xref: /llvm-project/llvm/test/CodeGen/Thumb2/mve-vst2-post.ll (revision b5b663aac17415625340eb29c8010832bfc4c21c) 
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp,+fp64 -verify-machineinstrs %s -o - | FileCheck %s
3
4; i32
5
6define ptr @vst2_v4i32(ptr %src, ptr %dst) {
7; CHECK-LABEL: vst2_v4i32:
8; CHECK:       @ %bb.0: @ %entry
9; CHECK-NEXT:    vldrw.u32 q1, [r0, #16]
10; CHECK-NEXT:    vldrw.u32 q0, [r0]
11; CHECK-NEXT:    vst20.32 {q0, q1}, [r1]
12; CHECK-NEXT:    vst21.32 {q0, q1}, [r1]!
13; CHECK-NEXT:    mov r0, r1
14; CHECK-NEXT:    bx lr
15entry:
16  %l1 = load <4 x i32>, ptr %src, align 4
17  %s2 = getelementptr <4 x i32>, ptr %src, i32 1
18  %l2 = load <4 x i32>, ptr %s2, align 4
19  %s = shufflevector <4 x i32> %l1, <4 x i32> %l2, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
20  store <8 x i32> %s, ptr %dst
21  %ret = getelementptr inbounds <8 x i32>, ptr %dst, i32 1
22  ret ptr %ret
23}
24
25; i16
26
27define ptr @vst2_v8i16(ptr %src, ptr %dst) {
28; CHECK-LABEL: vst2_v8i16:
29; CHECK:       @ %bb.0: @ %entry
30; CHECK-NEXT:    vldrw.u32 q1, [r0, #16]
31; CHECK-NEXT:    vldrw.u32 q0, [r0]
32; CHECK-NEXT:    vst20.16 {q0, q1}, [r1]
33; CHECK-NEXT:    vst21.16 {q0, q1}, [r1]!
34; CHECK-NEXT:    mov r0, r1
35; CHECK-NEXT:    bx lr
36entry:
37  %l1 = load <8 x i16>, ptr %src, align 4
38  %s2 = getelementptr <8 x i16>, ptr %src, i32 1
39  %l2 = load <8 x i16>, ptr %s2, align 4
40  %s = shufflevector <8 x i16> %l1, <8 x i16> %l2, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
41  store <16 x i16> %s, ptr %dst
42  %ret = getelementptr inbounds <16 x i16>, ptr %dst, i32 1
43  ret ptr %ret
44}
45
46; i8
47
48define ptr @vst2_v16i8(ptr %src, ptr %dst) {
49; CHECK-LABEL: vst2_v16i8:
50; CHECK:       @ %bb.0: @ %entry
51; CHECK-NEXT:    vldrw.u32 q1, [r0, #16]
52; CHECK-NEXT:    vldrw.u32 q0, [r0]
53; CHECK-NEXT:    vst20.8 {q0, q1}, [r1]
54; CHECK-NEXT:    vst21.8 {q0, q1}, [r1]!
