Table of Contents
Advertisement
and one access is required for each pair of charac-
ters. Buffer access time is 2 µs per byte pair.
Each 2250 is polled sequentially by address for
buffer service requests, starting with the 2250
having the next-higher address from the 2250 cur-
rently receiving service. Thus, all 2250' s have
equal priority in the polling sequence, regardless
of address assignment. Polling is accomplished
concurrently with buffer service; therefore, buffer
efficiency approaches 100 percent during periods
of high demand.
The execution time required for the order por-
tion of a display program can be computed by (1)
summing the number of order byte-pairs in the
individual display program for each 2250 and (2)
multiplying the total by 2 µs. The result of this
calculation is then subtracted from 25 ms to deter-
mine the time available for actual image display.
It
is therefore desirable to minimize the number of
orders in each 2250 display program.
Beam positioning time is obtained in one of two
ways:
(1) If N
<
16,
Positioning time
=
8
µs, or
(2) If N>
16,
Positioning time
=
8
+ 92 (N-16) µs,
1007
where: N
=
number uf
raste1'
units of the axis
(X or
Y)
having the greater change. For ex-
ample, when the X-axis change is 100 raster
units and the Y-axis change is 1, 023 raster
units (full scale deflection) , N is equal to
1023.
NOTE: Add 1 µs to the beam-positioning time cal-
culated for each unblanked point; this allows for
intensification after beam motion is complete.
These methods of computing beam-positioning time
can also be used to compute Character mode fly-
back time (the time to reposition the beam to the
start of a new line). However, the beam-position-
ing time in Graphic mode is not equivalent to actual
elapsed time between adjacent beam deflections and
cannot be used to compute execution time for a
display program. Fetching of the next Graphic mode
X-Y coordinate is overlapped with beam motion for
the current X-Y coordinate. This fetch time in-
cludes polling for 2250 buffer service requests,
buffer access, decision and data flow logic delays,
and cable delays. The sum of these delays often
exceeds beam -positioning time, resulting in a wait
between adjacent beam deflections.
Fetch time imposes a minimum effective Graphic
mode cycle time of 11. 8 µs per absolute deflection
and 9. 0 µs per incremental deflection. When several
consoles are operating simultaneously, the fetch
time increases, as does the minimum cycle time.
For example, assume that four 2250 's are simul-
taneously drawing absolute vectors that require 8 µs
each for beam-positioning. Eight buffer accesses,
totaling 16 µs, are required for each group of four
vectors. Therefore, the effective Graphic mode
cycle time is at least 16 µs even though beam-
positioning requires only
8
µs. If all four 2250' s
were drawing absolute vectors with beam position -
ing time greater than 16 µs, the Graphic mode cycle
time would be approximately equal to beam-position-
ing time.
Tables 3 and4 list the number of vectors that can
be displayed by a 2250 during one 25-ms (40 cps)
regeneration cycle and during one 33. 3-ms (30 cps)
regeneration cycle, respectively, when operating
in various system configurations. The messages
used
to
obtain the figures in these tables contained
one Enter Graphic Mode order, one Start Regener-
ation Timer order, and one Transfer Unconditional
order.
Character execution time is dependent on the
number of strokes required to draw the character.
An average of six strokes is required per visible
character. The space (blank) , null, and new line
characters are each equivalent to a one-stroke
visible character. A text message that contains a
normal complement of spaces will average five
strokes per character position. Flyback time can
be computed from the beam -positioning formula
presented earlier in this section. For practical
purposes, total message flyback time is propor-
tional to the number of character in a message and
is almost independent of the number of lines used
to display the message; this assumes that all lines
start at the left margin
(X
=
0000) and that new line
characters terminate each line after the last visible
character.
The number of characters that can be displayed
per console in 25 ms is determined by the number
of 2250's displaying simultaneously and by the cable
length. Buffer access delay does not impact char-
acter message size as heavily as it impacts graphics.
The reasons for this are twofold. Pirst, a character
is stored in the buffer as one byte of data; therefore,
one buffer access will fetch two characters (this is
one-fourth the number of accesses required for
absolute vectors). Secondly, the individual strokes
that compose the characters are generated by a
logic matrix which operates independently of the
buffer and which can service several 2250's simul-
taneously.
Tables 5 and 6 list the number of characters that
can be displayed per console in 25 ms (40 cps) and
in 33. 3 ms (30 cps), respectively, (as determined
for several system configurations). The message
2840 Operations with Attached 2250's
27
Advertisement
Table of Contents
