1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt < %s -mtriple=x86_64-unknown -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=SSE
3 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=SSE
4 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX,AVX1
5 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX,AVX2
6 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX,AVX256
7 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX512
8 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX512
10 @a64 = common global [8 x i64] zeroinitializer, align 64
11 @b64 = common global [8 x i64] zeroinitializer, align 64
12 @c64 = common global [8 x i64] zeroinitializer, align 64
13 @d64 = common global [8 x i64] zeroinitializer, align 64
14 @a32 = common global [16 x i32] zeroinitializer, align 64
15 @b32 = common global [16 x i32] zeroinitializer, align 64
16 @c32 = common global [16 x i32] zeroinitializer, align 64
17 @d32 = common global [16 x i32] zeroinitializer, align 64
18 @a16 = common global [32 x i16] zeroinitializer, align 64
19 @b16 = common global [32 x i16] zeroinitializer, align 64
20 @c16 = common global [32 x i16] zeroinitializer, align 64
21 @d16 = common global [32 x i16] zeroinitializer, align 64
22 @a8 = common global [64 x i8] zeroinitializer, align 64
23 @b8 = common global [64 x i8] zeroinitializer, align 64
24 @c8 = common global [64 x i8] zeroinitializer, align 64
25 @d8 = common global [64 x i8] zeroinitializer, align 64
27 declare i64 @llvm.fshl.i64(i64, i64, i64)
28 declare i32 @llvm.fshl.i32(i32, i32, i32)
29 declare i16 @llvm.fshl.i16(i16, i16, i16)
30 declare i8 @llvm.fshl.i8 (i8 , i8 , i8 )
32 define void @fshl_v8i64() {
33 ; SSE-LABEL: @fshl_v8i64(
34 ; SSE-NEXT: [[A0:%.*]] = load i64, ptr @a64, align 8
35 ; SSE-NEXT: [[A1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
36 ; SSE-NEXT: [[A2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
37 ; SSE-NEXT: [[A3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
38 ; SSE-NEXT: [[A4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
39 ; SSE-NEXT: [[A5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
40 ; SSE-NEXT: [[A6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
41 ; SSE-NEXT: [[A7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
42 ; SSE-NEXT: [[B0:%.*]] = load i64, ptr @b64, align 8
43 ; SSE-NEXT: [[B1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
44 ; SSE-NEXT: [[B2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
45 ; SSE-NEXT: [[B3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
46 ; SSE-NEXT: [[B4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
47 ; SSE-NEXT: [[B5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
48 ; SSE-NEXT: [[B6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
49 ; SSE-NEXT: [[B7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
50 ; SSE-NEXT: [[R0:%.*]] = call i64 @llvm.fshl.i64(i64 [[A0]], i64 [[A0]], i64 [[B0]])
51 ; SSE-NEXT: [[R1:%.*]] = call i64 @llvm.fshl.i64(i64 [[A1]], i64 [[A1]], i64 [[B1]])
52 ; SSE-NEXT: [[R2:%.*]] = call i64 @llvm.fshl.i64(i64 [[A2]], i64 [[A2]], i64 [[B2]])
53 ; SSE-NEXT: [[R3:%.*]] = call i64 @llvm.fshl.i64(i64 [[A3]], i64 [[A3]], i64 [[B3]])
54 ; SSE-NEXT: [[R4:%.*]] = call i64 @llvm.fshl.i64(i64 [[A4]], i64 [[A4]], i64 [[B4]])
55 ; SSE-NEXT: [[R5:%.*]] = call i64 @llvm.fshl.i64(i64 [[A5]], i64 [[A5]], i64 [[B5]])
56 ; SSE-NEXT: [[R6:%.*]] = call i64 @llvm.fshl.i64(i64 [[A6]], i64 [[A6]], i64 [[B6]])
57 ; SSE-NEXT: [[R7:%.*]] = call i64 @llvm.fshl.i64(i64 [[A7]], i64 [[A7]], i64 [[B7]])
58 ; SSE-NEXT: store i64 [[R0]], ptr @d64, align 8
59 ; SSE-NEXT: store i64 [[R1]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 1), align 8
60 ; SSE-NEXT: store i64 [[R2]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 2), align 8
61 ; SSE-NEXT: store i64 [[R3]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 3), align 8
62 ; SSE-NEXT: store i64 [[R4]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 4), align 8
63 ; SSE-NEXT: store i64 [[R5]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 5), align 8
64 ; SSE-NEXT: store i64 [[R6]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 6), align 8
65 ; SSE-NEXT: store i64 [[R7]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 7), align 8
68 ; AVX1-LABEL: @fshl_v8i64(
