1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
|
include "alldifferent.mzn";
include "cumulative.mzn";
%uint128_t x20 = x5 * x13;
%uint128_t x21 = x5 * x15;
%uint128_t x22 = x7 * x13;
%uint128_t x23 = x21 + x22;
%uint128_t x24 = x5 * x17;
%uint128_t x25 = x9 * x13;
%uint128_t x26 = x24 + x25;
%uint128_t x27 = x7 * x15;
%uint128_t x28 = x26 + x27;
%uint128_t x29 = x5 * x19;
%uint128_t x30 = x11 * x13;
%uint128_t x31 = x29 + x30;
%uint128_t x32 = x7 * x17;
%uint128_t x33 = x31 + x32;
%uint128_t x34 = x9 * x15;
%uint128_t x35 = x33 + x34;
%uint128_t x36 = x5 * x18;
%uint128_t x37 = x10 * x13;
%uint128_t x38 = x36 + x37;
%uint128_t x39 = x11 * x15;
%uint128_t x40 = x38 + x39;
%uint128_t x41 = x7 * x19;
%uint128_t x42 = x40 + x41;
%uint128_t x43 = x9 * x17;
%uint128_t x44 = x42 + x43;
%uint64_t x45 = x10 * 0x13;
%uint64_t x46 = x7 * 0x13;
%uint64_t x47 = x9 * 0x13;
%uint64_t x48 = x11 * 0x13;
%uint128_t x49 = x45 * x15;
%uint128_t x50 = x20 + x49;
%uint128_t x51 = x46 * x18;
%uint128_t x52 = x50 + x51;
%uint128_t x53 = x47 * x19;
%uint128_t x54 = x52 + x53;
%uint128_t x55 = x48 * x17;
%uint128_t x56 = x54 + x55;
%uint128_t x57 = x45 * x17;
%uint128_t x58 = x23 + x57;
%uint128_t x59 = x47 * x18;
%uint128_t x60 = x58 + x59;
%uint128_t x61 = x48 * x19;
%uint128_t x62 = x60 + x61;
%uint128_t x63 = x45 * x19;
%uint128_t x64 = x28 + x63;
%uint128_t x65 = x48 * x18;
%uint128_t x66 = x64 + x65;
%uint128_t x67 = x45 * x18;
%uint128_t x68 = x35 + x67;
%uint64_t x69 = x56 >> 0x33;
%uint64_t x70 = x56 & 0x7ffffffffffff;
%uint128_t x71 = x69 + x62;
%uint64_t x72 = x71 >> 0x33;
set of int: CORE = 1..3;
CORE: ADD_MUL = 1; CORE: MUL_CORE = 2; CORE: LEA_BW = 3; array[CORE] of string: CORE_NAMES = ["ADD_MUL", "MUL_CORE", "LEA_BW"];
set of int: INSTRUCTIONS = 1..53;
INSTRUCTIONS: x20 = 1; INSTRUCTIONS: x21 = 2; INSTRUCTIONS: x22 = 3; INSTRUCTIONS: x23 = 4; INSTRUCTIONS: x24 = 5; INSTRUCTIONS: x25 = 6; INSTRUCTIONS: x26 = 7; INSTRUCTIONS: x27 = 8; INSTRUCTIONS: x28 = 9; INSTRUCTIONS: x29 = 10; INSTRUCTIONS: x30 = 11; INSTRUCTIONS: x31 = 12; INSTRUCTIONS: x32 = 13; INSTRUCTIONS: x33 = 14; INSTRUCTIONS: x34 = 15; INSTRUCTIONS: x35 = 16; INSTRUCTIONS: x36 = 17; INSTRUCTIONS: x37 = 18; INSTRUCTIONS: x38 = 19; INSTRUCTIONS: x39 = 20; INSTRUCTIONS: x40 = 21; INSTRUCTIONS: x41 = 22; INSTRUCTIONS: x42 = 23; INSTRUCTIONS: x43 = 24; INSTRUCTIONS: x44 = 25; INSTRUCTIONS: x45 = 26; INSTRUCTIONS: x46 = 27; INSTRUCTIONS: x47 = 28; INSTRUCTIONS: x48 = 29; INSTRUCTIONS: x49 = 30; INSTRUCTIONS: x50 = 31; INSTRUCTIONS: x51 = 32; INSTRUCTIONS: x52 = 33; INSTRUCTIONS: x53 = 34; INSTRUCTIONS: x54 = 35; INSTRUCTIONS: x55 = 36; INSTRUCTIONS: x56 = 37; INSTRUCTIONS: x57 = 38; INSTRUCTIONS: x58 = 39; INSTRUCTIONS: x59 = 40; INSTRUCTIONS: x60 = 41; INSTRUCTIONS: x61 = 42; INSTRUCTIONS: x62 = 43; INSTRUCTIONS: x63 = 44; INSTRUCTIONS: x64 = 45; INSTRUCTIONS: x65 = 46; INSTRUCTIONS: x66 = 47; INSTRUCTIONS: x67 = 48; INSTRUCTIONS: x68 = 49; INSTRUCTIONS: x69 = 50; INSTRUCTIONS: x70 = 51; INSTRUCTIONS: x71 = 52; INSTRUCTIONS: x72 = 53; array[INSTRUCTIONS] of string: INSTRUCTIONS_NAMES = ["x20", "x21", "x22", "x23", "x24", "x25", "x26", "x27", "x28", "x29", "x30", "x31", "x32", "x33", "x34", "x35", "x36", "x37", "x38", "x39", "x40", "x41", "x42", "x43", "x44", "x45", "x46", "x47", "x48", "x49", "x50", "x51", "x52", "x53", "x54", "x55", "x56", "x57", "x58", "x59", "x60", "x61", "x62", "x63", "x64", "x65", "x66", "x67", "x68", "x69", "x70", "x71", "x72"];
