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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=r600 -mcpu=redwood | FileCheck --check-prefix=EG %s
define amdgpu_vs void @main(<4 x float> inreg %reg0, <4 x float> inreg %reg1) {
; EG-LABEL: main:
; EG: ; %bb.0: ; %main_body
; EG-NEXT: CALL_FS
; EG-NEXT: ALU 3, @8, KC0[CB0:0-32], KC1[]
; EG-NEXT: EXPORT T1.XYZW
; EG-NEXT: EXPORT T0.XYXX
; EG-NEXT: EXPORT T0.ZXXX
; EG-NEXT: EXPORT T0.XXWX
; EG-NEXT: EXPORT T0.XXXW
; EG-NEXT: CF_END
; EG-NEXT: ALU clause starting at 8:
; EG-NEXT: MOV * T0.W, KC0[0].X,
; EG-NEXT: MUL_IEEE T0.X, KC0[0].X, 0.0,
; EG-NEXT: MUL_IEEE T0.Y, KC0[0].Y, KC0[0].X,
; EG-NEXT: MUL_IEEE * T0.Z, KC0[0].Z, KC0[0].X,
main_body:
%0 = extractelement <4 x float> %reg1, i32 0
%1 = extractelement <4 x float> %reg1, i32 1
%2 = extractelement <4 x float> %reg1, i32 2
%3 = extractelement <4 x float> %reg1, i32 3
%4 = load <4 x float>, ptr addrspace(8) null
%5 = extractelement <4 x float> %4, i32 1
%6 = load <4 x float>, ptr addrspace(8) null
%7 = extractelement <4 x float> %6, i32 2
%8 = load <4 x float>, ptr addrspace(8) null
%9 = extractelement <4 x float> %8, i32 0
%10 = fmul float 0.000000e+00, %9
%11 = load <4 x float>, ptr addrspace(8) null
%12 = extractelement <4 x float> %11, i32 0
%13 = fmul float %5, %12
%14 = load <4 x float>, ptr addrspace(8) null
%15 = extractelement <4 x float> %14, i32 0
%16 = fmul float 0.000000e+00, %15
%17 = load <4 x float>, ptr addrspace(8) null
%18 = extractelement <4 x float> %17, i32 0
%19 = fmul float 0.000000e+00, %18
%20 = load <4 x float>, ptr addrspace(8) null
%21 = extractelement <4 x float> %20, i32 0
%22 = fmul float %7, %21
%23 = load <4 x float>, ptr addrspace(8) null
%24 = extractelement <4 x float> %23, i32 0
%25 = fmul float 0.000000e+00, %24
%26 = load <4 x float>, ptr addrspace(8) null
%27 = extractelement <4 x float> %26, i32 0
%28 = fmul float 0.000000e+00, %27
%29 = load <4 x float>, ptr addrspace(8) null
%30 = extractelement <4 x float> %29, i32 0
%31 = fmul float 0.000000e+00, %30
%32 = load <4 x float>, ptr addrspace(8) null
%33 = extractelement <4 x float> %32, i32 0
%34 = fmul float 0.000000e+00, %33
%35 = load <4 x float>, ptr addrspace(8) null
%36 = extractelement <4 x float> %35, i32 0
%37 = fmul float 0.000000e+00, %36
%38 = load <4 x float>, ptr addrspace(8) null
%39 = extractelement <4 x float> %38, i32 0
%40 = fmul float 1.000000e+00, %39
%41 = load <4 x float>, ptr addrspace(8) null
%42 = extractelement <4 x float> %41, i32 0
%43 = fmul float 0.000000e+00, %42
%44 = load <4 x float>, ptr addrspace(8) null
%45 = extractelement <4 x float> %44, i32 0
%46 = fmul float 0.000000e+00, %45
%47 = load <4 x float>, ptr addrspace(8) null
%48 = extractelement <4 x float> %47, i32 0
%49 = fmul float 0.000000e+00, %48
%50 = load <4 x float>, ptr addrspace(8) null
%51 = extractelement <4 x float> %50, i32 0
%52 = fmul float 0.000000e+00, %51
%53 = load <4 x float>, ptr addrspace(8) null
%54 = extractelement <4 x float> %53, i32 0
%55 = fmul float 1.000000e+00, %54
%56 = insertelement <4 x float> poison, float %0, i32 0
%57 = insertelement <4 x float> %56, float %1, i32 1
%58 = insertelement <4 x float> %57, float %2, i32 2
