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
| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 | FileCheck %s
; More than one 'arcp' division using a single divisor operand
; should be converted into a reciprocal and multiplication.
; Don't do anything for just one division.
define float @div1_arcp(float %x, float %y, float %z) {
; CHECK-LABEL: div1_arcp:
; CHECK: # %bb.0:
; CHECK-NEXT: divss %xmm1, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv arcp float %x, %y
ret float %div1
}
; All math instructions are 'arcp', so optimize.
define float @div2_arcp_all(float %x, float %y, float %z) {
; CHECK-LABEL: div2_arcp_all:
; CHECK: # %bb.0:
; CHECK-NEXT: movss {{.*#+}} xmm3 = mem[0],zero,zero,zero
; CHECK-NEXT: divss %xmm2, %xmm3
; CHECK-NEXT: mulss %xmm3, %xmm0
; CHECK-NEXT: mulss %xmm1, %xmm0
; CHECK-NEXT: mulss %xmm3, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv arcp float %x, %z
%mul = fmul arcp float %div1, %y
%div2 = fdiv arcp float %mul, %z
ret float %div2
}
; The first division is not 'arcp', so do not optimize.
define float @div2_arcp_partial1(float %x, float %y, float %z) {
; CHECK-LABEL: div2_arcp_partial1:
; CHECK: # %bb.0:
; CHECK-NEXT: divss %xmm2, %xmm0
; CHECK-NEXT: mulss %xmm1, %xmm0
; CHECK-NEXT: divss %xmm2, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv float %x, %z
%mul = fmul arcp float %div1, %y
%div2 = fdiv arcp float %mul, %z
ret float %div2
}
; The second division is not 'arcp', so do not optimize.
define float @div2_arcp_partial2(float %x, float %y, float %z) {
; CHECK-LABEL: div2_arcp_partial2:
; CHECK: # %bb.0:
; CHECK-NEXT: divss %xmm2, %xmm0
; CHECK-NEXT: mulss %xmm1, %xmm0
; CHECK-NEXT: divss %xmm2, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv arcp float %x, %z
%mul = fmul arcp float %div1, %y
%div2 = fdiv float %mul, %z
ret float %div2
}
; The multiply is not 'arcp', but that does not prevent optimizing the divisions.
define float @div2_arcp_partial3(float %x, float %y, float %z) {
; CHECK-LABEL: div2_arcp_partial3:
; CHECK: # %bb.0:
; CHECK-NEXT: movss {{.*#+}} xmm3 = mem[0],zero,zero,zero
; CHECK-NEXT: divss %xmm2, %xmm3
; CHECK-NEXT: mulss %xmm3, %xmm0
; CHECK-NEXT: mulss %xmm1, %xmm0
; CHECK-NEXT: mulss %xmm3, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv arcp float %x, %z
%mul = fmul float %div1, %y
%div2 = fdiv arcp float %mul, %z
ret float %div2
}
; If the reciprocal is already calculated, we should not
; generate an extra multiplication by 1.0.
define double @div3_arcp(double %x, double %y, double %z) {
; CHECK-LABEL: div3_arcp:
; CHECK: # %bb.0:
; CHECK-NEXT: movsd {{.*#+}} xmm2 = mem[0],zero
; CHECK-NEXT: divsd %xmm1, %xmm2
; CHECK-NEXT: mulsd %xmm2, %xmm0
; CHECK-NEXT: addsd %xmm2, %xmm0
; CHECK-NEXT: retq
%div1 = fdiv fast double 1.0, %y
%div2 = fdiv fast double %x, %y
%ret = fadd fast double %div2, %div1
ret double %ret
}
define float @div_select_constant_fold(i1 zeroext %arg) {
; CHECK-LABEL: div_select_constant_fold:
; CHECK: # %bb.0:
; CHECK-NEXT: testl %edi, %edi
; CHECK-NEXT: jne .LBB6_1
; CHECK-NEXT: # %bb.2:
; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: retq
; CHECK-NEXT: .LBB6_1:
; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: retq
%tmp = select i1 %arg, float 5.000000e+00, float 6.000000e+00
%B2 = fdiv nnan float %tmp, 2.000000e+00
ret float %B2
}
define float @div_select_constant_fold_zero(i1 zeroext %arg) {
; CHECK-LABEL: div_select_constant_fold_zero:
; CHECK: # %bb.0:
; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: retq
%tmp = select i1 %arg, float 5.000000e+00, float 6.000000e+00
%B2 = fdiv float %tmp, 0.000000e+00
ret float %B2
}
define void @PR24141() {
; CHECK-LABEL: PR24141:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: pushq %rax
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: # implicit-def: $xmm0
; CHECK-NEXT: .p2align 4, 0x90
; CHECK-NEXT: .LBB8_1: # %while.body
; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
; CHECK-NEXT: callq g
; CHECK-NEXT: divsd %xmm1, %xmm0
; CHECK-NEXT: jmp .LBB8_1
entry:
br label %while.body
while.body:
%x.0 = phi double [ undef, %entry ], [ %div, %while.body ]
%call = call { double, double } @g(double %x.0)
%xv0 = extractvalue { double, double } %call, 0
%xv1 = extractvalue { double, double } %call, 1
%div = fdiv arcp double %xv0, %xv1
br label %while.body
}
declare { double, double } @g(double)
|