69 ; AVX1-NEXT: [[A0:%.*]] = load i64, ptr @a64, align 8
70 ; AVX1-NEXT: [[A1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
71 ; AVX1-NEXT: [[A2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
72 ; AVX1-NEXT: [[A3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
73 ; AVX1-NEXT: [[A4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
74 ; AVX1-NEXT: [[A5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
75 ; AVX1-NEXT: [[A6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
76 ; AVX1-NEXT: [[A7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
77 ; AVX1-NEXT: [[B0:%.*]] = load i64, ptr @b64, align 8
78 ; AVX1-NEXT: [[B1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
79 ; AVX1-NEXT: [[B2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
80 ; AVX1-NEXT: [[B3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
81 ; AVX1-NEXT: [[B4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
82 ; AVX1-NEXT: [[B5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
83 ; AVX1-NEXT: [[B6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
84 ; AVX1-NEXT: [[B7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
85 ; AVX1-NEXT: [[R0:%.*]] = call i64 @llvm.fshl.i64(i64 [[A0]], i64 [[A0]], i64 [[B0]])
86 ; AVX1-NEXT: [[R1:%.*]] = call i64 @llvm.fshl.i64(i64 [[A1]], i64 [[A1]], i64 [[B1]])
87 ; AVX1-NEXT: [[R2:%.*]] = call i64 @llvm.fshl.i64(i64 [[A2]], i64 [[A2]], i64 [[B2]])
88 ; AVX1-NEXT: [[R3:%.*]] = call i64 @llvm.fshl.i64(i64 [[A3]], i64 [[A3]], i64 [[B3]])
89 ; AVX1-NEXT: [[R4:%.*]] = call i64 @llvm.fshl.i64(i64 [[A4]], i64 [[A4]], i64 [[B4]])
90 ; AVX1-NEXT: [[R5:%.*]] = call i64 @llvm.fshl.i64(i64 [[A5]], i64 [[A5]], i64 [[B5]])
91 ; AVX1-NEXT: [[R6:%.*]] = call i64 @llvm.fshl.i64(i64 [[A6]], i64 [[A6]], i64 [[B6]])
92 ; AVX1-NEXT: [[R7:%.*]] = call i64 @llvm.fshl.i64(i64 [[A7]], i64 [[A7]], i64 [[B7]])
93 ; AVX1-NEXT: store i64 [[R0]], ptr @d64, align 8
94 ; AVX1-NEXT: store i64 [[R1]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 1), align 8
95 ; AVX1-NEXT: store i64 [[R2]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 2), align 8
96 ; AVX1-NEXT: store i64 [[R3]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 3), align 8
97 ; AVX1-NEXT: store i64 [[R4]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 4), align 8
98 ; AVX1-NEXT: store i64 [[R5]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 5), align 8
99 ; AVX1-NEXT: store i64 [[R6]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 6), align 8
100 ; AVX1-NEXT: store i64 [[R7]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 7), align 8
101 ; AVX1-NEXT: ret void
103 ; AVX2-LABEL: @fshl_v8i64(
104 ; AVX2-NEXT: [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
105 ; AVX2-NEXT: [[TMP2:%.*]] = load <4 x i64>, ptr @b64, align 8
106 ; AVX2-NEXT: [[TMP3:%.*]] = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP1]], <4 x i64> [[TMP2]])
107 ; AVX2-NEXT: store <4 x i64> [[TMP3]], ptr @d64, align 8
108 ; AVX2-NEXT: [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
109 ; AVX2-NEXT: [[TMP5:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
110 ; AVX2-NEXT: [[TMP6:%.*]] = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> [[TMP4]], <4 x i64> [[TMP4]], <4 x i64> [[TMP5]])
111 ; AVX2-NEXT: store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 4), align 8
112 ; AVX2-NEXT: ret void
114 ; AVX256-LABEL: @fshl_v8i64(
115 ; AVX256-NEXT: [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
116 ; AVX256-NEXT: [[TMP2:%.*]] = load <4 x i64>, ptr @b64, align 8
117 ; AVX256-NEXT: [[TMP3:%.*]] = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP1]], <4 x i64> [[TMP2]])
118 ; AVX256-NEXT: store <4 x i64> [[TMP3]], ptr @d64, align 8
119 ; AVX256-NEXT: [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
120 ; AVX256-NEXT: [[TMP5:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
121 ; AVX256-NEXT: [[TMP6:%.*]] = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> [[TMP4]], <4 x i64> [[TMP4]], <4 x i64> [[TMP5]])
122 ; AVX256-NEXT: store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 4), align 8
123 ; AVX256-NEXT: ret void
125 ; AVX512-LABEL: @fshl_v8i64(
126 ; AVX512-NEXT: [[TMP1:%.*]] = load <8 x i64>, ptr @a64, align 8
127 ; AVX512-NEXT: [[TMP2:%.*]] = load <8 x i64>, ptr @b64, align 8
128 ; AVX512-NEXT: [[TMP3:%.*]] = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
