1 ; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s
3 define void @vst4i8(ptr %A, ptr %B) nounwind {
5 ;Check the alignment value. Max for this instruction is 256 bits:
6 ;CHECK: vst4.8 {d16, d17, d18, d19}, [r0:64]
7 %tmp1 = load <8 x i8>, ptr %B
8 call void @llvm.arm.neon.vst4.p0.v8i8(ptr %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, i32 8)
12 ;Check for a post-increment updating store with register increment.
13 define void @vst4i8_update(ptr %ptr, ptr %B, i32 %inc) nounwind {
14 ;CHECK-LABEL: vst4i8_update:
15 ;CHECK: vst4.8 {d16, d17, d18, d19}, [r{{[0-9]+}}:128], r2
16 %A = load ptr, ptr %ptr
17 %tmp1 = load <8 x i8>, ptr %B
18 call void @llvm.arm.neon.vst4.p0.v8i8(ptr %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, i32 16)
19 %tmp2 = getelementptr i8, ptr %A, i32 %inc
20 store ptr %tmp2, ptr %ptr
24 define void @vst4i16(ptr %A, ptr %B) nounwind {
25 ;CHECK-LABEL: vst4i16:
26 ;Check the alignment value. Max for this instruction is 256 bits:
27 ;CHECK: vst4.16 {d16, d17, d18, d19}, [r0:128]
28 %tmp1 = load <4 x i16>, ptr %B
29 call void @llvm.arm.neon.vst4.p0.v4i16(ptr %A, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, i32 16)
33 define void @vst4i32(ptr %A, ptr %B) nounwind {
34 ;CHECK-LABEL: vst4i32:
35 ;Check the alignment value. Max for this instruction is 256 bits:
36 ;CHECK: vst4.32 {d16, d17, d18, d19}, [r0:256]
37 %tmp1 = load <2 x i32>, ptr %B
38 call void @llvm.arm.neon.vst4.p0.v2i32(ptr %A, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, i32 32)
42 define void @vst4f(ptr %A, ptr %B) nounwind {
45 %tmp1 = load <2 x float>, ptr %B
46 call void @llvm.arm.neon.vst4.p0.v2f32(ptr %A, <2 x float> %tmp1, <2 x float> %tmp1, <2 x float> %tmp1, <2 x float> %tmp1, i32 1)
50 define void @vst4i64(ptr %A, ptr %B) nounwind {
51 ;CHECK-LABEL: vst4i64:
52 ;Check the alignment value. Max for this instruction is 256 bits:
53 ;CHECK: vst1.64 {d16, d17, d18, d19}, [r0:256]
54 %tmp1 = load <1 x i64>, ptr %B
55 call void @llvm.arm.neon.vst4.p0.v1i64(ptr %A, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, i32 64)
59 define void @vst4i64_update(ptr %ptr, ptr %B) nounwind {
60 ;CHECK-LABEL: vst4i64_update:
61 ;CHECK: vst1.64 {d16, d17, d18, d19}, [r{{[0-9]+}}]!
62 %A = load ptr, ptr %ptr
63 %tmp1 = load <1 x i64>, ptr %B
64 call void @llvm.arm.neon.vst4.p0.v1i64(ptr %A, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, i32 1)
65 %tmp2 = getelementptr i64, ptr %A, i32 4
66 store ptr %tmp2, ptr %ptr
70 define void @vst4i64_reg_update(ptr %ptr, ptr %B) nounwind {
71 ;CHECK-LABEL: vst4i64_reg_update:
72 ;CHECK: vst1.64 {d16, d17, d18, d19}, [r{{[0-9]+}}], r{{[0-9]+}}
73 %A = load ptr, ptr %ptr
74 %tmp1 = load <1 x i64>, ptr %B
75 call void @llvm.arm.neon.vst4.p0.v1i64(ptr %A, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, <1 x i64> %tmp1, i32 1)
76 %tmp2 = getelementptr i64, ptr %A, i32 1
77 store ptr %tmp2, ptr %ptr
81 define void @vst4Qi8(ptr %A, ptr %B) nounwind {
82 ;CHECK-LABEL: vst4Qi8:
83 ;Check the alignment value. Max for this instruction is 256 bits:
84 ;CHECK: vst4.8 {d16, d18, d20, d22}, [r0:256]!
85 ;CHECK: vst4.8 {d17, d19, d21, d23}, [r0:256]
86 %tmp1 = load <16 x i8>, ptr %B
87 call void @llvm.arm.neon.vst4.p0.v16i8(ptr %A, <16 x i8> %tmp1, <16 x i8> %tmp1, <16 x i8> %tmp1, <16 x i8> %tmp1, i32 64)
91 define void @vst4Qi16(ptr %A, ptr %B) nounwind {
92 ;CHECK-LABEL: vst4Qi16:
93 ;Check for no alignment specifier.
94 ;CHECK: vst4.16 {d16, d18, d20, d22}, [r0]!
95 ;CHECK: vst4.16 {d17, d19, d21, d23}, [r0]
96 %tmp1 = load <8 x i16>, ptr %B
97 call void @llvm.arm.neon.vst4.p0.v8i16(ptr %A, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 1)
101 define void @vst4Qi32(ptr %A, ptr %B) nounwind {
102 ;CHECK-LABEL: vst4Qi32:
105 %tmp1 = load <4 x i32>, ptr %B
106 call void @llvm.arm.neon.vst4.p0.v4i32(ptr %A, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, i32 1)
110 define void @vst4Qf(ptr %A, ptr %B) nounwind {
111 ;CHECK-LABEL: vst4Qf:
114 %tmp1 = load <4 x float>, ptr %B
115 call void @llvm.arm.neon.vst4.p0.v4f32(ptr %A, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, i32 1)
119 ;Check for a post-increment updating store.
120 define void @vst4Qf_update(ptr %ptr, ptr %B) nounwind {
121 ;CHECK-LABEL: vst4Qf_update:
122 ;CHECK: vst4.32 {d16, d18, d20, d22}, [r[[REG:[0-9]+]]]!
123 ;CHECK: vst4.32 {d17, d19, d21, d23}, [r[[REG]]]!
124 %A = load ptr, ptr %ptr
125 %tmp1 = load <4 x float>, ptr %B
126 call void @llvm.arm.neon.vst4.p0.v4f32(ptr %A, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, <4 x float> %tmp1, i32 1)
127 %tmp2 = getelementptr float, ptr %A, i32 16
128 store ptr %tmp2, ptr %ptr
132 declare void @llvm.arm.neon.vst4.p0.v8i8(ptr, <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>, i32) nounwind
133 declare void @llvm.arm.neon.vst4.p0.v4i16(ptr, <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>, i32) nounwind
134 declare void @llvm.arm.neon.vst4.p0.v2i32(ptr, <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, i32) nounwind
135 declare void @llvm.arm.neon.vst4.p0.v2f32(ptr, <2 x float>, <2 x float>, <2 x float>, <2 x float>, i32) nounwind
136 declare void @llvm.arm.neon.vst4.p0.v1i64(ptr, <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64>, i32) nounwind
138 declare void @llvm.arm.neon.vst4.p0.v16i8(ptr, <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, i32) nounwind
139 declare void @llvm.arm.neon.vst4.p0.v8i16(ptr, <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>, i32) nounwind
140 declare void @llvm.arm.neon.vst4.p0.v4i32(ptr, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, i32) nounwind
141 declare void @llvm.arm.neon.vst4.p0.v4f32(ptr, <4 x float>, <4 x float>, <4 x float>, <4 x float>, i32) nounwind