The Museum of Computing Machinery exhibits the most significant pieces of a very rich collection of machines linked to the history of Computer Science, unique in Italy and internationally important for its completeness and significance of the pieces.
The exhibition ranges from arithmometers of the nineteenth century to the great computers of the 1950s and 1960s, up to the most important products in the history of Personal Computers.
The Museum preserves unique examples such as:
– the Pisan Electronic Calculator (1961), around which the first Italian school of computer science was formed;
– the CINAC of the Institute of Calculus Applications in Rome (1964);
– iconic machines such as the Olivetti ELEA 6001 (1961), whose design was curated by Ettore Sottsass;
– the Cray X-MP (1982), on which the first Pixar short animated films were made.
You can take a fascinating journey into the history of a technology that is so pervasive today. The technology is so embedded in our daily lives that it is difficult for most people to think of information technology as a scientific discipline, which requires something different from installing applications on a smartphone or working on a laptop. The machines kept in the Museum, observed in operation or explained in detail in their behaviour, allow us to understand what are the principles that still guide modern tools today.
By wandering freely between the showcases or following a guided tour, visitors can learn the stories of the exhibits: the protagonists who made them, the social context in which they moved, the changes they brought about in the way they first understood reality and then manipulated reality. It will be possible to see the revolutions that led to our way of life.
In the section dedicated to calculators there are five subsections according to the type of machine. The sections include types of calculators that cover about five centuries of history: they range from compasses, electrical machines, mechanical machines and, finally, mini and personal computers to pocket calculators.
The section of calculators that need manual action to operate includes: rack and pinion adders, arithmometers, rulers, extended or reduced keyboard adders, drum machines and much more.
This section includes calculators that can do operations and write results just at the touch of a button, thanks to the presence of electrical circuits inside them. An example is the Friden SRW from 1952: the first calculator capable of extracting the square root with the press of a single key.
Mini and Personal Computer
Among them are exhibited machines that use electronic components to work: the Anita of 1963, the Olivetti Program 101 of 1965 (considered by many the first personal computer in the world), the first laptops of the 70s, the Macintosh of the 80s, the IBM compatible, the Commodore, the Spectrum and more.
In 1972 was produced the HP 35. It was the first scientific pocket calculator capable of calculating logarithms and trigonometric function values. This calculator is present in the Museum’s collection together with the Canon Palmtronic 85, the Texas Instrument 35 or the famous TI 57 (it was a great success thanks to its ease of programming).
The Pisan Electronic Calculator (CEP) is the showpiece of this section: one of the first large calculators built entirely in Italy. It started working at the end of the 50s at the University of Pisa, where it was built. Other important calculators ranging from the 50s (Gamma3 Bull) to the 90s (the Cray-research T90) complete this remarkable section.
Pisan Electronic Calculator
Origin: University of Pisa
The Pisan Electronic Calculator, also known as C.E.P. (Calcolatrice Elettronica Pisana), has been on display since 2002 in a dedicated room at the Calculation Tools Museum.
The C.E.P. represents for the Museum the most prestigious piece, at least in the category of large calculators, because it was the first calculator made in Italy (thanks to the commitment of the University of Pisa).
Gamma 3 BULL
Origin: CAED (Data Processing Acquisition Centre) – Mestre
Compagnie des Machines Bull’s Gamma 3 electronic calculator entered the market in March 1953 at Cré dit Lyonnais in Saint-Etiene (France). It had very similar functions to those of the IBM 604.
The series of Gamma calculators, often referred to as Gxx, began with the 1951 Gamma 2. This series was developed as a complement to Bull’s 150 Series punch card readers, but later these computers became real computers that used card readers as peripherals.
The commands (instructions) were stored in a connection panel that could contain a maximum of 64 commands called program steps. The clock of the machine was synchronized with the frequency of the electromechanical equipment’s motor (usually a tabulator or a punch card reader) connected by means of large parallel cables, called buoys, at 48 bits. The processor used about 400 vacuum tubes and germanium diodes which helped to improve reliability compared to a fully-valve machine.