129 ; AVX512-NEXT: store <8 x i64> [[TMP3]], ptr @d64, align 8
130 ; AVX512-NEXT: ret void
132 %a0 = load i64, ptr @a64, align 8
133 %a1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
134 %a2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
135 %a3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
136 %a4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
137 %a5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
138 %a6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
139 %a7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
140 %b0 = load i64, ptr @b64, align 8
141 %b1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
142 %b2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
143 %b3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
144 %b4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
145 %b5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
146 %b6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
147 %b7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
148 %r0 = call i64 @llvm.fshl.i64(i64 %a0, i64 %a0, i64 %b0)
149 %r1 = call i64 @llvm.fshl.i64(i64 %a1, i64 %a1, i64 %b1)
150 %r2 = call i64 @llvm.fshl.i64(i64 %a2, i64 %a2, i64 %b2)
151 %r3 = call i64 @llvm.fshl.i64(i64 %a3, i64 %a3, i64 %b3)
152 %r4 = call i64 @llvm.fshl.i64(i64 %a4, i64 %a4, i64 %b4)
153 %r5 = call i64 @llvm.fshl.i64(i64 %a5, i64 %a5, i64 %b5)
154 %r6 = call i64 @llvm.fshl.i64(i64 %a6, i64 %a6, i64 %b6)
155 %r7 = call i64 @llvm.fshl.i64(i64 %a7, i64 %a7, i64 %b7)
156 store i64 %r0, ptr @d64, align 8
157 store i64 %r1, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 1), align 8
158 store i64 %r2, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 2), align 8
159 store i64 %r3, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 3), align 8
160 store i64 %r4, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 4), align 8
161 store i64 %r5, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 5), align 8
162 store i64 %r6, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 6), align 8
163 store i64 %r7, ptr getelementptr inbounds ([8 x i64], ptr @d64, i32 0, i64 7), align 8
167 define void @fshl_v16i32() {
168 ; SSE-LABEL: @fshl_v16i32(
169 ; SSE-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
170 ; SSE-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
171 ; SSE-NEXT: [[A2:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 2), align 4
172 ; SSE-NEXT: [[A3:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 3), align 4
173 ; SSE-NEXT: [[A4:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4), align 4
174 ; SSE-NEXT: [[A5:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 5), align 4
175 ; SSE-NEXT: [[A6:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 6), align 4
176 ; SSE-NEXT: [[A7:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 7), align 4
177 ; SSE-NEXT: [[A8:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
178 ; SSE-NEXT: [[A9:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 9), align 4
179 ; SSE-NEXT: [[A10:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 10), align 4
180 ; SSE-NEXT: [[A11:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 11), align 4
181 ; SSE-NEXT: [[A12:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
182 ; SSE-NEXT: [[A13:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 13), align 4
183 ; SSE-NEXT: [[A14:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 14), align 4
184 ; SSE-NEXT: [[A15:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 15), align 4
185 ; SSE-NEXT: [[B0:%.*]] = load i32, ptr @b32, align 4
186 ; SSE-NEXT: [[B1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1), align 4
187 ; SSE-NEXT: [[B2:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 2), align 4
188 ; SSE-NEXT: [[B3:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 3), align 4
189 ; SSE-NEXT: [[B4:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4), align 4
190 ; SSE-NEXT: [[B5:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 5), align 4
191 ; SSE-NEXT: [[B6:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 6), align 4
192 ; SSE-NEXT: [[B7:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 7), align 4
193 ; SSE-NEXT: [[B8:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
