Printing - IBM 1130 User Manual

the first 20 columns of 500 of them. For the 1442-6,
the breakdown now becomes:
Operation Milliseconds
Read 2100 cards 420,000
Punch 20 columns, 131,250
500 cards
Something else
Total
With the 1442-7, it becomes:
48,750
600,000
Read 2100 cards 210,000
Punch 20 col. 500 cards 68,250
Something else 48,750
Total 327,000
or 5.5 minutes
N ate that the times shown apply only to the time
actually spent punching. If the card being punched
was previously read, the punch time may be simply
added to the total. If the card being punched was
not previously read, you must add 200 or 150
milliseconds of read time per card to allow for the
(feeding of cards past the read station, even
though they were not read. This will always be
the case with the 1442-5, which cannot read cards.
Section Subsections Page
90
30
I
20 03
Printing
Three different line printers may be attached to the
1130 system, each having different print and skip
times:
Printer
Approximate Time in Milliseconds
Print 1 Line
Skip 1 Line
1132 750
16
1403 Model
61
175 (3.6-
or
microsec-
ond CPU) 5
Model 7
100 (2.2-
microsec-
ond CPU)
To illustrate the improvement pOSSIble
III
this
area, let us take an example similar to the last one.
Suppose you have a program that is essentially a card
listing job. In ten minutes it reads 600 cards,
prints 600 lines, and skips 100 lines. This can be
broken down as follows:
Operation
Read card (1442-6)
Print (1132) 600
@
750
Skip 100
@
16
"Everything else"
Total
Milliseconds
120,000
450,000
1,600
28,400
600,000 (10
minutes)
If
you replace the 1132 with a 1403-6, your print
and skip times drop:
Print (600
@
175) 105,000
Skip 100
@
5 500
Added to the card read time and the "everything else"
time, which remains the same, this results in a total
time of 253,900 milliseconds, or about 4 1/4 minutes,
as opposed to 10 minutes.
Note that despite this dramatic increase in
throughput, the 1403 is printing at only 141 (600/4.25)
lines per minute, far below its rated speed of 340.
The 1132 was also below its rated speed of 80 lpm,
since it printed 600 lines in ten minutes, or 60 lpm.
This shows again that rated speeds of cards per
minute, lines per minute, etc, cannot be used when
investigating alternate approaches to improving
throughput. The only usable figure is the length of
time the CPU is tied up-- that is, prevented from
doing something else.
This example has assumed a 3. 6-microsecond
CPU; if the 1130 were a Model C (2.2 microseconds),
a 1403 time of 100 milliseconds would be used. The
overall time would drop to 3.5 minutes, for a speed
of 172 lpm.
In all cases of 1403 timing investigations, you
must calculate the resulting lines per minute to make
sure that it does not exceed the rated speed of the