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
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vp2intersect -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,X86
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vp2intersect -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,X64
; Test with more than four live mask pairs
define void @test(<16 x i32> %a0, <16 x i32> %b0, <16 x i32> %a1, <16 x i32> %b1, <16 x i32> %a2, <16 x i32> %b2, <16 x i32> %a3, <16 x i32> %b3, <16 x i32> %a4, <16 x i32> %b4, i16* nocapture %m0, i16* nocapture %m1) nounwind {
; X86-LABEL: test:
; X86: # %bb.0: # %entry
; X86-NEXT: pushl %ebp
; X86-NEXT: movl %esp, %ebp
; X86-NEXT: pushl %edi
; X86-NEXT: pushl %esi
; X86-NEXT: andl $-64, %esp
; X86-NEXT: subl $64, %esp
; X86-NEXT: movl 456(%ebp), %esi
; X86-NEXT: vmovaps 328(%ebp), %zmm3
; X86-NEXT: vmovaps 200(%ebp), %zmm4
; X86-NEXT: vmovaps 72(%ebp), %zmm5
; X86-NEXT: vp2intersectd %zmm1, %zmm0, %k0
; X86-NEXT: kmovw %k0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill
; X86-NEXT: kmovw %k1, {{[0-9]+}}(%esp)
; X86-NEXT: vp2intersectd 8(%ebp), %zmm2, %k0
; X86-NEXT: kmovw %k0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill
; X86-NEXT: kmovw %k1, {{[0-9]+}}(%esp)
; X86-NEXT: vp2intersectd 136(%ebp), %zmm5, %k0
; X86-NEXT: kmovw %k0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill
; X86-NEXT: kmovw %k1, {{[0-9]+}}(%esp)
; X86-NEXT: vp2intersectd 264(%ebp), %zmm4, %k0
; X86-NEXT: kmovw %k0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill
; X86-NEXT: kmovw %k1, {{[0-9]+}}(%esp)
; X86-NEXT: vp2intersectd 392(%ebp), %zmm3, %k0
; X86-NEXT: kmovw %k0, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill
; X86-NEXT: kmovw %k1, {{[0-9]+}}(%esp)
; X86-NEXT: vzeroupper
; X86-NEXT: calll dummy
; X86-NEXT: kmovw {{[-0-9]+}}(%e{{[sb]}}p), %k0 # 4-byte Folded Reload
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1
; X86-NEXT: kmovw %k0, %eax
; X86-NEXT: kmovw {{[-0-9]+}}(%e{{[sb]}}p), %k0 # 4-byte Folded Reload
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1
; X86-NEXT: kmovw %k0, %ecx
; X86-NEXT: kmovw {{[-0-9]+}}(%e{{[sb]}}p), %k0 # 4-byte Folded Reload
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1
; X86-NEXT: kmovw %k0, %edx
; X86-NEXT: kmovw {{[-0-9]+}}(%e{{[sb]}}p), %k0 # 4-byte Folded Reload
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1
; X86-NEXT: kmovw %k0, %edi
; X86-NEXT: addl %edi, %eax
; X86-NEXT: kmovw {{[-0-9]+}}(%e{{[sb]}}p), %k2 # 4-byte Folded Reload
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k3
; X86-NEXT: kmovw %k2, %edi
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: kmovw %k1, %ecx
; X86-NEXT: addl %edi, %ecx
; X86-NEXT: addl %eax, %ecx
; X86-NEXT: addl %edx, %ecx
; X86-NEXT: movw %cx, (%esi)
; X86-NEXT: leal -8(%ebp), %esp
; X86-NEXT: popl %esi
; X86-NEXT: popl %edi
; X86-NEXT: popl %ebp
; X86-NEXT: retl
;
; X64-LABEL: test:
; X64: # %bb.0: # %entry
; X64-NEXT: pushq %rbp
; X64-NEXT: movq %rsp, %rbp
; X64-NEXT: pushq %r14
; X64-NEXT: pushq %rbx
; X64-NEXT: andq $-64, %rsp
; X64-NEXT: subq $64, %rsp
; X64-NEXT: movq %rdi, %r14
; X64-NEXT: vmovaps 16(%rbp), %zmm8