194 ; SSE-NEXT: [[B9:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 9), align 4
195 ; SSE-NEXT: [[B10:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 10), align 4
196 ; SSE-NEXT: [[B11:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 11), align 4
197 ; SSE-NEXT: [[B12:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
198 ; SSE-NEXT: [[B13:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 13), align 4
199 ; SSE-NEXT: [[B14:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 14), align 4
200 ; SSE-NEXT: [[B15:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 15), align 4
201 ; SSE-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 [[B0]])
202 ; SSE-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 [[B1]])
203 ; SSE-NEXT: [[R2:%.*]] = call i32 @llvm.fshl.i32(i32 [[A2]], i32 [[A2]], i32 [[B2]])
204 ; SSE-NEXT: [[R3:%.*]] = call i32 @llvm.fshl.i32(i32 [[A3]], i32 [[A3]], i32 [[B3]])
205 ; SSE-NEXT: [[R4:%.*]] = call i32 @llvm.fshl.i32(i32 [[A4]], i32 [[A4]], i32 [[B4]])
206 ; SSE-NEXT: [[R5:%.*]] = call i32 @llvm.fshl.i32(i32 [[A5]], i32 [[A5]], i32 [[B5]])
207 ; SSE-NEXT: [[R6:%.*]] = call i32 @llvm.fshl.i32(i32 [[A6]], i32 [[A6]], i32 [[B6]])
208 ; SSE-NEXT: [[R7:%.*]] = call i32 @llvm.fshl.i32(i32 [[A7]], i32 [[A7]], i32 [[B7]])
209 ; SSE-NEXT: [[R8:%.*]] = call i32 @llvm.fshl.i32(i32 [[A8]], i32 [[A8]], i32 [[B8]])
210 ; SSE-NEXT: [[R9:%.*]] = call i32 @llvm.fshl.i32(i32 [[A9]], i32 [[A9]], i32 [[B9]])
211 ; SSE-NEXT: [[R10:%.*]] = call i32 @llvm.fshl.i32(i32 [[A10]], i32 [[A10]], i32 [[B10]])
212 ; SSE-NEXT: [[R11:%.*]] = call i32 @llvm.fshl.i32(i32 [[A11]], i32 [[A11]], i32 [[B11]])
213 ; SSE-NEXT: [[R12:%.*]] = call i32 @llvm.fshl.i32(i32 [[A12]], i32 [[A12]], i32 [[B12]])
214 ; SSE-NEXT: [[R13:%.*]] = call i32 @llvm.fshl.i32(i32 [[A13]], i32 [[A13]], i32 [[B13]])
215 ; SSE-NEXT: [[R14:%.*]] = call i32 @llvm.fshl.i32(i32 [[A14]], i32 [[A14]], i32 [[B14]])
216 ; SSE-NEXT: [[R15:%.*]] = call i32 @llvm.fshl.i32(i32 [[A15]], i32 [[A15]], i32 [[B15]])
217 ; SSE-NEXT: store i32 [[R0]], ptr @d32, align 4
218 ; SSE-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
219 ; SSE-NEXT: store i32 [[R2]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 2), align 4
220 ; SSE-NEXT: store i32 [[R3]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 3), align 4
221 ; SSE-NEXT: store i32 [[R4]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 4), align 4
222 ; SSE-NEXT: store i32 [[R5]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 5), align 4
223 ; SSE-NEXT: store i32 [[R6]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 6), align 4
224 ; SSE-NEXT: store i32 [[R7]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 7), align 4
225 ; SSE-NEXT: store i32 [[R8]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 8), align 4
226 ; SSE-NEXT: store i32 [[R9]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 9), align 4
227 ; SSE-NEXT: store i32 [[R10]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 10), align 4
228 ; SSE-NEXT: store i32 [[R11]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 11), align 4
229 ; SSE-NEXT: store i32 [[R12]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 12), align 4
230 ; SSE-NEXT: store i32 [[R13]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 13), align 4
231 ; SSE-NEXT: store i32 [[R14]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 14), align 4
232 ; SSE-NEXT: store i32 [[R15]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 15), align 4
235 ; AVX-LABEL: @fshl_v16i32(
236 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x i32>, ptr @a32, align 4
237 ; AVX-NEXT: [[TMP2:%.*]] = load <8 x i32>, ptr @b32, align 4
238 ; AVX-NEXT: [[TMP3:%.*]] = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP1]], <8 x i32> [[TMP2]])
239 ; AVX-NEXT: store <8 x i32> [[TMP3]], ptr @d32, align 4
240 ; AVX-NEXT: [[TMP4:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
241 ; AVX-NEXT: [[TMP5:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
242 ; AVX-NEXT: [[TMP6:%.*]] = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> [[TMP4]], <8 x i32> [[TMP4]], <8 x i32> [[TMP5]])
243 ; AVX-NEXT: store <8 x i32> [[TMP6]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 8), align 4
246 ; AVX512-LABEL: @fshl_v16i32(
247 ; AVX512-NEXT: [[TMP1:%.*]] = load <16 x i32>, ptr @a32, align 4
