• IBM 7030 - "Stretch"
• Harwood Kolsky (hardware), Fran Allen (software), Ann Hardy (software), ?, Fred Brooks (software manager) in front of the CHM STRETCH maintenance and operator console after the presentation "IBM Stretch: The Forgotten Computer that Helped Spark a Revolution" at CHM September 11, 2008. Fred Brooks is normally very alert - the long flight and long day must have had effect ;-)) - photo by Robert Garner -

Word Length: 64 bits plus 8 bits for parity 
and error checking.

Memory Size: 1 to 8 16K core memory
stacks, self-contained wach with its own
clock, addressing circuits, data registers 
and checking circuits, addressing of up 
to 256k word locations.

Data Transfer Rate:  Addressing of memories
and transfer of information from and to
memories ba a memory bus permits new 
addresses, information, or both to pass 
through the bus every 0.220 micro-seconds.

Central Processor: The processor consists 
of the instruction unit, the look-ahead unit,
a parallel arithmetic unit and a serial arithmetic 
unit.  Multi-programming through program
interruption and address monitoring, and 
over-lapped or parallel execution of instructions
is possible.

Instruction Format: Half-word formats
accommodiate indexing and floating-point
instructions.  Full-word formats are used by
variable-field-length instructions.  Five
instruction sets and 765 different types of
instructions are used.

Technology:  Standard Modular System
Transistor Cards.  Used 150,000 high-speed drift
transistors, and provided interleaved magnetic
core memory with 2.18 usec access cycle.

Comments from Bob Bellizzi 10/3/2008

Regarding Gordon Bell's notes on memory The core array was in an aluminum tank, immersed in an oil bath. The oil was circulated through a system that maintened a constant temperature; it had both heating and cooling. The oil allowed the system to maintain the cores at the optimum point for fastest switching which required maximum current and a stable temperature. The most interesting maintenance issue occurred in the POK mfg building where we put together the first system to ship. We were in the final test phase of manufacturing. Everything was hooked together and the integration was being tested prior to turning it over for acceptance testing. This was not a raised floor so you had to step over masses of cables or end up crushing some of the inter-gate coax. The memory failed about 10 pm on a Sunday night. Since we were operating a 4 shift, rotation with 6 days on and one day off, we were all punchy. We called Pete Perez back in; he had just left after working 72 hours with less than 12 off per each of the 6 days. He opened the small gates above the core tank. He ran some simple tests and decreed that he had to drain the tank and remove the array (all the joints were soldered). He stuck his head inside the memory rologon frame, removed the appropriate attachments. He bent down, set up a drain system and started to stand up when he hit the latch on the small gate that was swung out. He lost control of the drain and also started bleeding all over the place. The cut wasn't much but blood pours out of the scalp when it gets cut. After he was taken to the hospital we had to clean everything up; took a whole shift to get back on the air. George Werner, who was in charge of the integration team was livid because he had made a schedule that none of us agreed we could meet but still sold it to management. He seemed to think poor Pete did it to give us a break

Comment by Ed Thelen

• IBM Stretch (7030) -- Aggressive Uniprocessor Parallelism"
• "Computer Structures: Readings and Examples", (1971 edition) by C. Gordon Bell & Allen Newell, Part 5, Section 2, Chapter 34
• Planning a Computer System - Project Stretch

http://wotug.ukc.ac.uk/parallel/documents/misc/timeline/timeline.txt
========1956========

IBM starts 7030 project (known as STRETCH), with the goal of producing
a machine with 100 times the performance of the IBM 704, initiated by
Atomic Energy Commission at Los Alamos.  (MW,HGK: IBM, STRETCH)

from Computer Museum History Center "CORE" 1.2

In the early to mid 1950s, IBM and UNIVAC, the only two 
large companies building computers, were considering
the use of transistors in their products. Though the transistor 
effect had been discovered in 1947 at Bell Labs,
vacuum tubes remained commonplace in computer hardware, 
while American manufacturers struggled to make a
reliable, mass-producible transistor.

Today it may seem surprising that IBM was undergoing 
tremendous turmoil about its role in the new field of
computers. However, the public had begun to associate 
the UNIVAC name (not IBM) with computers. CBS's
1952 election coverage included a UNIVAC machine that 
correctly predicted Eisenhower's victory. And, when
former IBM customers started assigning key contracts 
to UNIVAC, IBM executives took notice.
Steve "Red" Dunwell and Werner Buchholz, two senior 
IBM engineers, proposed a new machine, code-named
"Datatron:" Based on transistors, the machine would 
enable IBM to leap ahead of UNIVAC and would embody
many new architectural concepts.