; X64-NEXT: vp2intersectd %zmm1, %zmm0, %k0
; X64-NEXT: kmovw %k0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
; X64-NEXT: kmovw %k1, {{[0-9]+}}(%rsp)
; X64-NEXT: vp2intersectd %zmm3, %zmm2, %k0
; X64-NEXT: kmovw %k0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
; X64-NEXT: kmovw %k1, {{[0-9]+}}(%rsp)
; X64-NEXT: vp2intersectd %zmm5, %zmm4, %k0
; X64-NEXT: kmovw %k0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
; X64-NEXT: kmovw %k1, {{[0-9]+}}(%rsp)
; X64-NEXT: vp2intersectd %zmm7, %zmm6, %k0
; X64-NEXT: kmovw %k0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
; X64-NEXT: kmovw %k1, {{[0-9]+}}(%rsp)
; X64-NEXT: vp2intersectd 80(%rbp), %zmm8, %k0
; X64-NEXT: kmovw %k0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
; X64-NEXT: kmovw %k1, {{[0-9]+}}(%rsp)
; X64-NEXT: vzeroupper
; X64-NEXT: callq dummy
; X64-NEXT: kmovw {{[-0-9]+}}(%r{{[sb]}}p), %k0 # 4-byte Folded Reload
; X64-NEXT: kmovw {{[0-9]+}}(%rsp), %k1
; X64-NEXT: kmovw %k0, %eax
; X64-NEXT: kmovw {{[-0-9]+}}(%r{{[sb]}}p), %k0 # 4-byte Folded Reload
; X64-NEXT: kmovw {{[0-9]+}}(%rsp), %k1
; X64-NEXT: kmovw %k0, %ecx
; X64-NEXT: kmovw {{[-0-9]+}}(%r{{[sb]}}p), %k0 # 4-byte Folded Reload
; X64-NEXT: kmovw {{[0-9]+}}(%rsp), %k1
; X64-NEXT: kmovw %k0, %edx
; X64-NEXT: kmovw {{[-0-9]+}}(%r{{[sb]}}p), %k0 # 4-byte Folded Reload
; X64-NEXT: kmovw {{[0-9]+}}(%rsp), %k1
; X64-NEXT: kmovw %k0, %esi
; X64-NEXT: kmovw {{[-0-9]+}}(%r{{[sb]}}p), %k0 # 4-byte Folded Reload
; X64-NEXT: kmovw {{[0-9]+}}(%rsp), %k1
; X64-NEXT: kmovw %k0, %edi
; X64-NEXT: kmovw %k1, %ebx
; X64-NEXT: addl %edi, %eax
; X64-NEXT: addl %ecx, %edx
; X64-NEXT: leal (%rbx,%rsi), %ecx
; X64-NEXT: addl %eax, %ecx
; X64-NEXT: addl %edx, %ecx
; X64-NEXT: movw %cx, (%r14)
; X64-NEXT: leaq -16(%rbp), %rsp
; X64-NEXT: popq %rbx
; X64-NEXT: popq %r14
; X64-NEXT: popq %rbp
; X64-NEXT: retq
entry:
%0 = call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %a0, <16 x i32> %b0)
%1 = call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %a1, <16 x i32> %b1)
%2 = call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %a2, <16 x i32> %b2)
%3 = call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %a3, <16 x i32> %b3)
%4 = call { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32> %a4, <16 x i32> %b4)
%5 = extractvalue { <16 x i1>, <16 x i1> } %0, 0
%6 = extractvalue { <16 x i1>, <16 x i1> } %1, 0
%7 = extractvalue { <16 x i1>, <16 x i1> } %2, 0
%8 = extractvalue { <16 x i1>, <16 x i1> } %3, 0
%9 = extractvalue { <16 x i1>, <16 x i1> } %4, 0
%10 = extractvalue { <16 x i1>, <16 x i1> } %0, 1
%11 = extractvalue { <16 x i1>, <16 x i1> } %1, 1
call void @dummy()
%12 = bitcast <16 x i1> %5 to i16
%13 = bitcast <16 x i1> %6 to i16
%14 = bitcast <16 x i1> %7 to i16
%15 = bitcast <16 x i1> %8 to i16
%16 = bitcast <16 x i1> %9 to i16
%17 = bitcast <16 x i1> %10 to i16
%18 = bitcast <16 x i1> %11 to i16
%19 = add i16 %12, %13
%20 = add i16 %14, %15
%21 = add i16 %16, %17
%22 = add i16 %19, %21
%23 = add i16 %22, %20
store i16 %23, i16* %m0, align 16
ret void
}
declare { <16 x i1>, <16 x i1> } @llvm.x86.avx512.vp2intersect.d.512(<16 x i32>, <16 x i32>)
declare void @dummy()
|