Perform I/O

return to main page

The 1401 Performs I/O For Other Computers

Historical Introduction

UNIVAC was the early electronic computer leader. The early UNIVAC designers recognized that electronic computers could process business data faster than the card reader, card punches, and printers of the day could get data into and out of the computer.
https://en.wikipedia.org/wiki/UNITYPER
The difficulty, costs, and delays of developing the UNITYPER and the off-line mag tape to printer greatly hampered the new company, and helped cause such financial problems that the company was taken over by Remington Rand.
The card readers and printers attached to IBM fast computers seriously slowed the throughput of the systems for data processing. The speed of magnetic tape systems much better matched the processing speed, especially when several card images or print lines were placed in one tape "block" or patch of magnetic information.

Introducing the IBM 1401, October 5, 1959

There was no question in anyone's mind, customers or computer vendors/sales that the 1401 had a great set of peripherals:

The IBM 1402 Card Reader set a new industry standard in speed, accuracy, and operator convenience
The IBM 1402 Card Punch also set a new industry standard for speed and accuracy
The IBM 1403 Line Printer had much higher speed than any previous IBM Printer,
and set a new industry standard in print quality, and operator convenience
The IBM 729 Magnetic Tape unit was already the industry standard for speed, capacity and accuracy
An IBM 1401 system with several magnetic tape units was a price and performance bargain.

Table of Contents:
     - a) Carry magnetic tapes between machines
     - b) Share magnetic tape units
     - c) Share a disc
     (No IBM 1401 was known to be direct coupled to other machines as in HASP)

- a) Carry magnetic tapes between machines

Seemingly the most common, most labor intensive, most error prone.
People carried/carted spools of magnetic tape between tape drives attached to the:
- Fast main machine (which read and wrote to its magnetic tape drives
- Slow I/O machine with the card reader, card punch, printer
To establish that a standard 12 inch diameter magnetic tape reel can hold enough data to make the above feasible, read here.
Jay Jeager estimates that it takes four minutes for an operator to unload a magnetic tape from one machine and reload the same tape on a near by machine.
- b) Share magnetic tape units

One of "our" tape units has an extra socket and logic
Table of Contents":
     - Feature
     - Further Questions
     - German 729 tape with dual interface
     - Why
     - Switched
     - Cabling
     - Power
Feature
From Allen Palmer, March 28, 2017

IBM provided a feature on the 729 tape drive that allowed a tape drive to be switched from one computer system to another without having to manually uncable the drive from one CPU and attach it with cables from another CPU.
Tape drives were connected to the TAU ( Tape Adaptor Unit) in the CPU through a large cable with a (get someone to check on the number of pins in the 729 ‘shoe) xx pin shoe. There are two ‘shoes’ – an ‘IN’ shoes and an ‘OUT’ shoe on the back of each drive. The ‘string’ of drives was serially connected together through these cables. A cable ran from the CPU to the first drive in the line connecting to the IN shoe, then a separate cable was connected the OUT shoe of the first drive to the IN shoe of the second drive and so on down the line, with the last drive having a ‘terminator’ box attached to its OUT shoe.
In order to switch drives from one CPU system to another the drive was built with four shoes. One set of shoes ran to one CPU and the other set ran to the other CPU. In addition to the extra set of shoes there was a bank of relays that switched the internal circuitry of the tape drive from one set of shoes (CPU) to the other set of shoes (CPU). The drive could not be functionally connected to both CPU’s at the same time. The transfer was mechanical in terms that bank of relays were either ‘open’ or ‘closed’ thereby connecting the drive to either CPU.
What this allowed was to physically move a tape drive and its reel of tape and data from one CPU to another without having to either go behind the drive and unattach one set of cables and attach another set or to unload the reel of tape and physically mount it to another drive attached to the other CPU. This transfer was done electrically.
Now ED, the question you have is how was it done. There is a switch on the back of the drive to switch from the A set of shoes to the B set but that would have been very impractical so there must have been a way another way but I can not remember it. Ask the guys to look through the logic systems to find it.

Further questions ...

Now - is there some extra status logic and status conditions that help control who does what with which tape drive ? "Idle, powered up, threaded, rewound, and ready" "I take control of this tape unit" "Unit being controlled by another entity" "Command rejected, unit being controlled by another entity" "I'm rewound, with data for you, and available for you to use" "I have completed processing the data, rewind complete, and available for you to use" and such. Initialization, Status available on control panel for operator ?? The two processors have to communicate in some way, and this might be the simplest. A wired inter-processor link is typically expensive and complicated.

German 729 tape with dual interface
From Ignacio Menendez, Mon, Apr 10, 2017
Ignacio must have read the book "How to win friends, And influence people" ;-))

Ed, abusing your kindness and good nature, could you please add the following Paragraphs which document the dual tape drive interface that exists in only one of our 729s at CHM, tape in position 3 of German (DE) 1401 System. (See below). Thanks, Iggy
Like how can you say NO to that ?? ;-))
--------------------------------------- German 729 tape with dual interface Viewed from rear, 4 biscuit connectors:
Interface |...A...+...B...| | | | |Top IN |Top IN.| |-------+-------| |Bottom |Bottom | | OUT. | OUT. | |-------+-------|
Terminators go on bottom, or cable to next drive down the line, away from 1401. At CHM, tape #3 side A is used, top IN cable comes from tape #2, side A. OUT goes to emulator box, which has terminators inside. At bottom left of tape with dual interface, there are two switches.... for our CHM purpose, these should be ser to Interface 'A,'and 'Local.' Interface B is NOT used at CHM, so BOTH TOP and BOTOM biscuit connectors on the RIGHT SIDE (B) are left unused, and empty, devoid of cables or terminators. The emulator (box under PC) MUST BE ON for the tapes 1,2,and3 to work properly with TAU on German 1401. The emulator can have up to 6 virtual 729s, and a virtual TAU.