The processor clock is at 281 kHz. Loading a 16-bit instruction takes 520 ms and its execution time ranges from 0.6 ms to 10 ms. The main memory consisted of 15 scrolling registers (each containing a word of 12 decimal places, i.e., 48 bits) with magnetostrictive technology. The basic register clock time (time required to move one bit inside the register) was 172 ms. Two of the registers acted as accumulators for fixed-point decimal operations. These models followed: Gamma 3A, Gamma 3M, Gamma 3B, Gamma AET, Gamma AET-ACIE, Gamma ET.
Calculator of the National Institute for Calculation Applications
Origin: CNR of Rome
In 1955 the National Institute for Calculation Applications of the C.N.R. in Rome bought a Mark I calculator (marketed in Europe in 1951) from the English company Ferranti. The calculator was renamed FINAC and remained in operation until 1966 when the Institute acquired an Elea 9104. In order not to lose the work and experience acquired by the Institute staff with the first computer, a FINAC hardware-software simulator was built and interfaced with the Elea 9104. It was given the name CINAC (Calcolatore Istituto Nazionale per le Applicazioni del Calcolo). CINAC therefore consisted of Olivetti Elea 9104 and the FINAC simulator.
A series of teletypewriters, printers and bandwidth readers were used to prepare and read programs suitable for both hardware platforms. The FINAC made use of a circular hole band while the Elea 9104 made use of a band with several holes and a different shape (square). FINAC made sounds when it was running and the programmers got used to interpreting them and were able to tell whether everything was going well or not. In order to do the same with the new computer, a loudspeaker with a suitable circuit was mounted on a support of the Elea’s central unit so that the FINAC sounds could be reproduced.
The Elea 9104 was never programmed with the native Olivetti code but worked in emulation, i.e. it was always programmed with the FINAC code.
The CINAC was decommissioned on June 30, 1970.
Origin: CAED (Data Processing Acquisition Centre) – Mestre
The Elea 6001 Processing System was built in 1961 by Olivetti in its basic composition and is composed as follows:
– central unit containing the magnetic core memory of 10,000 positions and the sequence logic matrix with 256 positions;
– command and control table;
– typewriter with built-in band perforator.
The basic composition is expandable by means of:
– additional memory modules from 10,000-20,000 positions up to a total capacity of 100,000 positions;
– directly connected input and output units;
– fast band perforator;
– magnetic tape units up to 8 units in total;
– in-line introduction and extraction units up to 7 units in total, 4 of which are input (card readers, band readers) and 3 output (card punchers, printers).
The Elea 6001 system, depending on the connected units, assumes one of the following two compositions:
– 6001/S for technical-scientific applications;
– 6001/C for both technical-scientific and business applications.
Origin: IEI (currently ISTI of CNR of Pisa)
The TAU2-TAUMUS system is a complex hardware-software that allowed the storage, the composition, the reworking and the execution of music tracks in real time.
This system was built at the C.N.R. – I.E.I. of Pisa from 1973 to 1975 where it worked until 1987. Then it was donated to the L. Cherubini Conservatory of Florence and then returned to Pisa to the Calculation Tools Museum.
The TAUMUS program resided on an IBM370/168 at the CNUCE Computer Center in via Santa Maria 36 while the TAU2 audio terminal was at the I.E.I., at the time in via Santa Maria 46.
The computer and the audio terminal were connected by a parallel cable. It was possible to access the IBM computer either from the TAU2 room using a serial cable or from another location using a modem-data line on the SIP network. The output of the TAU2 could be listened to in the room itself or remotely through the SIP network phonic line.
CRAY / XMP – YMP2E – T90
X-MP ( CRAY – Research Italia through the ENEL Research Centre of Pisa – Mestre ).
YMP2E and T90 ( ENEL Research Centre of Pisa ).
In 1961 Seymour Cray designed the fastest computer in the world at that time, the first supercomputer, the Control Data Corporation’s CDC 6600 model, of which he was one of the founders in 1957. Later he also designed the CDC 7600.
