ENIAC World Wide Web
BRLESC Ballistic Research Laboratories Electronic Scientific Computer (U.S. Amry Photo)
Exterior ballistics problems such as high altitude, solar and lunar trajectories, computation for the preparation of firing tables and guidance control data for Ordnance weapons, including free flight and guided missiles.
Interior ballistic problems, including projectile, propellant and launcher behavior, e.g., physical characteristics of solid propellants, equilibrium composition and thermodynamic properties of rocket propellants, computation of detonation waves for reflected shock waves vibration of gun barrels and the flow of fluids in porous media.
Terminal ballistic problems, including nuclear, fragmentation and penetration effects in such areas as explosion kinetics, shaped charge behavior, ignition, and heat transfer.
Ballistic measurement problems, including photogrammetric, ionospheric, and damping of satellite spin calculations, reduction of satellite doppler tracking data, and computation of satellite orbital elements.
Weapon systems evaluation problems, inlcuding antiaircraft and anti- missile evaluation, war game problems, linear programming for solution of Army logistical problems, probabilities of mine detonations and lethal area and kill probabilities of mine detonations, and lethal area and kill probability studies of missiles.
Internal number system Binary Binary digits/word 68 + 4 parity Binary digits/instruction 68 Instructions/word 1 Instructions decoded 33 Arithmetic system Fixed and floating point Instruction type Three-address Instruction word format 4 4 6 14 6 14 6 14 Order Para- Index (alpha)-Ad- Index (beta)-Ad- Index (gamma)-Ad- type meter dress dress dress Number word format Fixed Point 3 1 4 . 60 Tag Sign Binary Point Number word format Floating Point 3 1 4 . 52 8 Tag Sign Binary Point Coefficient Biased Exp of 16
In additon to the standard set of jump instructions, three more jump instructions have been included which will be used in connection with the "permanent" storage of "built-in" subroutines. These are Jump to "permanent" instruction, Jump to "built-in" subroutine, and Set index and jump to main memory.
The machine will have 63-one microsecond access index registers, addressable by the alpha, beta, and gamma addresses of the instruction words.
The parameter bits of the instruction word are used to indicate variations of the basic order type.
All three arithmetic registers are 68 bits. Tag bits enter these registers only on the logical instructions and the shift instruction if it is cyclic or is a Boolean shift. On arithmetic orders, the tag bits are saved in a separate three bit register and the three extra bits in the arithmetic registers are used for checking overflow. Thus the range of numbers in the arithmetic unit is -128 less than or equal to N less than 128.
Add and subtract are performed the same as for normalized arithmetic, except the result is never shifted left at the end of the operation.
Before multiply is done, the coefficient that has the largest absolute value is normalized. There is no left normalization after the operation. Thus the result has approximately the same number of significant digits as the operand that had the smaller number of significant digits. It does tend to retain an average of about two or more bits than it should, however.
Before divide is done, both operands are normalized but the number of divide steps performed is reduced accordingly so that the result has approximately the same number of significant digits as the operand that had the smaller number of significant digits.
Microseconds Operation Excl A T Incl A T Fixed point add or subtract 1 5 Fixed or floating multiply 20 25 Fixed or floating divide 60 65 Floating add or subtract 3.0 6 Boolean logic operation 1 5 Indexing and control 2 2(Avg)
The arithmetic unit is constructed of standard vacuum tube logical packages, with tube driven, crystal diode logical gating. The arithmetic unit only is constructed of 1,727 vacuum tubes of 4 types, 853 transistors of 3 types, 46,500 diodes of 2 types and 1,600 pulse transformers of 1 type.
Arithmetic mode Parallel Timing Synchronous
Logical events are controlled by a five-phase clock, permitting decisions at a 5 Mc rate.
Indexing and control will be concurrent with arithmetic operations.
Except for arithmetic or Boolean compare instructions, the test overflow instructions with P_33 = 1, or any arithmetic order that stores in any index register or stores in the location of the next instruction, the machine always gets its next instruction from the memory while it is doing the previous instruction. If this next instruction is one of the control and indexing orders, it is immediately done, unless it is an input-output order or a test overflow order. If it is done, it proceeds to get another instruction and do it, if possible. Thus almost all of the control and indexing orders can be done concurrently with the arithmetic or logical orders. Only the arithmetic and logical orders require the use of the main arithmetic unit of the machine.
