ORACLE (computer)



The ORACLE or Oak Ridge Automatic Computer and Logical Engine, an early computer built by Oak Ridge National Laboratory, was based on the IAS architecture developed by John von Neumann.

Summary
The Oak Ridge National Laboratory Review states: [Oak Ridge National] Laboratory engineers assisted Argonne [National Laboratory] during the early 1950s in design and fabrication of the Oak Ridge Automatic Computer and Logical Engine. Its name was selected with reference to a lyrical acronym from Greek mythology—ORACLE, defined as 'a shrine in which a deity reveals hidden knowledge.'

Assembled before the development of transistors and microchips, the ORACLE was a large scientific digital computer that used vacuum tubes. It had an original storage capacity of 1024 words of 40 bits each (later doubled to 2048 words). The computer also contained a magnetic-tape auxiliary memory and an on-line cathode-tube plotter, a recorder, and a typewriter. Operational in 1954, for a time the ORACLE had the fastest speed and largest data storage capacity of any computer in the world. Problems that would have required two mathematicians with electric calculators three years to solve could be done on the ORACLE in 20 minutes.

[Alston] Householder and the Mathematics Panel used the ORACLE to analyze radiation and shielding problems. In 1957, Hezz Stringfield and Ward Foster, both of the Budget Office, also adopted the ORACLE for more mundane but equally important tasks—annual budgeting and monthly financial accounting. As one of the last 'homemade computers,' the ORACLE became obsolete by the 1960s. The Laboratory then purchased or leased its mainframe computers from commercial suppliers. From the initial applications of the ORACLE to nuclear aircraft problems, computer enthusiasm spread like lightning throughout the Laboratory, and in time, use of the machines became common in all the Laboratory's divisions.

ORACLE was operational (passed acceptance test) in 1953,    and replaced the USAF-Fairchild Computer (or "SPEC" - Special Purpose Electronic Computer).

As with all computers of its era, the ORACLE computer was a one-of-a-kind machine that could not exchange programs with other computers (even other IAS machines). It used vacuum tubes, transistors, and diodes. It used a Williams tube for 2048 words of memory. Its addition time was 70 microseconds, the multiplication time was 370-590 microseconds, and the division time was 590 microseconds. These times include the storage access time, which was about 62 microseconds.

The ORACLE pre-dated input from disks and the use of punch cards with computers; it used paper tape for input and breakage of the tape was a frequent problem.

<!--== Personal memories == These are not facts for which I can provide citations, just reminiscences from a visit I made when I was a child. Maybe something in this can inspire someone else with more "real" information! In 1953, or maybe 1952, there was a math meeting (a sectional meeting of the AMS or MAA national mathematics societies, but at this remove I don't remember which and I have been unable to find any relevant listings at the websites of those two societies) at Oak Ridge: My father was a math prof at U-Tennessee Knoxville. The meeting arranged a tour of the Oracle lab, and my father was able to smuggle me in. (That was still not totally trivial, Oak Ridge as a city was still closed, not just the labs.) So I got to see the machine, it was demonstrated to us, and these are some of its characteristics that I remember. (I was 10 or 11, but they did do a good job of explaining things and for that matter almost all of the older mathematicians were as new to computers as I was.)

The system as we saw it had a desk-like user station across one end of a room that must have been about 35 feet square. Going back perpendicular to that end were three rows of cabinetry. On the right side (looking at the desk) were the tape drives. On the left and going down the middle of the room were rows of cabinets containing the electronic chassis and cooling blowers (about which more below).

There were several, I am remembering about 6, tape drives. But the tapes were not on reels. The tape itself was two inches wide and a drive (by engaging/disengaging two different pressure rollers on capstans) could fairly quickly start the tape moving in either direction. I think there were 12 tracks on the tape, and something like 50 feet of tape on each drive. The tape was pulled from a glass-walled vacuum column on one side to a similar column on the other, and just piled up in the column: Thus there was very little mass to accelerate or decelerate, mostly the tape itself at the head and for a few feet on either side. As I understood it then the tapes were backup storage to the Williams tubes: One did not dismount a tape and mount another in its place, the way tape was used in later decades. The "fast" Williams tube memory was very limited: I think there were 2K words of high speed storage, each 40 bits. So data and code were shuffled back and forth to tape to make space available.

There was a console typewriter for simple I/O. I am pretty sure there were a paper tape reader and punch, but I don't know width or format - maybe someone else can supply that, it would be interesting as a sidelight on how character sets developed. There was also a graphics display. I remember that it was an intensive user of computing resources, and I think the main arithmetic units had to calculate X/Y coordinates at which to position a beam and light up a dot. (We certainly did not say "pixel".) There was some provision for audio output, perhaps like the CDC 1604, I know because as a demonstration they played Christmas music and displayed a Christmas tree.

As an interesting fact about hardware failures: A few years later I know the ORACLE developed some intermittent errors that were very hard to track down. As part of its cooling the machine had large "squirrel cage" blowers, like in a home furnace, inside the electronics cabinets. The blowers were spun by electric motors through V belts. The motors, and probably the blower housings, were mounted on rubber vibration-isolation mounts. After a lot of searching it was found that the belts were operating as van de Graaf generators, building high static charges until a small "lightning bolt" would let it discharge. That of course would cause havoc with the nearby, high impedance, computing circuitry.

I am sorry this is fragmented and imprecise, but maybe others can build on it! This machine was a significant step in the development of computing and deserves better documentation. !-->