xref: /llvm-project/llvm/test/CodeGen/AArch64/urem-seteq-vec-nonzero.ll (revision cc82f1290a1e2157a6c0530d78d8cc84d2b8553d)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc -mtriple=aarch64-unknown-linux-gnu < %s | FileCheck %s
3
4define <4 x i1> @t32_3(<4 x i32> %X) nounwind {
5; CHECK-LABEL: t32_3:
6; CHECK:       // %bb.0:
7; CHECK-NEXT:    adrp x8, .LCPI0_0
8; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI0_0]
9; CHECK-NEXT:    mov w8, #43691 // =0xaaab
10; CHECK-NEXT:    movk w8, #43690, lsl #16
11; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
12; CHECK-NEXT:    dup v1.4s, w8
13; CHECK-NEXT:    adrp x8, .LCPI0_1
14; CHECK-NEXT:    mul v0.4s, v0.4s, v1.4s
15; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI0_1]
16; CHECK-NEXT:    cmhs v0.4s, v1.4s, v0.4s
17; CHECK-NEXT:    xtn v0.4h, v0.4s
18; CHECK-NEXT:    ret
19  %urem = urem <4 x i32> %X, <i32 3, i32 3, i32 3, i32 3>
20  %cmp = icmp eq <4 x i32> %urem, <i32 0, i32 1, i32 2, i32 2>
21  ret <4 x i1> %cmp
22}
23
24define <4 x i1> @t32_5(<4 x i32> %X) nounwind {
25; CHECK-LABEL: t32_5:
26; CHECK:       // %bb.0:
27; CHECK-NEXT:    adrp x8, .LCPI1_0
28; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI1_0]
29; CHECK-NEXT:    mov w8, #52429 // =0xcccd
30; CHECK-NEXT:    movk w8, #52428, lsl #16
31; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
32; CHECK-NEXT:    dup v1.4s, w8
33; CHECK-NEXT:    mov w8, #13106 // =0x3332
34; CHECK-NEXT:    movk w8, #13107, lsl #16
35; CHECK-NEXT:    mul v0.4s, v0.4s, v1.4s
36; CHECK-NEXT:    dup v1.4s, w8
37; CHECK-NEXT:    cmhs v0.4s, v1.4s, v0.4s
38; CHECK-NEXT:    xtn v0.4h, v0.4s
39; CHECK-NEXT:    ret
40  %urem = urem <4 x i32> %X, <i32 5, i32 5, i32 5, i32 5>
41  %cmp = icmp eq <4 x i32> %urem, <i32 1, i32 2, i32 3, i32 4>
42  ret <4 x i1> %cmp
43}
44
45define <4 x i1> @t32_6_part0(<4 x i32> %X) nounwind {
46; CHECK-LABEL: t32_6_part0:
47; CHECK:       // %bb.0:
48; CHECK-NEXT:    adrp x8, .LCPI2_0
49; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI2_0]
50; CHECK-NEXT:    mov w8, #43691 // =0xaaab
51; CHECK-NEXT:    movk w8, #43690, lsl #16
52; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
53; CHECK-NEXT:    dup v1.4s, w8
54; CHECK-NEXT:    mul v0.4s, v0.4s, v1.4s
55; CHECK-NEXT:    movi v1.16b, #170
56; CHECK-NEXT:    shl v2.4s, v0.4s, #31
57; CHECK-NEXT:    fneg v1.4s, v1.4s
58; CHECK-NEXT:    usra v2.4s, v0.4s, #1
59; CHECK-NEXT:    cmhs v0.4s, v1.4s, v2.4s
60; CHECK-NEXT:    xtn v0.4h, v0.4s
61; CHECK-NEXT:    ret
62  %urem = urem <4 x i32> %X, <i32 6, i32 6, i32 6, i32 6>
63  %cmp = icmp eq <4 x i32> %urem, <i32 0, i32 1, i32 2, i32 3>
64  ret <4 x i1> %cmp
65}
66
67define <4 x i1> @t32_6_part1(<4 x i32> %X) nounwind {
68; CHECK-LABEL: t32_6_part1:
69; CHECK:       // %bb.0:
70; CHECK-NEXT:    adrp x8, .LCPI3_0
71; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI3_0]
72; CHECK-NEXT:    mov w8, #43691 // =0xaaab
73; CHECK-NEXT:    movk w8, #43690, lsl #16
74; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
75; CHECK-NEXT:    dup v1.4s, w8
76; CHECK-NEXT:    adrp x8, .LCPI3_1
77; CHECK-NEXT:    mul v0.4s, v0.4s, v1.4s
78; CHECK-NEXT:    shl v1.4s, v0.4s, #31
79; CHECK-NEXT:    usra v1.4s, v0.4s, #1
80; CHECK-NEXT:    ldr q0, [x8, :lo12:.LCPI3_1]
81; CHECK-NEXT:    cmhs v0.4s, v0.4s, v1.4s
82; CHECK-NEXT:    xtn v0.4h, v0.4s
83; CHECK-NEXT:    ret
84  %urem = urem <4 x i32> %X, <i32 6, i32 6, i32 6, i32 6>
85  %cmp = icmp eq <4 x i32> %urem, <i32 4, i32 5, i32 0, i32 0>
86  ret <4 x i1> %cmp
87}
88
89define <4 x i1> @t32_tautological(<4 x i32> %X) nounwind {
90; CHECK-LABEL: t32_tautological:
91; CHECK:       // %bb.0:
92; CHECK-NEXT:    adrp x8, .LCPI4_0
93; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI4_0]
94; CHECK-NEXT:    mov w8, #43691 // =0xaaab
95; CHECK-NEXT:    movk w8, #43690, lsl #16
96; CHECK-NEXT:    sub v0.4s, v0.4s, v1.4s
97; CHECK-NEXT:    dup v1.4s, w8
98; CHECK-NEXT:    adrp x8, .LCPI4_1
99; CHECK-NEXT:    mul v0.4s, v0.4s, v1.4s
100; CHECK-NEXT:    ldr q1, [x8, :lo12:.LCPI4_1]
101; CHECK-NEXT:    cmhs v0.4s, v1.4s, v0.4s
102; CHECK-NEXT:    movi d1, #0x00ffffffff0000
103; CHECK-NEXT:    xtn v0.4h, v0.4s
104; CHECK-NEXT:    eor v0.8b, v0.8b, v1.8b
105; CHECK-NEXT:    ret
106  %urem = urem <4 x i32> %X, <i32 1, i32 1, i32 2, i32 3>
107  %cmp = icmp eq <4 x i32> %urem, <i32 0, i32 1, i32 2, i32 2>
108  ret <4 x i1> %cmp
109}
110