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
|
; RUN: llc -march=mipsel -mcpu=mips32r5 -mattr=+fp64,+msa,-nooddspreg \
; RUN: -verify-machineinstrs -no-integrated-as -relocation-model=pic < %s | \
; RUN: FileCheck %s -check-prefixes=ALL,ODDSPREG
; RUN: llc -march=mipsel -mcpu=mips32r5 -mattr=+fp64,+msa,+nooddspreg \
; RUN: -verify-machineinstrs -no-integrated-as -relocation-model=pic < %s | \
; RUN: FileCheck %s -check-prefixes=ALL,NOODDSPREG
@v4f32 = global <4 x float> zeroinitializer
define void @msa_insert_0(float %a) {
entry:
; Force the float into an odd-numbered register using named registers and
; load the vector.
%b = call float asm sideeffect "mov.s $0, $1", "={$f13},{$f12}" (float %a)
%0 = load volatile <4 x float>, <4 x float>* @v4f32
; Clobber all except $f12/$w12 and $f13
;
; The intention is that if odd single precision registers are permitted, the
; allocator will choose $f12/$w12 for the vector and $f13 for the float to
; avoid the spill/reload.
;
; On the other hand, if odd single precision registers are not permitted, it
; must copy $f13 to an even-numbered register before inserting into the
; vector.
call void asm sideeffect "teqi $$zero, 1", "~{$f0},~{$f1},~{$f2},~{$f3},~{$f4},~{$f5},~{$f6},~{$f7},~{$f8},~{$f9},~{$f10},~{$f11},~{$f14},~{$f15},~{$f16},~{$f17},~{$f18},~{$f19},~{$f20},~{$f21},~{$f22},~{$f23},~{$f24},~{$f25},~{$f26},~{$f27},~{$f28},~{$f29},~{$f30},~{$f31}"()
%1 = insertelement <4 x float> %0, float %b, i32 0
store <4 x float> %1, <4 x float>* @v4f32
ret void
}
; ALL-LABEL: msa_insert_0:
; ALL: mov.s $f13, $f12
; NOODDSPREG: mov.s $f[[F0:[0-9]+]], $f13
; ALL: lw $[[R0:[0-9]+]], %got(v4f32)(
; ALL: ld.w $w[[W0:[0-9]+]], 0($[[R0]])
; NOODDSPREG: insve.w $w[[W0]][0], $w[[F0]][0]
; ODDSPREG: insve.w $w[[W0]][0], $w13[0]
; ALL: teqi $zero, 1
; ALL-NOT: sdc1
; ALL-NOT: ldc1
; ALL: st.w $w[[W0]], 0($[[R0]])
define void @msa_insert_1(float %a) {
entry:
; Force the float into an odd-numbered register using named registers and
; load the vector.
%b = call float asm sideeffect "mov.s $0, $1", "={$f13},{$f12}" (float %a)
%0 = load volatile <4 x float>, <4 x float>* @v4f32
; Clobber all except $f12/$w12 and $f13
;
; The intention is that if odd single precision registers are permitted, the
; allocator will choose $f12/$w12 for the vector and $f13 for the float to
; avoid the spill/reload.
;
; On the other hand, if odd single precision registers are not permitted, it
; must copy $f13 to an even-numbered register before inserting into the
; vector.