248 ; AVX512-NEXT: [[TMP2:%.*]] = load <16 x i32>, ptr @b32, align 4
249 ; AVX512-NEXT: [[TMP3:%.*]] = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
250 ; AVX512-NEXT: store <16 x i32> [[TMP3]], ptr @d32, align 4
251 ; AVX512-NEXT: ret void
253 %a0 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 0 ), align 4
254 %a1 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1 ), align 4
255 %a2 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 2 ), align 4
256 %a3 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 3 ), align 4
257 %a4 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4 ), align 4
258 %a5 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 5 ), align 4
259 %a6 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 6 ), align 4
260 %a7 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 7 ), align 4
261 %a8 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8 ), align 4
262 %a9 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 9 ), align 4
263 %a10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 10), align 4
264 %a11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 11), align 4
265 %a12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
266 %a13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 13), align 4
267 %a14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 14), align 4
268 %a15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 15), align 4
269 %b0 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 0 ), align 4
270 %b1 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1 ), align 4
271 %b2 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 2 ), align 4
272 %b3 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 3 ), align 4
273 %b4 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4 ), align 4
274 %b5 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 5 ), align 4
275 %b6 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 6 ), align 4
276 %b7 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 7 ), align 4
277 %b8 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8 ), align 4
278 %b9 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 9 ), align 4
279 %b10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 10), align 4
280 %b11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 11), align 4
281 %b12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
282 %b13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 13), align 4
283 %b14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 14), align 4
284 %b15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 15), align 4
285 %r0 = call i32 @llvm.fshl.i32(i32 %a0 , i32 %a0 , i32 %b0 )
286 %r1 = call i32 @llvm.fshl.i32(i32 %a1 , i32 %a1 , i32 %b1 )
287 %r2 = call i32 @llvm.fshl.i32(i32 %a2 , i32 %a2 , i32 %b2 )
288 %r3 = call i32 @llvm.fshl.i32(i32 %a3 , i32 %a3 , i32 %b3 )
289 %r4 = call i32 @llvm.fshl.i32(i32 %a4 , i32 %a4 , i32 %b4 )
290 %r5 = call i32 @llvm.fshl.i32(i32 %a5 , i32 %a5 , i32 %b5 )
291 %r6 = call i32 @llvm.fshl.i32(i32 %a6 , i32 %a6 , i32 %b6 )
292 %r7 = call i32 @llvm.fshl.i32(i32 %a7 , i32 %a7 , i32 %b7 )
293 %r8 = call i32 @llvm.fshl.i32(i32 %a8 , i32 %a8 , i32 %b8 )
294 %r9 = call i32 @llvm.fshl.i32(i32 %a9 , i32 %a9 , i32 %b9 )
295 %r10 = call i32 @llvm.fshl.i32(i32 %a10, i32 %a10, i32 %b10)
296 %r11 = call i32 @llvm.fshl.i32(i32 %a11, i32 %a11, i32 %b11)
297 %r12 = call i32 @llvm.fshl.i32(i32 %a12, i32 %a12, i32 %b12)
298 %r13 = call i32 @llvm.fshl.i32(i32 %a13, i32 %a13, i32 %b13)
299 %r14 = call i32 @llvm.fshl.i32(i32 %a14, i32 %a14, i32 %b14)
300 %r15 = call i32 @llvm.fshl.i32(i32 %a15, i32 %a15, i32 %b15)
301 store i32 %r0 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 0 ), align 4
302 store i32 %r1 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1 ), align 4
303 store i32 %r2 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 2 ), align 4
304 store i32 %r3 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 3 ), align 4
305 store i32 %r4 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 4 ), align 4
306 store i32 %r5 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 5 ), align 4
307 store i32 %r6 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 6 ), align 4
308 store i32 %r7 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 7 ), align 4
