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
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
|
class R600Reg <string name, bits<16> encoding> : Register<name> {
let Namespace = "AMDGPU";
let HWEncoding = encoding;
}
class R600RegWithChan <string name, bits<9> sel, string chan> :
Register <name> {
field bits<2> chan_encoding = !if(!eq(chan, "X"), 0,
!if(!eq(chan, "Y"), 1,
!if(!eq(chan, "Z"), 2,
!if(!eq(chan, "W"), 3, 0))));
let HWEncoding{8-0} = sel;
let HWEncoding{10-9} = chan_encoding;
let Namespace = "AMDGPU";
}
class R600Reg_128<string n, list<Register> subregs, bits<16> encoding> :
RegisterWithSubRegs<n, subregs> {
field bits<2> chan_encoding = 0;
let Namespace = "AMDGPU";
let SubRegIndices = [sub0, sub1, sub2, sub3];
let HWEncoding{8-0} = encoding{8-0};
let HWEncoding{10-9} = chan_encoding;
}
class R600Reg_64<string n, list<Register> subregs, bits<16> encoding> :
RegisterWithSubRegs<n, subregs> {
field bits<2> chan_encoding = 0;
let Namespace = "AMDGPU";
let SubRegIndices = [sub0, sub1];
let HWEncoding = encoding;
let HWEncoding{8-0} = encoding{8-0};
let HWEncoding{10-9} = chan_encoding;
}
class R600Reg_64Vertical<int lo, int hi, string chan> : R600Reg_64 <
"V"#lo#hi#"_"#chan,
[!cast<Register>("T"#lo#"_"#chan), !cast<Register>("T"#hi#"_"#chan)],
lo
>;
foreach Index = 0-127 in {
foreach Chan = [ "X", "Y", "Z", "W" ] in {
// 32-bit Temporary Registers
def T#Index#_#Chan : R600RegWithChan <"T"#Index#"."#Chan, Index, Chan>;
// Indirect addressing offset registers
def Addr#Index#_#Chan : R600RegWithChan <"T("#Index#" + AR.x)."#Chan,
Index, Chan>;
}
// 128-bit Temporary Registers
def T#Index#_XYZW : R600Reg_128 <"T"#Index#"",
[!cast<Register>("T"#Index#"_X"),
!cast<Register>("T"#Index#"_Y"),
!cast<Register>("T"#Index#"_Z"),
!cast<Register>("T"#Index#"_W")],
Index>;
def T#Index#_XY : R600Reg_64 <"T"#Index#"",
[!cast<Register>("T"#Index#"_X"),
!cast<Register>("T"#Index#"_Y")],
Index>;
}
foreach Chan = [ "X", "Y", "Z", "W"] in {
let chan_encoding = !if(!eq(Chan, "X"), 0,
!if(!eq(Chan, "Y"), 1,
!if(!eq(Chan, "Z"), 2,
!if(!eq(Chan, "W"), 3, 0)))) in {
def V0123_#Chan : R600Reg_128 <"V0123_"#Chan,
[!cast<Register>("T0_"#Chan),
!cast<Register>("T1_"#Chan),
!cast<Register>("T2_"#Chan),
!cast<Register>("T3_"#Chan)],
0>;
def V01_#Chan : R600Reg_64Vertical<0, 1, Chan>;
def V23_#Chan : R600Reg_64Vertical<2, 3, Chan>;
}
}
// KCACHE_BANK0
foreach Index = 159-128 in {
foreach Chan = [ "X", "Y", "Z", "W" ] in {
// 32-bit Temporary Registers
def KC0_#Index#_#Chan : R600RegWithChan <"KC0["#!add(Index,-128)#"]."#Chan, Index, Chan>;
}
// 128-bit Temporary Registers
def KC0_#Index#_XYZW : R600Reg_128 <"KC0["#!add(Index, -128)#"].XYZW",
[!cast<Register>("KC0_"#Index#"_X"),
!cast<Register>("KC0_"#Index#"_Y"),
!cast<Register>("KC0_"#Index#"_Z"),
!cast<Register>("KC0_"#Index#"_W")],
Index>;
}
// KCACHE_BANK1
foreach Index = 191-160 in {
foreach Chan = [ "X", "Y", "Z", "W" ] in {
// 32-bit Temporary Registers
def KC1_#Index#_#Chan : R600RegWithChan <"KC1["#!add(Index,-160)#"]."#Chan, Index, Chan>;
}
// 128-bit Temporary Registers
def KC1_#Index#_XYZW : R600Reg_128 <"KC1["#!add(Index, -160)#"].XYZW",
[!cast<Register>("KC1_"#Index#"_X"),
!cast<Register>("KC1_"#Index#"_Y"),
!cast<Register>("KC1_"#Index#"_Z"),
!cast<Register>("KC1_"#Index#"_W")],
Index>;
}
// Array Base Register holding input in FS
foreach Index = 448-480 in {
def ArrayBase#Index : R600Reg<"ARRAY_BASE", Index>;
}
// Special Registers
def OQA : R600Reg<"OQA", 219>;
def OQB : R600Reg<"OQB", 220>;
def OQAP : R600Reg<"OQAP", 221>;
def OQBP : R600Reg<"OQAP", 222>;
def LDS_DIRECT_A : R600Reg<"LDS_DIRECT_A", 223>;
def LDS_DIRECT_B : R600Reg<"LDS_DIRECT_B", 224>;
def ZERO : R600Reg<"0.0", 248>;
def ONE : R600Reg<"1.0", 249>;
def NEG_ONE : R600Reg<"-1.0", 249>;
def ONE_INT : R600Reg<"1", 250>;
def HALF : R600Reg<"0.5", 252>;
def NEG_HALF : R600Reg<"-0.5", 252>;
def ALU_LITERAL_X : R600RegWithChan<"literal.x", 253, "X">;
def ALU_LITERAL_Y : R600RegWithChan<"literal.y", 253, "Y">;