In 1972 S. Cray left the CDC and founded the Cray Research Corporation in Chippewa Falls, Wisconsin. In 1976 the CRAY-1 supercomputer was announced (image 04 on page:
http://www.fondazionegalileogalilei.it/museo/collezioni/grandi_calcolatori/schede_grandi_calc/cray.html). It was 10 times more powerful than the CDC 7600. It was first installed in the Los Alamos National Laboratory for $8.8 million.
The CRAY-1 had a power of 160 megaflops and could perform 160 million floating point operations per second. It also had a main memory of 8 megabytes (1 million words of 8 bytes).
In 1982 the company developed the first multiprocessor supercomputer: the CRAY X-MP (image 01).
In 1985 was released the CRAY-2 (image 05), 10 times more powerful than the CRAY-1.
In 1988 Cray Research introduced the CRAY Y-MP supercomputer, the first computer in the world to work with many gigaflops applications.
The Y-MP2E model (image 02) was the first CRAY supercomputer to have air cooling.
In the early 1990s the CRAY C90 model (image 06) was followed with 16 gigaflops, which used 16 x 1 gigaflops processors and a central memory of 256 million words.
The company also produced its first minisupercomputer, the CRAY XMS system, followed by the CRAY Y-MP EL series (image 07), and CRAY J90 (image 08).
In 1993 the CRAY T3D (image 09) launched on the market, the first CRAY MPP (Massively Parallel Processing) which soon became the market leader. The T3E and T3E-1200E models followed.
In 1994 the first supercomputer in the wireless world was announced:
the CRAY T90 (image 03).
APE – Array Processor Experiment
Origin: INFN ( Pisa section )
In 1984 the National Institute of Nuclear Physics (INFN) started the APE program, created to solve complicated problems of theoretical physics of elementary particles.
It was necessary to make a computer suitable for the study of quark interactions in order to explain the structure and behavior of protons and neutrons.
The project was coordinated by INFN President Nicola Cabibbo and Giorgio Parisi. In 1987 this research resulted in the creation of the APE parallel calculator, capable of one billion operations per second, equal to the best offered by the world market but at a much lower cost.
The arithmetic of APE was based on an innovation that years later found application in all modern microprocessors: performing several operations for each machine cycle. APE performed 8 operations per machine cycle and therefore derived 64 million operations per second, which with sixteen parallel computing units became about one billion operations per second.
In 1989 a second parallel calculator was created from this program: APE100, a hundred times more powerful than APE. This great increase in computing power was achieved by making almost a thousand processors work together.
In APE commercial electronic components were used while in APE100 there was a chip designed directly by INFN (the MAD, Multiply and Adder Device) and containing about 150 thousand transistors that performed the functions that in the first APE required 300 chips. This family of computers was industrialized and marketed under the name Quadrics by Alenia Spazio.
In more recent years, APEmille has continued the APE tradition. APEmille calculation systems are installed in various locations in Italy, Germany, France and the UK. A new generation of APE systems (apeNEXT) is at an advanced stage of development.
Origin: CNUCE (currently ISTI of CNR of Pisa)
In the early 1990s, CNUCE submitted a feasibility study to the Governing Bodies of CNR in order to acquire a massive parallelism architecture to be installed at CNUCE, as the host institute both the primary provider pole of the CNR processing centre system and the parallel computing support initiative.
The nCUBE 2 parallel supercomputer, produced by nCube Corporation of Beaverton, Oregon (USA), was purchased by CNR in January 1991 and given to the CNUCE Parallel Computing Group (Dr. Domenico Laforenza).
The model purchased was the nCUBE 2 6400, with 128 processors (each with the same power as a VAX/780), having a hypercube interconnection topology.
The nCUBE supercomputer had a proprietary 20 MHz processor, and the number of such processors was scalable from 8 to 8192 with a memory of 4 Mbytes each.
The computer was always available online for qualified researchers and was mainly used by mathematicians, chemists and engineers. It was decommissioned at the end of 1998.
The photographic archive of the University of Pisa contains a Tools Section, dedicated to iconographic and documentary material from the Museum of Calculation Tools.
It also contains material related to the CEP.