set of int: OPS = 1..6;
OPS: SHL = 1; OPS: AND = 2; OPS: ADD = 3; OPS: MUL = 4; OPS: OR = 5; OPS: SHR = 6; array[OPS] of string: OPS_NAMES = ["SHL", "AND", "ADD", "MUL", "OR", "SHR"];
int: MAX_LOC = 183;
set of int: LOCATIONS = 1..MAX_LOC;
array[INSTRUCTIONS] of var LOCATIONS: output_locations;
array[INSTRUCTIONS] of var int: output_data_latency;
array[INSTRUCTIONS] of var int: output_core_latency;
array[INSTRUCTIONS] of var CORE: output_cores;
array[INSTRUCTIONS] of OPS: input_ops = [MUL, MUL, MUL, ADD, MUL, MUL, ADD, MUL, ADD, MUL, MUL, ADD, MUL, ADD, MUL, ADD, MUL, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, MUL, MUL, MUL, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, MUL, ADD, SHL, AND, ADD, SHL];
array[INSTRUCTIONS] of var int: output_ADD_MUL_core_latency;
array[INSTRUCTIONS] of var 0..1: output_ADD_MUL_core_use;
constraint forall (i in INSTRUCTIONS) (0 <= output_ADD_MUL_core_latency[i]);
constraint forall (i in INSTRUCTIONS) (output_ADD_MUL_core_use[i] == 1 -> output_core_latency[i] == output_ADD_MUL_core_latency[i]);
array[INSTRUCTIONS] of var int: output_MUL_CORE_core_latency;
array[INSTRUCTIONS] of var 0..1: output_MUL_CORE_core_use;
constraint forall (i in INSTRUCTIONS) (0 <= output_MUL_CORE_core_latency[i]);
constraint forall (i in INSTRUCTIONS) (output_MUL_CORE_core_use[i] == 1 -> output_core_latency[i] == output_MUL_CORE_core_latency[i]);
array[INSTRUCTIONS] of var int: output_LEA_BW_core_latency;
array[INSTRUCTIONS] of var 0..1: output_LEA_BW_core_use;
constraint forall (i in INSTRUCTIONS) (0 <= output_LEA_BW_core_latency[i]);
constraint forall (i in INSTRUCTIONS) (output_LEA_BW_core_use[i] == 1 -> output_core_latency[i] == output_LEA_BW_core_latency[i]);
var LOCATIONS: RET_loc;
constraint alldifferent(output_locations);
constraint output_locations[x21] + output_data_latency[x21] <= output_locations[x23];
constraint output_locations[x22] + output_data_latency[x22] <= output_locations[x23];
constraint output_locations[x24] + output_data_latency[x24] <= output_locations[x26];
constraint output_locations[x25] + output_data_latency[x25] <= output_locations[x26];
constraint output_locations[x26] + output_data_latency[x26] <= output_locations[x28];
constraint output_locations[x27] + output_data_latency[x27] <= output_locations[x28];
constraint output_locations[x29] + output_data_latency[x29] <= output_locations[x31];
constraint output_locations[x30] + output_data_latency[x30] <= output_locations[x31];
constraint output_locations[x31] + output_data_latency[x31] <= output_locations[x33];
constraint output_locations[x32] + output_data_latency[x32] <= output_locations[x33];
constraint output_locations[x33] + output_data_latency[x33] <= output_locations[x35];
constraint output_locations[x34] + output_data_latency[x34] <= output_locations[x35];