Why
From: Marc Verdiell

Iggy,
Thanks for clarifying, the 4 connectors always get me mystified.
What is the intent of the extra B connectors, why and how would you use them?
Connect to a backup controller?
Marc

Switched
From Stan Paddock, Tue, Apr 11, 2017

IBM provided a feature on the 729 tape drive that allowed a tape drive to be switched from one computer system to another without having to manually uncable the drive from one CPU and attach it with cables from another CPU.
Tape drives were connected to the TAU ( Tape Adaptor Unit) in the CPU through a large cable with a (get someone to check on the number of pins in the 729 ‘shoe) xx pin shoe. There are two ‘shoes’ – an ‘IN’ shoes and an ‘OUT’ shoe on the back of each drive. The ‘string’ of drives was serially connected together through these cables. A cable ran from the CPU to the first drive in the line connecting to the IN shoe, then a separate cable was connected the OUT shoe of the first drive to the IN shoe of the second drive and so on down the line, with the last drive having a ‘terminator’ box attached to its OUT shoe.
In order to switch drives from one CPU system to another the drive was built with four shoes. One set of shoes ran to one CPU and the other set ran to the other CPU. In addition to the extra set of shoes there was a bank of relays that switched the internal circuitry of the tape drive from one set of shoes (CPU) to the other set of shoes (CPU). The drive could not be functionally connected to both CPU’s at the same time. The transfer was mechanical in terms that bank of relays were either ‘open’ or ‘closed’ thereby connecting the drive to either CPU.
What this allowed was to physically move a tape drive and its reel of tape and data from one CPU to another without having to either go behind the drive and unattach one set of cables and attach another set or to unload the reel of tape and physically mount it to another drive attached to the other CPU. This transfer was done electrically.
Now ED, the question you have is how was it done. There is a switch on the back of the drive to switch from the A set of shoes to the B set but that would have been very impractical so there must have been a way another way but I can not remember it. Ask the guys to look through the logic systems to find it.
=================================================================
The switches on the back of out one switchable drive are: Local/Remote
Switch to A bank Switch to B Bank.

Our drive is set to Local and A bank

If it is set to Remote, there was a box the operator can switch the drives without having to go behind the tape drive.
We don't have one of these boxes.

The 709X systems supported the IBM 729 tape drives,
This switching would make it easy to use 1401 as a 'spooler' without having to move tapes between systems.

Stan

Cabling
From Ignacio Menendez, Tue, Apr 11, 2017

Marc, to add to what Stan wrote....
That particular drive #3 in DE 1401, could be cabled to be not only #3 on DE, but #5 on Connecticut. This is not an AND, but an OR possibility, depending on the switch position on the lower left of the tape drive rear frame.
The problems to do this may be....
1- cable length spec. and availability restrictions.
2- grounding AC AND DC, since DE System is WYE 220, and CONN. Is DELTA 220 vac. As has been seen, and shown to us, by Bob Feretich, the emulator needs differently adjusted parts to deal with this problem, in order to not get data read problems.
3- the switching is accomplished with wire contact relays, many of them, to switch the hundreds of wires on both biscuit connectors..... We may possibly experience lots of poor contacts on these relays, that remained dormant for years. Not only that, but if switched to side B, when returned to side A, we may also experience connection problems.
The cable needed for remote switching, a feature, may or may not be found. Also, I did not see a socket to connect this cable.
4- power would need to be ON in both systems, if this drive would be #5 on Connecticut System, to provide the 50 Hz. 220 vac to this drive.
5- We need to find ALDs that show the point to point wiring, from biscuits to relays, and from relays to the SMS logic gates.
6- Some other unknown gotchas.
Bob and anyone, please chime in on your observations and thoughts, perhaps this is more trouble than it is worth ?
Iggy

Power
From Bob Feretich, Tue, Apr 11, 2017

Iggy is 100% correct. The different power sources may create issues.
Also, when we first brought the drive up, the relay contacts were a problem. We never tried to debug the other side relay contacts.
Regards,
Bob
- c) Share a disc
Paper Robert Garner's description:

" The STL Integrated Computer Operating System” that describes a 7094/1401 installation at TRW that shared a 1301 disk between a 1410 and a 7094, instead of shlepping tapes between several 1401s and two 7090s.
Their 7090s initially handled 300 - 400 jobs daily, with jobs taking up to an 1 hour, 80% under 5 minutes, with an average of 3.6 minutes (needing 2.7 tapes).
TRW's new system that “returned conventional data I/O functions to on-line processes” improved job throughput by 3x, apparently eliminating the need for a second 7094!
Interesting simulation results and tables…"

Started March 20, 2017
Updated March 30, 2017