Section 3

The Instruction Repertory

In this Section are given detailed descriptions of the various functions in the Orion instruction repertory. Each description is given in a precise symbolic form and also verbally. Exceptions and special cases are given and brief examples of the uses of some instructions. When discussing some of the simpler groups of functions, a general introduction to the group is given, followed by a short description of the individual functions within that group. The more complex instructions are described individually in detail.

It is assumed that the reader is familiar with the differences between the 2-address and 3-address types and with the modification and replacement facilities. In these descriptions the X- and Y-addresses are the effective addresses at the time the instruction is obeyed, after all replacements and/or modifications have been done, including possibly pre-modification by one or more 116 and/or 117 instructions.

Timing of Orion Programs

Orion 1, like other machines of its generation, does not have fixed instruction times as do simple serial machines. The following are some of the more important factors which introduce variations into instruction times.

1) Carries

Orion has a parallel adder with a carry detection circuit and any micro-program operation which refers to the result of an addition or subtraction may be subject to a delay whilst waiting for this carry to die down. This applies particularly to the carries generated in a modification and the carry produced as a result of arithmetic orders although in this latter case, due to the carry-speed circuits and the fact that the next instruction is read before the result is written away, the maximum delay is 8 microsecs.

2) Address Checking

The address-checking circuits on Orion, for both lock-outs and reservations, operate in a synchronous serial manner. Reading from the core store can take place without waiting for the address checking to be completed, but at the end of each instruction that jumps or performs a write there may be a delay for this checking to be completed. The core store operates in 12 microsecs. and an address can be checked in 16 microsecs so that instructions which do 4 operations on three addresses, such as the two address 00 instruction, can operate in 48 microsecs although this depends on starting the instruction at the correct phasing with respect to the address checking. If an address which is being checked for lock-outs agrees with one of the lock-out addresses in the most significant 7 bits of the address (this address having been left justified through n places when the store has 2^15-n words) then the lock-out checking will take an extra l6 microsecs even though the word may not be locked out.

3) Hesitations

When a peripheral transfer is initiated the number of words or characters called for is added into the program timer although when reading magnetic tape or using modes 1 or 21 on paper tape there may be less actual hesitations. When a program is running at the same time as a peripheral device the peripheral device control unit has priority over the central machine in use of the core store and whenever the peripheral uses a core store cycle the timer does not get augmented; this happens whether or not the computer required access to the store at that time. The occurrence of hesitations prevents the computer from making effective use of the 12 microsec core store cycle, on the other hand if a hesitation occurs at the time a jump instruction is about to jump, the jump can actually take place quicker than it otherwise would.

The time taken to obey a section of program on Orion is not directly the sum of the times for the individual instructions since each instruction time is a function of the conditions at entry and so depends on the preceding instructions in the program. None-the-less it is hoped that the figures given below are useful averages when due notice is taken of the effects described above. While efforts have been made to ensure accuracy, the times required by instructions may, in the end, be different for technical reasons from those given here.

3.A Instruction Times for Orion 1

This section gives the average time taken to obey instructions. For those which involve an arithmetical process (this includes comparisons made for jumps and table searching etc.) the time may be faster or slower than that stated, depending on the carries involved, i.e. on the operands. The timing for the floating-point instructions depends to a great extent on the operands.

3.A.1 Orion 1 Instruction Times ( Times in micro seconds)

Replacement of an address takes 16 microseconds.

	3-address	2-address	2-add unmodified
Group 0	80	80	64
(03 and 04)	64	80	64

Group 1	64	64	48
(13 and 14)	48	64	48

Group 2	As for Group 0

Group 3
30,31	208	208	192
32	240	240	224
33	272	288	272
34	80+F	80+F	64+F

where where n is the number of significant bits in Y.

Group 4 (average times for 4 runs)

40	765	773	755
41	700	718	686
42	1725	1737	1728
43	588	598	571
44	620	626	608
45	621	631	610

Group 5

n is the number of places shifted.

50 to 53

n<24	64+4n	64+4n	48+4n
n≥24	64+4(n-24)	64+4(n-24)	48+4(n-24)

54 to 57

n<48	128+4n	140+4n	112+4n
n>48	128+4(n-48)	140+4(n-48)	112+4(n-48)
54 for n=48	129+192	140+192	112+192
55-57 for n=48	As for n>48

For 54 and 55 if x*<0 then partial justification takes 16 microseconds.

	3-address	2-address	2-add unmodified
Group 6
	Jump, No Jump	Jump, No Jump	Jump, No Jump
60 to 62	120, 80	96, 64	64, 32
63 and 64	112, 80	96, 64	64, 32
65	120, 64	96, 64	64, 32
66	120, 64	96, 64	72, 32
67	104, 80	96, 64	72, 32
Group 7
	Jump, No Jump	Jump, No Jump	Jump, No Jump
70 to 73	96, 64	80, 48	64, 32
74	80, 64	--, 48	--, 32
75	96, 48	80, --	64, --
76 and 77	96, 48	80, 48	64, 32
Group 8
	Jump, No Jump	Jump, No Jump	Jump, No Jump
80	128, 64	144, 80	128, 64
81	80, 112	96, 128	80, 112
82	112, 80	128, 96	112, 80
83	96, 96	120, 112	96, 96
84	96, 166	128, 192	96, 166
85	176, 80	208, 112	176, 80
86	64	80	64
87	72	96	64
Group 9
90	160	170	153
91	176	186	169
92	160	168	152
93	134	144	120
94	272	287	256
95	400	416	397
Group 10
100	560	560	52
101	1203	1251	1203
102	129+4n	129+4n	113+4n
where n = no. of shifts to standardise
103	98+4n	98+4n	82+4n
n=m or m-24, m is no. of places shifted
104	400	400	384
Group 11
110 and 112	96	80	64
111	88	80	64
114	112	112	96
115	80	92	64
116	28	32	28
117	32	32	32