constraint output_locations[x36] + output_data_latency[x36] <= output_locations[x38];
constraint output_locations[x37] + output_data_latency[x37] <= output_locations[x38];
constraint output_locations[x38] + output_data_latency[x38] <= output_locations[x40];
constraint output_locations[x39] + output_data_latency[x39] <= output_locations[x40];
constraint output_locations[x40] + output_data_latency[x40] <= output_locations[x42];
constraint output_locations[x41] + output_data_latency[x41] <= output_locations[x42];
constraint output_locations[x42] + output_data_latency[x42] <= output_locations[x44];
constraint output_locations[x43] + output_data_latency[x43] <= output_locations[x44];
constraint output_locations[x45] + output_data_latency[x45] <= output_locations[x49];
constraint output_locations[x20] + output_data_latency[x20] <= output_locations[x50];
constraint output_locations[x49] + output_data_latency[x49] <= output_locations[x50];
constraint output_locations[x46] + output_data_latency[x46] <= output_locations[x51];
constraint output_locations[x50] + output_data_latency[x50] <= output_locations[x52];
constraint output_locations[x51] + output_data_latency[x51] <= output_locations[x52];
constraint output_locations[x47] + output_data_latency[x47] <= output_locations[x53];
constraint output_locations[x52] + output_data_latency[x52] <= output_locations[x54];
constraint output_locations[x53] + output_data_latency[x53] <= output_locations[x54];
constraint output_locations[x48] + output_data_latency[x48] <= output_locations[x55];
constraint output_locations[x54] + output_data_latency[x54] <= output_locations[x56];
constraint output_locations[x55] + output_data_latency[x55] <= output_locations[x56];
constraint output_locations[x45] + output_data_latency[x45] <= output_locations[x57];
constraint output_locations[x23] + output_data_latency[x23] <= output_locations[x58];
constraint output_locations[x57] + output_data_latency[x57] <= output_locations[x58];
constraint output_locations[x47] + output_data_latency[x47] <= output_locations[x59];
constraint output_locations[x58] + output_data_latency[x58] <= output_locations[x60];
constraint output_locations[x59] + output_data_latency[x59] <= output_locations[x60];
constraint output_locations[x48] + output_data_latency[x48] <= output_locations[x61];
constraint output_locations[x60] + output_data_latency[x60] <= output_locations[x62];
constraint output_locations[x61] + output_data_latency[x61] <= output_locations[x62];
constraint output_locations[x45] + output_data_latency[x45] <= output_locations[x63];
constraint output_locations[x28] + output_data_latency[x28] <= output_locations[x64];
constraint output_locations[x63] + output_data_latency[x63] <= output_locations[x64];
constraint output_locations[x48] + output_data_latency[x48] <= output_locations[x65];
constraint output_locations[x64] + output_data_latency[x64] <= output_locations[x66];
constraint output_locations[x65] + output_data_latency[x65] <= output_locations[x66];
constraint output_locations[x45] + output_data_latency[x45] <= output_locations[x67];
constraint output_locations[x35] + output_data_latency[x35] <= output_locations[x68];