def ALU_LITERAL_Z : R600RegWithChan<"literal.z", 253, "Z">;
def ALU_LITERAL_W : R600RegWithChan<"literal.w", 253, "W">;
def PV_X : R600RegWithChan<"PV.X", 254, "X">;
def PV_Y : R600RegWithChan<"PV.Y", 254, "Y">;
def PV_Z : R600RegWithChan<"PV.Z", 254, "Z">;
def PV_W : R600RegWithChan<"PV.W", 254, "W">;
def PS: R600Reg<"PS", 255>;
def PREDICATE_BIT : R600Reg<"PredicateBit", 0>;
def PRED_SEL_OFF: R600Reg<"Pred_sel_off", 0>;
def PRED_SEL_ZERO : R600Reg<"Pred_sel_zero", 2>;
def PRED_SEL_ONE : R600Reg<"Pred_sel_one", 3>;
def AR_X : R600Reg<"AR.x", 0>;
def INDIRECT_BASE_ADDR : R600Reg <"INDIRECT_BASE_ADDR", 0>;
def R600_ArrayBase : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "ArrayBase%u", 448, 480))>;
// special registers for ALU src operands
// const buffer reference, SRCx_SEL contains index
def ALU_CONST : R600Reg<"CBuf", 0>;
// interpolation param reference, SRCx_SEL contains index
def ALU_PARAM : R600Reg<"Param", 0>;
let isAllocatable = 0 in {
def R600_Addr : RegisterClass <"AMDGPU", [i32], 32, (add (sequence "Addr%u_X", 0, 127))>;
// We only use Addr_[YZW] for vertical vectors.
// FIXME if we add more vertical vector registers we will need to ad more
// registers to these classes.
def R600_Addr_Y : RegisterClass <"AMDGPU", [i32], 32, (add Addr0_Y)>;
def R600_Addr_Z : RegisterClass <"AMDGPU", [i32], 32, (add Addr0_Z)>;
def R600_Addr_W : RegisterClass <"AMDGPU", [i32], 32, (add Addr0_W)>;
def R600_LDS_SRC_REG : RegisterClass<"AMDGPU", [i32], 32,
(add OQA, OQB, OQAP, OQBP, LDS_DIRECT_A, LDS_DIRECT_B)>;
def R600_KC0_X : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC0_%u_X", 128, 159))>;
def R600_KC0_Y : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC0_%u_Y", 128, 159))>;
def R600_KC0_Z : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC0_%u_Z", 128, 159))>;
def R600_KC0_W : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC0_%u_W", 128, 159))>;
def R600_KC0 : RegisterClass <"AMDGPU", [f32, i32], 32,
(interleave R600_KC0_X, R600_KC0_Y,
R600_KC0_Z, R600_KC0_W)>;
def R600_KC1_X : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC1_%u_X", 160, 191))>;
def R600_KC1_Y : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC1_%u_Y", 160, 191))>;
def R600_KC1_Z : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC1_%u_Z", 160, 191))>;
def R600_KC1_W : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "KC1_%u_W", 160, 191))>;
def R600_KC1 : RegisterClass <"AMDGPU", [f32, i32], 32,
(interleave R600_KC1_X, R600_KC1_Y,
R600_KC1_Z, R600_KC1_W)>;
} // End isAllocatable = 0
def R600_TReg32_X : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "T%u_X", 0, 127), AR_X)>;
def R600_TReg32_Y : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "T%u_Y", 0, 127))>;
def R600_TReg32_Z : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "T%u_Z", 0, 127))>;
def R600_TReg32_W : RegisterClass <"AMDGPU", [f32, i32], 32,
(add (sequence "T%u_W", 0, 127))>;
def R600_TReg32 : RegisterClass <"AMDGPU", [f32, i32], 32,
(interleave R600_TReg32_X, R600_TReg32_Y,
R600_TReg32_Z, R600_TReg32_W)>;
def R600_Reg32 : RegisterClass <"AMDGPU", [f32, i32], 32, (add
R600_TReg32,
R600_ArrayBase,
R600_Addr,
R600_KC0, R600_KC1,
ZERO, HALF, ONE, ONE_INT, PV_X, ALU_LITERAL_X, NEG_ONE, NEG_HALF,
ALU_CONST, ALU_PARAM, OQAP, INDIRECT_BASE_ADDR
)>;
def R600_Predicate : RegisterClass <"AMDGPU", [i32], 32, (add
PRED_SEL_OFF, PRED_SEL_ZERO, PRED_SEL_ONE)>;
def R600_Predicate_Bit: RegisterClass <"AMDGPU", [i32], 32, (add
PREDICATE_BIT)>;
def R600_Reg128 : RegisterClass<"AMDGPU", [v4f32, v4i32], 128,
(add (sequence "T%u_XYZW", 0, 127))> {
let CopyCost = -1;
}
def R600_Reg128Vertical : RegisterClass<"AMDGPU", [v4f32, v4i32], 128,
(add V0123_W, V0123_Z, V0123_Y, V0123_X)
>;
def R600_Reg64 : RegisterClass<"AMDGPU", [v2f32, v2i32, i64, f64], 64,
(add (sequence "T%u_XY", 0, 63))>;
def R600_Reg64Vertical : RegisterClass<"AMDGPU", [v2f32, v2i32], 64,
(add V01_X, V01_Y, V01_Z, V01_W,
V23_X, V23_Y, V23_Z, V23_W)>;
|