Group 12

120
Y ≥ 0, n = Y 68+4n 76+4n 48+4n
Y < 0, n = 3-Y

121
0 ≤ Y ≤ 23, n = Y
24 ≤ Y ≤ 255, n = Y-24 64+4n 64+4n 48+4n
-23 ≤ Y ≤ -1, n = 3-Y
-255 ≤ Y ≤ -24, n = -21-Y

122 80 96 112

123 144+24E 172+24E 176

where E=8-n for n≠0 and E=0 for n=0, n being the number of characters to be changed.

124 96+4n 76+4n 48+4n

where n is the no. of places shifted. If Y<0 add 12 microsecs

125 224+4m

where m is the no. of shifts needed to standardise.

126 113 124 101

Group 14

142 112+20Y 128+20Y 112+20Y

143 96+30n 96+34n 96+30n

n is the number of words

144,145,146 112+20n illegal illegal

3.A.2 Orion 2 Instruction Times

Times in microseconds

Replacement of an address takes 2.5 microseconds.
Modification of an address takes 3.5 microseconds.
Modification of two addresses takes 4.5 microseconds.

* An asterisk indicates a typical value.

3-address and 2-address unmodified

Group 0

00, 01, 02, 05, 07             12
03                                  10
04                                   8.5
06                                  11

Group 1

10, 12, 15                      10
11                                 11
13                                  8.5
14                                  6.5
16                                  8.5
17                                 10.5

Group 2

20, 21, 22, 25, 27            10.5
23                                   9.5
24                                   8.5
26                                 10

Add 1 microsecond for an odd pseudo-register.

Group 3

30                                 41.5
31                                 38
32                                 42
33                                 50
34                                 40.5

Group 4

40                               152
41                               150
42                               294
43                               150
44                               153
45                               153

Group 5

L = left shift, R = right shift, where L and R are directions of shift performed by the computer.

n = shift number

[x] = integral part of x.

Add 1 microsecond if shift number is negative.

d(x) = 1, x ≠ 0; d(x) = 0, x = 0.

50, 51 L 11 + n

50, 51 R 11 + + (n-1)mod 2 + d(n)

52,53 L 11 + 2 + + n mod 2

52, 53 R 11 + + + n mod 2

54, 55 L 19.5 + 2n

54,55 R 19.5 + + 2((n-1)mod 2) + 2 d(n)

56, 57 L 18.5 + 4 + 2 + 2(n mod 2)

56, 57 R 18.5 + 2 + + 2(n mod 2)

Add 1.6 microseconds for a partial justification.

3-address 2-address unmodified

Group 6

60 to 65 10.5 8.5
≠66, 67 10.5 8.5

≠ Add 1 microsecond for an odd pseudo-register.

3-address 2-address unmodified

Group 7

70 to 75 8.5 6
76, 77 10.5 6

3-address or 2-address unmodified

Group 8

80 to 83	11
84 Successful jump ) 85 Unsuccessful jump )	10.5 + 2.8 ((n+1)mod 8)
84 Unsuccessful jump ) 85 Successful jump )	30
86	7.0
87	7.5

Group 9

90, 91	24 + n + 1.6m
92	25 + n + 1.6m
93	16.5 + 1.6m
94*	54
95*	160
97	26 + n + 1.6m

n is the difference in exponents
m is the number of places of shifting needed for restandardisation
For 90, 91, 92, 93, and 97, add 1 microsecond if y_e > x_e.

Group 10

100*	130
101*	140
102*	380
103*	288
104*	50

	3-address	2-address unmodified
Group 11
110	15	12
111, 112 ≠	14	10.5
114	15	15
115	12.5	12.5
116, 117	8.5	8.5
≠ Add 1 microsecond for an odd pseudo-register.

Group 12

120	13.5 + 1.6n + m n is the number of places shifted.	13.5 + 1.6n + m m is the number of 1’s.
121	11 + + + n mod 2 n is the number of places shifted.	Same as 3-address.
122	10.5 + m m is the number of characters shifted.	13 + m
123	21 + 2m (m≠0); 14.5 (m=0) m is the number of characters unchanged.	23.5+2m (m≠0); 16.5 (m=0)
124	16 + 2.6n Add 1 microsecond if a 1 is found. Add 1 microsecond of Y < 0 n is the number of places shifted.	12.5 + 2.6n
125	217 + 70m m is the number of places shifting needed for restandardisation.	217 + 70m
* 126*	220	220

Group 14

142 pair	36.5 + 6.1n	36.5 + 6.1n
143	17.5 + 2.9n	17.5 + 2.9n
144, 145	13.5 + 4.9n	illegal
146	16 + 7.5n	illegal