IBM 709 General Information Manual page 18

Using Figure 28, assume that the magnetic tape unit
marked "1" is reading data. Because the tape data
are arranged in six-bit groups across the width of the
tape, it takes six of these groups to make up a 36-bit
word. As these groups are read from tape, they are as-
sembled in the tape control. When a full word is as-
sembled, this word is sent to the central processing
unit and from there to the core storage location speci-
fied by the instruction that caused the tape to be read.
Actually, two instructions are required to read data
from any input-output device into core storage. The
first is a read-tape-I instruction with "read" as its op-
eration part and "tape 1" as its address part. The
execution of this instruction: (1) selects the proper
tape unit, (2) puts it into read status, and (3) starts
the tape moving in the proper direction. The second
instruction needed is called a copy instruction. "Copy"
is the operation part with the location that the data
should enter (core storage) as the address part. If a
copy-1000 instruction were executed, the data read
would be entered into core storage location 1000. A
single copy instruction moves one full word into stor-
age. If two or more words are to be moved, two or
more copy instructions must be furnished, each with
a different address part.
INSTRUCTION ADDRESS
Read Tape I
Copy 1000
Copy 1001
Copy 5000
REMARKS
Gets tape 1 ready and moving
Puts that word into location 1000
Puts second word into location 1001
Puts third word into location 5000
This type of instruction group is called a "copy
loop" or a copy routine. A true "loop" is fully ex-
plained under the heading "Index Registers."
The information paths and the components used in
the 709 system are shown in Figure 29. Comparing
Dota Synchronizer Unit
Channel
II
A" Channel"
BII
Figure 29. Information Flow,
IBM
709 System
Figures 28 and 29, note that the major difference be-
tween the two systems is the fact that data coming
from or going to core storage do not pass through the
central processing unit. Instead, a new device called
a data synchronizer is used as the go-between for in-
put-output devices and core storage.
With the 709 computer, the stored program starts
an input-output operation by defining which device
is to transmit (read tape) and what area in storage is
to receive the data; then the central processing unit is
free to do other calculations. Note the main differ-
ence between the 704 and the 709. In the 704, the cen-
tral processing unit must wait until all data have been
transmitted while the 709 merely starts the operation
and is then free to do other work. This subject is ex-
panded under the section
"IBM
766 Data Synchro-
nizer."
The 7090 computer system uses basically the same
information paths as does the 709. Two new units are
added in the 7090 system. They replace the data syn-
chronizer's operation. They are called "multiplexor"
and "data channel." Their operation is explained in
the section
"IBM
766 Data Synchronizer." Figure 30
shows information flow and components in the 7090
system.
There may be a maximum of eight data channels at-
tached to the multiplexor. Each data channel may
have the same complement of input·output devices as
shown in Figure 30. This feature gives a maximum of
80 magnetic tape units, eight card readers, eight card
punches, and eight printers.
Figure 30 shows that the central processing unit of
the 7090 system has even less control over core storage
and the input·output devices than does the 709 system.
The main difference in all three systems is the pro-
gression from a synchronous computer (704) to an
asynchronous computer (709-7090), as far as the in-
put·output controls are concerned. Here, synchronous
is defined as one operation happening after another
has finished, while asynchronous means simultaneous
occurrence of several operations.
L ~~~~J~-"""1Central
Processing
Unit
Figure 30. Information Flow,
IBM
7090 System
Central Processing Unit 17