%59 = insertelement <4 x float> %58, float %3, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %59, i32 60, i32 1)
%60 = insertelement <4 x float> poison, float %10, i32 0
%61 = insertelement <4 x float> %60, float %13, i32 1
%62 = insertelement <4 x float> %61, float %16, i32 2
%63 = insertelement <4 x float> %62, float %19, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %63, i32 0, i32 2)
%64 = insertelement <4 x float> poison, float %22, i32 0
%65 = insertelement <4 x float> %64, float %25, i32 1
%66 = insertelement <4 x float> %65, float %28, i32 2
%67 = insertelement <4 x float> %66, float %31, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %67, i32 1, i32 2)
%68 = insertelement <4 x float> poison, float %34, i32 0
%69 = insertelement <4 x float> %68, float %37, i32 1
%70 = insertelement <4 x float> %69, float %40, i32 2
%71 = insertelement <4 x float> %70, float %43, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %71, i32 2, i32 2)
%72 = insertelement <4 x float> poison, float %46, i32 0
%73 = insertelement <4 x float> %72, float %49, i32 1
%74 = insertelement <4 x float> %73, float %52, i32 2
%75 = insertelement <4 x float> %74, float %55, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %75, i32 3, i32 2)
ret void
}
define amdgpu_vs void @main2(<4 x float> inreg %reg0, <4 x float> inreg %reg1) {
; EG-LABEL: main2:
; EG: ; %bb.0: ; %main_body
; EG-NEXT: CALL_FS
; EG-NEXT: ALU 11, @6, KC0[CB0:0-32], KC1[]
; EG-NEXT: EXPORT T1.XY__
; EG-NEXT: EXPORT T0.ZXY0
; EG-NEXT: CF_END
; EG-NEXT: PAD
; EG-NEXT: ALU clause starting at 6:
; EG-NEXT: MULADD_IEEE * T0.W, KC0[1].X, literal.x, 0.5,
; EG-NEXT: 1042479491(1.591549e-01), 0(0.000000e+00)
; EG-NEXT: FRACT * T0.W, PV.W,
; EG-NEXT: ADD * T0.W, PV.W, literal.x,
; EG-NEXT: -1090519040(-5.000000e-01), 0(0.000000e+00)
; EG-NEXT: COS * T0.Z, PV.W,
; EG-NEXT: ADD T1.X, T1.X, literal.x,
; EG-NEXT: MOV T0.Y, KC0[0].Y,
; EG-NEXT: MOV * T0.X, KC0[0].X,
; EG-NEXT: 1075838976(2.500000e+00), 0(0.000000e+00)
; EG-NEXT: MUL_IEEE * T1.Y, T1.Y, literal.x,
; EG-NEXT: 1080033280(3.500000e+00), 0(0.000000e+00)
main_body:
%0 = extractelement <4 x float> %reg1, i32 0
%1 = extractelement <4 x float> %reg1, i32 1
%2 = fadd float %0, 2.5
%3 = fmul float %1, 3.5
%4 = load <4 x float>, ptr addrspace(8) getelementptr ([1024 x <4 x float>], ptr addrspace(8) null, i64 0, i32 1)
%5 = extractelement <4 x float> %4, i32 0
%6 = call float @llvm.cos.f32(float %5)
%7 = load <4 x float>, ptr addrspace(8) null
%8 = extractelement <4 x float> %7, i32 0
%9 = load <4 x float>, ptr addrspace(8) null
%10 = extractelement <4 x float> %9, i32 1
%11 = insertelement <4 x float> poison, float %2, i32 0
%12 = insertelement <4 x float> %11, float %3, i32 1
call void @llvm.r600.store.swizzle(<4 x float> %12, i32 60, i32 1)
%13 = insertelement <4 x float> poison, float %6, i32 0
%14 = insertelement <4 x float> %13, float %8, i32 1
%15 = insertelement <4 x float> %14, float %10, i32 2
%16 = insertelement <4 x float> %15, float 0.000000e+00, i32 3
call void @llvm.r600.store.swizzle(<4 x float> %16, i32 0, i32 2)
ret void
}
; Function Attrs: nounwind readonly
declare float @llvm.cos.f32(float) #1
declare void @llvm.r600.store.swizzle(<4 x float>, i32, i32)
attributes #1 = { nounwind readonly }
|