Engineering Laboratories
FRANCE
Paris-Gambetta (1931-1993)
It is difficult to assign a founding date for the Gambetta laboratory. But
the first electronics engineers were hired in 1949. This lab developed electronic
calculators and computers. In the late 1950s, the hiring
busted when Compagnie des
Machines Bull started the development of Gamma 60. A few ex-SEA engineers joined the
company, but the majority were freshmen from French Grandes Ecoles, discovering digital
technology on the spot. In 1963-1964, many left either to the subsidiaries
of Univac or CDC
or to create the start-ups like SETI Pallas or CAE. The last pure French products
developed at Gambetta were the Gamma 10, the small Gamma 5 computer (ancestor of GE-58,
and the Gamma M-40.
Bull started to constitute a software team for the Gamma 60. It was enlarged in
1963 by the sales support team of the Service Technico Commercial.
After the GE merger, the Gambetta lab embarked on the short
lived Gamma 140 project. When the project was halted in 1967, many engineers, specially in
the software team, left the company to the burgeoning CII. Other move to some GE-400
tasks.
The separate work done with Claude Bouvier on small machines reached a milestone
with the announcement of GE-58 and, although Claude Bouvier and some co-workers quit
temporarily for CII, the small systems engineering closely integrated with marketing
remained
alive on that product line until the late 1970s.
A small team, originating essentially from M-40, was the
core of the future level 64 program.
The team went on working on advanced paper designs, while an embryonic software
team
attempted to design a modern operating system for the same hardware. The adoption of the
project by Honeywell triggers a new burst of hiring of logic designers and software
developers that culminated in 1972. For a few years, Gambetta did continuation engineering
for Level 64, taking over the parts of the development that have been done originally in
the Boston area.
Circa 1974, there was a plan to move eventually the Gambetta engineering lab to the
south suburbs of Evry, a reason for that plan is that floor area in XXe arrondissement
became expensive, because French government subsidized companies to move out of Paris, and
also because the personal administration desired to get some attrition from the massive
hiring. In fact, the CII-HB merger obsoleted that plan before any implementation.
After the CII merger, the Gambetta hardware team specialized in P7G processor design and system integration, while part of the GCOS7 software was distributed to Louveciennes. P7G was the last processor designed in Gambetta and hardware people were transferred to Les Clayes after the end of that project. It was also intended to move also software. To compromise with the resulting social unrest, the plan was amended to move progressively people to Louveciennes. Software people stayed in Gambetta until the last month of 1993.
Saint-Ouen
Louveciennes (1964-1992)
Louveciennes site was originally CII headquarters as well as marketing. It
was to be the software development site for CII Siris operating systems including its
development in data communications networks.
After the CII-HB merger, in 1976, the ex-CII software team
was split in parts, one dealing with Networking, another in GCOS7 software and
Siris emulation and a smaller team worked on GCOS6 complements.
In 1983, started the implementation of a transfer plan of GCOS7 software
from
Paris-Gambetta to Louveciennes. It allowed to better merge teams working on the same
operating system and to integrate GCOS7 as a DSA system. A planned side effect of the plan
was to trigger a significant attrition in the 1970 hired Gambetta team.
Circa 1990, Bull envisioned to clear completely the Louveciennes site and to move all engineering to Les Clayes, where it is still there. In fact, the Louveciennes campus was reactivated in 1993 when Descarpentries decided to vacate tour Bull. At that time, the only engineers staying in Louveciennes were the Bull-CP8 people.
SEMS had another engineering site in Louveciennes-Village, staffed with some ex-CII engineers who worked on projects not included in CII-HB merger. When SEMS became part of Groupe Bull in 1982, the engineering activities were moved to Grenoble. Some engineers left with François Michel (Bull-SEMS president) to found Copernique.
Les Clayes sous Bois (1965-current)
Les Clayes was the main hardware laboratory of CII.
In 1977, when Unidata X-systems were discontinued, the bulk
of ex-CII logic designers embark on the P7G development where they were assigned mainly
the IOC and the channels.
During DPS-7 development, Les Clayes technologists, not very happy with the choice
of Honeywell CML as the company bipolar technology, started to push the development of the
MOS technology. Their intent of developing a experimental NMOS wafer production line was
cancelled by management in 1980, but the design of a VLSI DPS-7 (code named Archer) goes
on. And eventually this program succeeded in delivering DPS-7000 (Ares) in 1986. Logic
designers of Ares were ex-CII engineers who got reinforced in 1982 by ex-Honeywell-Bull
engineers who were transferred from the Gambetta hardware laboratory that was closed down.
Massy
Massy was the site of engineering of CGE-Alcatel group, Transac -Alcatel
was a part of. It joined Groupe Bull in 1983
Angers
An engineering hardware laboratory was created within Angers plant in
~1978, when the plant began investing for the DPS-7 micropackaging technology. Under
Christian Joly, the Saint-Ouen prototype lab was transferred
to Angers with almost all its
engineers. The proximity of engineers and manufacturing was one of the factor of the
successful introduction of that technology revolution. It helped also to clear the
technical problems that came later to plague DPS-7 production, particularly the discovery
of electro migration when copper was substituted to gold in the micropacks substrates.
Christian Joly had not been very happy by the added complexity and the cost
evolution of the P7G program: after a bootleg phase, he succeeded to make approved a
cost-reduced model, based on the Level-64 hardware architecture. The P7G-CCR (Taurus) was
completely designed in Angers and Chris Joly succeeded
to have it in production several
months before the bigger
P7G. Angers became close to become a systems house, competing with
Les Clayes. In 1982, Angers started to redesign a version of P7G, making a repackaged
version of the processor and extending the system to a full quad multiprocessor, that was
the Lyra system.
Christian Joly was then promoted to head hardware engineering in Paris, and the
micropackaging technology began to yield to emerging CMOS and DPS-7 was to be succeeded by
Les Clayes designed DPS-7000.
The Angers lab began to phase out in 1984 (engineers hired in the 1950s started to
retire and were not replaced, others moved to manufacturing) and was dissolved ~1990.
Belfort
Grenoble-Echirolles
ITALY
Pregnana-Milanese
USA
Phoenix (195x-current)
Cambridge
Billerica
Framingham (195x-1973)
Waltham
other
Hemel-Hempstead (UK)
Bangalore (India)
NEC
Fuchu
NEC main computer laboratory is located in Fuchu City, in the western
Tokyo suburbs. This lab is not specialized in computers, but performs also
telecommunications and other work. Instead of being entirely organized by product line,
all hardware engineers for main frame report to a single organization, all technology and
packaging work being common to ACOS and DIPS product lines. Software people were organized
by operating system, although people were moved between those teams when a product line
was growing or was being phased out. For instance, UNIX team was largely built from ACOS
teams.
Tokyo
ACOS6 software team came originally from Toshiba and, probably because
engineers were not hired for the Fuchu area, they stayed in Tokyo, at or near the NEC
headquarters.
Revision : 29 juin 2001