indirect addressing is supported. Character data are 8-bits EBCDIC. An optional floating point unit was proposed.
The I/O operations use the concept of " channel programs"
running simultaneously with central processor. Two types of channels are available one for
tape (and discs) controllers, one for unit records (card equipment, paper tape and
printer). Note that backwards reading was allowed on tapes.
The disks were to be GE-15 removable discs. The card and printer devices were Bull produced.
The Tape Operating System main functions were:
- system initialization
- interrupt handling
- operator interface
- program libraries handling
- Job Control language (control cards) interpreting and job management.
Tape files use IBM standard labels. Disc files, handled through a VTOC on each disc store data in sequential, partitioned, index sequential and random, inspired by IBM DOS/360.
Programming languages are Assembler (used for software development), COBOL (with a compiler written in Cobol) and Fortran.
Revision : 19 février 2002.