All types of input-output orders can be done concurrently with other instructions. Automatic interlocks are provided so as to prevent timing conflict. Reference to a main memory position within the range of either an input or output instruction will halt the computer until the input or output transfer has occurred at that memory position. The computer is released as soon as the transfer of that particular word has been made and does not wait for the entire transfer to be completed. There is no interlock on the index memory when it is used as index registers. Only the effective addresses alpha, beta, and gamma are conflict checked. The programmer can easily make the computer wait until such a transfer is complete by using the last address in the index range of the inout order in the A, B, or C addresses of a dummy order. An input-output instruction is not started until the previous arithmetic instruction is finished, hence the last arithmetic result may be included in the range of any input-output order.
As many as five input-output orders can be operating concurrently with computing and with each other. There is a separate trunk for reading cards, punching cards, using drum, and two separate trunks for using magnetic tape and all five of these trunks can operate concurrently.
No. of Digits Access Media Words per Word Microsec Magnetic Core (Main) 4,096 72 binary 2 Magnetic Core (Index) 63 16 binary 1 Magnetic Drums (Two) 24,576 Magnetic Tapes (Six) No. of units that can be connected 16 Units No. of chars/linear inch 400 Char/in Channels or tracks on the tape 16 Tracks/tape Blank tape separating each record 0.80 Inches Tape speed 150 Inches/sec Transfer rate 120,000 Char/sec Start time 3.0 Millisec Stop time 3.0 Millisec Average time for experienced operator to change reel 60 Seconds Physical properties of tape Width 1.0 Inch Length of reel 2,500 Feet Composition 0.43 Magnetic coating 1.45 Mil
Provision is made for up to 16,384 words of high speed memory and system can be expanded to 28 tape stations.
Media Speed Card Reader 800 cards/min Magnetic Tape See "Storage"
Media Speed Card Punch 250 cards/min Magnetic Tape See "Storage"
A single unit that is capable of converting alphanumerical characters from cards to tape, tape to high speed printer, tape to cards, cards to high speed printer and paper to magnetic tape.
Type Quantity Tubes 5847 5,600 6197 110 6C4 110 6AQ5 220 Misc 80 Diodes LD70/CTP309 12,600 LD71 100,000 Misc 13,700 Transistors 2N697 600 2N1143 240 2N398 1,600 Misc 6,300
Code checking features will include stopping on any selected address, the display of the contents of any memory cell, the display of normal or abnormal conditions, the ability to manually store in any selected memory cell, and the ability to transfer control to any part of the system. Parity checking is performed in each of the four 17-bit groups in each word.
Power, computing system 35 Kw Power, air conditioner 20 Kw Space, computing system Plenum is 30 ft x 40 ft Space, air conditioner (see below) Capacity, air conditioner 25 Tons
Chilled water is sent two flights up to computer site to heat exchanger, transferring heat from computer closed loop air to closed loop chilled water. On ground floor, compressor refrigerant absorbs heat from chilled water. An evaporative system absorbs heat from refrigerant in a cooling tower. Compressor located two floors below. Liquid coolant piped upstairs. Heat exchanger, computer closed-loop air-to-coolant at computer site, and coolant-to-outside air downstairs.
Number of systems produced to date 1 Operational date anticipated as 1 April 1961.
The approximate cost, including an additional bank of 4,096 words of high speed memory, 6 tape stations, the sytem as described, with all peripheral converters and input-output equipment, site preparation, overhead and other related costs will be approximately 2.0 million dollars.
Three 8-Hour Shifts Supervisors 6 Analysts 3 Programmers and Coders 14 Clerks 1 Engineers 1 Technicians 6
No engineers are assigned to the operation of the machine, but are used for development and design of additions to the machine. The technicians consult the engineers when a total break-down occurs.
A high degree of reliability is achieved by utilizing standard logical plug-in packages, a riggedized, long life, driver tube, derated components and point-to-point soldered connections.
Computing Laboratory Ballistic Research Laboratories Aberdeen Proving Ground, MD