Manchester Small-Scale Experimental Machine

From Wikipedia, the free encyclopedia

The Manchester Small-Scale Experimental Machine (SSEM), nicknamed Baby, was the world's first stored-program computer. Developed by Frederic C. Williams, Tom Kilburn and Geoff Tootill, at the Victoria University of Manchester, it ran its first program on June 21, 1948.^[1]

It was designed and built as a test-bed for the Williams tube, not as a practical computer. However, its success resulted in the Ferranti Mark I, the world's second commercially available general-purpose computer.

Contents 1 Origins 2 The first program 3 Later developments 4 References 4.1 Notes 4.2 Bibliography 5 Further reading 6 External links

[edit] Origins

The computer was built as a mechanism to test the Williams tube, a particular type of cathode ray tube (CRT) which had been developed by Williams at the Telecommunications Research Establishment in July–November 1946, before he joined the University of Manchester in December 1946, rather than as a practical computer. After seeing an experiment at Bell Labs to remove the ground echoes that occurred in radar systems, Williams had seen the possibility of using CRTs as storage devices.^[2]

Working with Kilburn at the university they increased the storage capacity of the CRT from one bit to 2048 bits by October 1947 using a 64 by 32 array.^[3] This could be used for a computer's memory, with the advantage of allowing random access to memory, rather than the sequential access of the delay line memory units.

By June 1948, the SSEM had been built.^[3] It used one CRT to provide a single 32 by 32-bit word store, a second to hold a single 32-bit accumulator, and a third to hold the current instruction and its address. A fourth CRT acted as the output device, displaying the bit pattern of any chosen storage tube. The input device was a set of 32 buttons, with manual switches to set the bit pattern of any word.^[3]

Each instruction was stored in a 32-bit word. Bits 0–12 represented the memory address of the operand to be used, and bits 13–15 specifed the operation to be executed; the remaining 16 bits were unused. The second operand of any operation was the accumulator i.e. the SSEM had a single operand architecture. An instruction was executed in 1.2 milliseconds. The main store could be read, written or refreshed in 300ms and it was refreshed every 16 instructions.^[3]

Storage on the SSEM was very limited; it could store 32 numbers and 32 instructions. The instruction set was also very limited. As only three bits were available to identify the operation, there was a maximum of eight (2³) different instructions. The initial seven were:

• Jump to the instruction at the specified memory address.
• Relative jump indirect.
• Take the number from the specified memory address, negate it, and load it into the accumulator.
• Store the number in the accumulator at the specified memory address.
• Subtract the value at the specified memory address from the accumulator, and store the result in the accumulator.
• If the value in the accumulator is negative, skip the next instruction.
• Stop.

There was no add instruction. This was due to the need to minimise the size of the instruction set; an "add" could be constructed using the subtract instruction.

[edit] The first program

The first program consisted of 17 instructions. Written by Kilburn, it was designed to find the highest proper factor of 2¹⁸ by trying every integer from 2¹⁸ - 1 downwards. It took 3.5 million operations and 52 minutes to produce the answer.^[4]

[edit] Later developments

The SSEM developed into the Manchester Mark I, which led to the Ferranti Mark I, the world's second commercially available general-purpose computer.^{[citation needed]} At around the same time EDSAC was being developed at the University of Cambridge Mathematical Laboratory.

A working replica of the SSEM was created in 1998 to celebrate the 50th anniversary of the running of its first program. This is on display at the Museum of Science and Industry in Manchester.

Defining characteristics of some early digital computers of the 1940s (See History of computing hardware)

Name	First operational	Numeral system	Computing mechanism	Programming	Turing complete
Zuse Z3 (Germany)	May 1941	Binary	Electro-mechanical	Program-controlled by punched film stock	Yes (1998)
Atanasoff–Berry Computer (USA)	Summer 1941	Binary	Electronic	Not programmable—single purpose	No
Colossus (UK)	December 1943	Binary	Electronic	Program-controlled by patch cables and switches	No
Harvard Mark I – IBM ASCC (USA)	1944	Decimal	Electro-mechanical	Program-controlled by 24-channel punched paper tape (but no conditional branch)	Yes (1998)
ENIAC (USA)	November 1945	Decimal	Electronic	Program-controlled by patch cables and switches	Yes
Manchester Small-Scale Experimental Machine (UK)	June 1948	Binary	Electronic	Stored-program in Williams cathode ray tube memory	Yes
Modified ENIAC (USA)	September 1948	Decimal	Electronic	Program-controlled by patch cables and switches plus a primitive read-only stored programming mechanism using the Function Tables as program ROM	Yes
EDSAC (UK)	May 1949	Binary	Electronic	Stored-program in mercury delay line memory	Yes
Manchester Mark I (UK)	October 1949	Binary	Electronic	Williams cathode ray tube memory and magnetic drum memory	Yes
CSIRAC (Australia)	November 1949	Binary	Electronic	Stored-program in mercury delay line memory	Yes

[edit] References

[edit] Notes

[edit] Bibliography

[edit] Further reading

• Anderson, David (2004-06-04). "Was the Manchester Baby conceived at Bletchley Park?". Alan Mathison Turing 2004: A celebration of his life and achievements, University of Manchester: British Computer Society. 

[edit] External links

• Computer 50 - A website celebrating the 50th anniversary of the SSEM in 1998.
• A brief history of the Small-Scale Experimental Machine
• Manchester Baby Simulator software