call void asm sideeffect "teqi $$zero, 1", "~{$f0},~{$f1},~{$f2},~{$f3},~{$f4},~{$f5},~{$f6},~{$f7},~{$f8},~{$f9},~{$f10},~{$f11},~{$f14},~{$f15},~{$f16},~{$f17},~{$f18},~{$f19},~{$f20},~{$f21},~{$f22},~{$f23},~{$f24},~{$f25},~{$f26},~{$f27},~{$f28},~{$f29},~{$f30},~{$f31}"()
%1 = insertelement <4 x float> %0, float %b, i32 1
store <4 x float> %1, <4 x float>* @v4f32
ret void
}
; ALL-LABEL: msa_insert_1:
; ALL: mov.s $f13, $f12
; NOODDSPREG: mov.s $f[[F0:[0-9]+]], $f13
; ALL: lw $[[R0:[0-9]+]], %got(v4f32)(
; ALL: ld.w $w[[W0:[0-9]+]], 0($[[R0]])
; NOODDSPREG: insve.w $w[[W0]][1], $w[[F0]][0]
; ODDSPREG: insve.w $w[[W0]][1], $w13[0]
; ALL: teqi $zero, 1
; ALL-NOT: sdc1
; ALL-NOT: ldc1
; ALL: st.w $w[[W0]], 0($[[R0]])
define float @msa_extract_0() {
entry:
%0 = load volatile <4 x float>, <4 x float>* @v4f32
%1 = call <4 x float> asm sideeffect "move.v $0, $1", "={$w13},{$w12}" (<4 x float> %0)
; Clobber all except $f12, and $f13
;
; The intention is that if odd single precision registers are permitted, the
; allocator will choose $f13/$w13 for the vector since that saves on moves.
;
; On the other hand, if odd single precision registers are not permitted, it
; must move it to $f12/$w12.
call void asm sideeffect "teqi $$zero, 1", "~{$f0},~{$f1},~{$f2},~{$f3},~{$f4},~{$f5},~{$f6},~{$f7},~{$f8},~{$f9},~{$f10},~{$f11},~{$f14},~{$f15},~{$f16},~{$f17},~{$f18},~{$f19},~{$f20},~{$f21},~{$f22},~{$f23},~{$f24},~{$f25},~{$f26},~{$f27},~{$f28},~{$f29},~{$f30},~{$f31}"()
%2 = extractelement <4 x float> %1, i32 0
ret float %2
}
; ALL-LABEL: msa_extract_0:
; ALL: lw $[[R0:[0-9]+]], %got(v4f32)(
; ALL: ld.w $w12, 0($[[R0]])
; ALL: move.v $w[[W0:13]], $w12
; NOODDSPREG: move.v $w[[W0:12]], $w13
; ALL: teqi $zero, 1
; ALL-NOT: st.w
; ALL-NOT: ld.w
; ALL: mov.s $f0, $f[[W0]]
define float @msa_extract_1() {
entry:
%0 = load volatile <4 x float>, <4 x float>* @v4f32
%1 = call <4 x float> asm sideeffect "move.v $0, $1", "={$w13},{$w12}" (<4 x float> %0)
; Clobber all except $f13
;
; The intention is that if odd single precision registers are permitted, the
; allocator will choose $f13/$w13 for the vector since that saves on moves.
;
; On the other hand, if odd single precision registers are not permitted, it
; must be spilled.
call void asm sideeffect "teqi $$zero, 1", "~{$f0},~{$f1},~{$f2},~{$f3},~{$f4},~{$f5},~{$f6},~{$f7},~{$f8},~{$f9},~{$f10},~{$f11},~{$f12},~{$f14},~{$f15},~{$f16},~{$f17},~{$f18},~{$f19},~{$f20},~{$f21},~{$f22},~{$f23},~{$f24},~{$f25},~{$f26},~{$f27},~{$f28},~{$f29},~{$f30},~{$f31}"()
%2 = extractelement <4 x float> %1, i32 1
ret float %2
}
; ALL-LABEL: msa_extract_1:
; ALL: lw $[[R0:[0-9]+]], %got(v4f32)(
; ALL: ld.w $w12, 0($[[R0]])
; ALL: splati.w $w[[W0:[0-9]+]], $w13[1]
; NOODDSPREG: st.w $w[[W0]], 0($sp)
; ODDSPREG-NOT: st.w
; ODDSPREG-NOT: ld.w
; ALL: teqi $zero, 1
; ODDSPREG-NOT: st.w
; ODDSPREG-NOT: ld.w
; NOODDSPREG: ld.w $w0, 0($sp)
; ODDSPREG: mov.s $f0, $f[[W0]]
|