100 TIMES FASTER
In a famous memo dated October 25, 1954, Dunwell wrote: 
"The Datatron program is intended to assure IBM a
preeminent position in the field..." and will "take 
a giant step and make substantial advances on all fronts.
" A team of senior IBM technical and management staff 
met to consider building what John von Neumann had earlier
exhorted them to create: "the most advanced machine... 
possible in the present state of the art." Besides
allowing IBM to leapfrog its main competitor, Dunwell 
argued that the machine would allow IBM to unify its 
various computer products - roughly divided along scientific 
and business lines - thus greatly reducing manufacturing
costs and simplifying IBM's engineering and production processes.


After great internal debate and a contract from Los Alamos 
Scientific Laboratory, the project went ahead. Now
codenamed "Stretch," the machine was to be "100 times 
faster than the most advanced computer working
today," and President Tom Watson proudly noted that 
the new machine could complete "100 billion computations
in a day."

THE NEWS SPREADS
The first machine (officially named the IBM 7030) was 
delivered to Los Alamos on April 16, 1961. Although far
short of being 100 times faster than competing machines, 
it was accepted and ran for the next 10 years, with the
thenastonishing average reliability of 17 hours before failure.
While customers were generally happy with the machine's 
performance, internally, Stretch was considered a
failure for not meeting its speed benchmark. IBM reduced 
the price from $13.5 million to $7.78 million, thus
guaranteeing that every machine was built at a loss. 
Dunwell's star within IBM fell dramatically, and he was given
fewer responsibilities.

As time went on, however, attitudes within IBM changed. 
From a lagging position in industry, IBM had moved into
the forefront through the manufacturing, packaging, and 
architectural innovations Stretch had fostered. Dunwell's
exile ended in 1966, when the contributions Stretch had 
made to the development of other IBM machines
including the monumentally successful System/360 product line 
- became evident. Dunwell was made an IBM
Fellow that year, the company's highest honor.

THE SUCCESSFUL FAILURE
The Stretch story is only one of many in the history of 
computing that shows how triumphs are built upon the
ashes of "failures:" Stretch is one of the hallmark 
machines - despite its near invisibility to history 
- that defined the limits of the possible for later 
generations of computer designers and users. You may 
recognize many Stretch innovations in present-day products: 
	Multiprogramming 
	Memory protection 
	Generalized interrupt system
	Pipelining 
	Memory interleaving 
	Speculative execution 
	Lookahead (overlap of memory and arithmetic ops) 
	Concept of a memory bus 
	Coupling two computers to a single memory 
	Large core memory ( 1MB) 
	The eight-bit character (the "byte") 
	Variable word length 
	Standard I/0 interface

Ironically, microprocessor companies 20 or 30 years later 
"re-invented" most of these innovations. The Computer
Museum History Center has parts of the original Stretch 
machine (serial number 1) from Los Alamos and a
complete. Stretch (minus core memory unit) from the Lawrence 
Livermore National Laboratory.

STRETCH SPECS
The Stretch covered 2,500 square feet, the size of the 
average American home, and weighed approximately
40,000 Ibs. The CPU alone was 900 square feet 
(30' x 6' x 5'). Nine machines were ultimately produced 
and sold for $7.78 million each (1961 dollars). 
The processing units alone used 21kW.

Stretch employed aggressive uniprocessor parallelism; 
had an instruction set of 735 instructions (including
modes) of variable field length; used magnetic core memory 
(6 x 16KW, 2.1us cycle time): and had 169,200
transistors. The basic machine cycle was 300ns (3.3 MHz), 
and it performed at approximately 500 KIPS (code
dependent). Stretch accommodated word lengths of 64 + 8 check bits 
(SECDED), had a disk of 2MW and 8Mbps, and used 
magnetic tape in its 12 x IBM 729 IV tape drives. The 
machine had a 1,000 cpm (card per minute) card reader; 
a 600 Lpm printer; and a 250 cpm card punch.

FURTHER READING
Bashe, Charles, et al. IBM's Early Computers. Cambridge: 
MIT Press, 1986, pp. 416-468.

Blaauw, Gerritt, & Brooks, Frederick. Computer Architecture: 
Concepts and Evolution. New York: Addison Wesley, 1997.

Buchholz, Werner. Planning a Computer System: Project
Stretch. New York: McGraw-Hill Book Company, 1962.
Out of print.

Dunwell, S. W. "Design Objectives for the IBM Stretch
Computer." Proc. Eastern Joint Computer Conference.
December 1956, pp. 20-22.

 - - - - - - - - - - - - - -

 Dag Spicer is Curator & Manager of Historical Collections -
at The Computer Museum History Center
 
A version of this article first appeared in
Dr. Dobbs Journal online.