constraint output_locations[x67] + output_data_latency[x67] <= output_locations[x68];
constraint output_locations[x56] + output_data_latency[x56] <= output_locations[x69];
constraint output_locations[x56] + output_data_latency[x56] <= output_locations[x70];
constraint output_locations[x69] + output_data_latency[x69] <= output_locations[x71];
constraint output_locations[x62] + output_data_latency[x62] <= output_locations[x71];
constraint output_locations[x71] + output_data_latency[x71] <= output_locations[x72];
constraint max([ output_locations[i] + output_data_latency[i] | i in INSTRUCTIONS ]) <= RET_loc;
constraint forall (i in INSTRUCTIONS) (input_ops[i] == SHL -> ((output_cores[i] == LEA_BW /\ output_LEA_BW_core_use[i] == 1 /\ output_LEA_BW_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0)));
constraint forall (i in INSTRUCTIONS) (input_ops[i] == AND -> ((output_cores[i] == LEA_BW /\ output_LEA_BW_core_use[i] == 1 /\ output_LEA_BW_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0)));
constraint forall (i in INSTRUCTIONS) (input_ops[i] == ADD -> ((output_cores[i] == ADD_MUL /\ output_ADD_MUL_core_use[i] == 1 /\ output_ADD_MUL_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0 /\ output_LEA_BW_core_use[i] == 0 /\ output_LEA_BW_core_latency[i] == 0) \/ (output_cores[i] == LEA_BW /\ output_LEA_BW_core_use[i] == 1 /\ output_LEA_BW_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0)));
constraint forall (i in INSTRUCTIONS) (input_ops[i] == MUL -> ((output_cores[i] == ADD_MUL /\ output_ADD_MUL_core_use[i] == 1 /\ output_ADD_MUL_core_latency[i] == 4 /\ output_data_latency[i] == 12 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0 /\ output_LEA_BW_core_use[i] == 0 /\ output_LEA_BW_core_latency[i] == 0) \/ (output_cores[i] == MUL_CORE /\ output_MUL_CORE_core_use[i] == 1 /\ output_MUL_CORE_core_latency[i] == 4 /\ output_data_latency[i] == 12 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_LEA_BW_core_use[i] == 0 /\ output_LEA_BW_core_latency[i] == 0)));
constraint forall (i in INSTRUCTIONS) (input_ops[i] == OR -> ((output_cores[i] == LEA_BW /\ output_LEA_BW_core_use[i] == 1 /\ output_LEA_BW_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0)));
constraint forall (i in INSTRUCTIONS) (input_ops[i] == SHR -> ((output_cores[i] == LEA_BW /\ output_LEA_BW_core_use[i] == 1 /\ output_LEA_BW_core_latency[i] == 4 /\ output_data_latency[i] == 4 /\ output_ADD_MUL_core_use[i] == 0 /\ output_ADD_MUL_core_latency[i] == 0 /\ output_MUL_CORE_core_use[i] == 0 /\ output_MUL_CORE_core_latency[i] == 0)));
constraint cumulative(output_locations, output_ADD_MUL_core_latency, output_ADD_MUL_core_use, 2);
constraint cumulative(output_locations, output_MUL_CORE_core_latency, output_MUL_CORE_core_use, 1);
constraint cumulative(output_locations, output_LEA_BW_core_latency, output_LEA_BW_core_use, 2);
solve minimize RET_loc;
output [ "(" ++ show(INSTRUCTIONS_NAMES[i]) ++ ", " ++ show(CORE_NAMES[fix(output_cores[i])]) ++ ", " ++ show(output_locations[i]) ++ ", " ++ show(output_data_latency[i]) ++ ", " ++ show(output_core_latency[i]) ++ ") ,\n"
| i in INSTRUCTIONS ];
output [ "RET_loc: " ++ show(RET_loc) ];
|