55; CHECK-NEXT:    mov r0, r1
56; CHECK-NEXT:    bx lr
57entry:
58  %l1 = load <16 x i8>, ptr %src, align 4
59  %s2 = getelementptr <16 x i8>, ptr %src, i32 1
60  %l2 = load <16 x i8>, ptr %s2, align 4
61  %s = shufflevector <16 x i8> %l1, <16 x i8> %l2, <32 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23, i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27, i32 12, i32 28, i32 13, i32 29, i32 14, i32 30, i32 15, i32 31>
62  store <32 x i8> %s, ptr %dst
63  %ret = getelementptr inbounds <32 x i8>, ptr %dst, i32 1
64  ret ptr %ret
65}
66
67; i64
68
69define ptr @vst2_v2i64(ptr %src, ptr %dst) {
70; CHECK-LABEL: vst2_v2i64:
71; CHECK:       @ %bb.0: @ %entry
72; CHECK-NEXT:    vldrw.u32 q0, [r0, #16]
73; CHECK-NEXT:    vldrw.u32 q1, [r0]
74; CHECK-NEXT:    vmov.f64 d5, d0
75; CHECK-NEXT:    vmov.f64 d0, d3
76; CHECK-NEXT:    vmov.f64 d4, d2
77; CHECK-NEXT:    vstrw.32 q0, [r1, #16]
78; CHECK-NEXT:    vstrw.32 q2, [r1], #32
79; CHECK-NEXT:    mov r0, r1
80; CHECK-NEXT:    bx lr
81entry:
82  %l1 = load <2 x i64>, ptr %src, align 4
83  %s2 = getelementptr <2 x i64>, ptr %src, i32 1
84  %l2 = load <2 x i64>, ptr %s2, align 4
85  %s = shufflevector <2 x i64> %l1, <2 x i64> %l2, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
86  store <4 x i64> %s, ptr %dst
87  %ret = getelementptr inbounds <4 x i64>, ptr %dst, i32 1
88  ret ptr %ret
89}
90
91; f32
92
93define ptr @vst2_v4f32(ptr %src, ptr %dst) {
94; CHECK-LABEL: vst2_v4f32:
95; CHECK:       @ %bb.0: @ %entry
96; CHECK-NEXT:    vldrw.u32 q1, [r0, #16]
97; CHECK-NEXT:    vldrw.u32 q0, [r0]
98; CHECK-NEXT:    vst20.32 {q0, q1}, [r1]
99; CHECK-NEXT:    vst21.32 {q0, q1}, [r1]!
100; CHECK-NEXT:    mov r0, r1
101; CHECK-NEXT:    bx lr
102entry:
103  %l1 = load <4 x float>, ptr %src, align 4
104  %s2 = getelementptr <4 x float>, ptr %src, i32 1
105  %l2 = load <4 x float>, ptr %s2, align 4
106  %s = shufflevector <4 x float> %l1, <4 x float> %l2, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
107  store <8 x float> %s, ptr %dst
108  %ret = getelementptr inbounds <8 x float>, ptr %dst, i32 1
109  ret ptr %ret
110}
111
112; f16
113
114define ptr @vst2_v8f16(ptr %src, ptr %dst) {
115; CHECK-LABEL: vst2_v8f16:
116; CHECK:       @ %bb.0: @ %entry
117; CHECK-NEXT:    vldrw.u32 q1, [r0, #16]
118; CHECK-NEXT:    vldrw.u32 q0, [r0]
119; CHECK-NEXT:    vst20.16 {q0, q1}, [r1]
120; CHECK-NEXT:    vst21.16 {q0, q1}, [r1]!
121; CHECK-NEXT:    mov r0, r1
122; CHECK-NEXT:    bx lr
123entry:
124  %l1 = load <8 x half>, ptr %src, align 4
125  %s2 = getelementptr <8 x half>, ptr %src, i32 1
126  %l2 = load <8 x half>, ptr %s2, align 4
127  %s = shufflevector <8 x half> %l1, <8 x half> %l2, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
128  store <16 x half> %s, ptr %dst
129  %ret = getelementptr inbounds <16 x half>, ptr %dst, i32 1
130  ret ptr %ret
131}
132
133; f64
134
135define ptr @vst2_v2f64(ptr %src, ptr %dst) {
136; CHECK-LABEL: vst2_v2f64:
137; CHECK:       @ %bb.0: @ %entry
138; CHECK-NEXT:    vldrw.u32 q0, [r0, #16]
139; CHECK-NEXT:    vldrw.u32 q1, [r0]
140; CHECK-NEXT:    vmov.f64 d5, d0
141; CHECK-NEXT:    vmov.f64 d0, d3
142; CHECK-NEXT:    vmov.f64 d4, d2
143; CHECK-NEXT:    vstrw.32 q0, [r1, #16]
144; CHECK-NEXT:    vstrw.32 q2, [r1], #32
145; CHECK-NEXT:    mov r0, r1
146; CHECK-NEXT:    bx lr
147entry:
148  %l1 = load <2 x double>, ptr %src, align 4
149  %s2 = getelementptr <2 x double>, ptr %src, i32 1
150  %l2 = load <2 x double>, ptr %s2, align 4
151  %s = shufflevector <2 x double> %l1, <2 x double> %l2, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
152  store <4 x double> %s, ptr %dst
153  %ret = getelementptr inbounds <4 x double>, ptr %dst, i32 1
154  ret ptr %ret
155}
156