309 store i32 %r8 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 8 ), align 4
310 store i32 %r9 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 9 ), align 4
311 store i32 %r10, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 10), align 4
312 store i32 %r11, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 11), align 4
313 store i32 %r12, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 12), align 4
314 store i32 %r13, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 13), align 4
315 store i32 %r14, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 14), align 4
316 store i32 %r15, ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 15), align 4
320 define void @fshl_v32i16() {
321 ; AVX-LABEL: @fshl_v32i16(
322 ; AVX-NEXT: [[TMP1:%.*]] = load <16 x i16>, ptr @a16, align 2
323 ; AVX-NEXT: [[TMP2:%.*]] = load <16 x i16>, ptr @b16, align 2
324 ; AVX-NEXT: [[TMP3:%.*]] = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP1]], <16 x i16> [[TMP2]])
325 ; AVX-NEXT: store <16 x i16> [[TMP3]], ptr @d16, align 2
326 ; AVX-NEXT: [[TMP4:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
327 ; AVX-NEXT: [[TMP5:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
328 ; AVX-NEXT: [[TMP6:%.*]] = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> [[TMP4]], <16 x i16> [[TMP4]], <16 x i16> [[TMP5]])
329 ; AVX-NEXT: store <16 x i16> [[TMP6]], ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 16), align 2
332 ; AVX512-LABEL: @fshl_v32i16(
333 ; AVX512-NEXT: [[TMP1:%.*]] = load <32 x i16>, ptr @a16, align 2
334 ; AVX512-NEXT: [[TMP2:%.*]] = load <32 x i16>, ptr @b16, align 2
335 ; AVX512-NEXT: [[TMP3:%.*]] = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
336 ; AVX512-NEXT: store <32 x i16> [[TMP3]], ptr @d16, align 2
337 ; AVX512-NEXT: ret void
339 %a0 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 0 ), align 2
340 %a1 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 1 ), align 2
341 %a2 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 2 ), align 2
342 %a3 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 3 ), align 2
343 %a4 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 4 ), align 2
344 %a5 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 5 ), align 2
345 %a6 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 6 ), align 2
346 %a7 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 7 ), align 2
347 %a8 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 8 ), align 2
348 %a9 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 9 ), align 2
349 %a10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 10), align 2
350 %a11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 11), align 2
351 %a12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 12), align 2
352 %a13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 13), align 2
353 %a14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 14), align 2
354 %a15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 15), align 2
355 %a16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
356 %a17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 17), align 2
357 %a18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 18), align 2
358 %a19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 19), align 2
359 %a20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 20), align 2
360 %a21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 21), align 2
361 %a22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 22), align 2
362 %a23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 23), align 2
363 %a24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 24), align 2
364 %a25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 25), align 2
365 %a26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 26), align 2
366 %a27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 27), align 2
367 %a28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 28), align 2
368 %a29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 29), align 2
369 %a30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 30), align 2
370 %a31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 31), align 2
371 %b0 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 0 ), align 2
372 %b1 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 1 ), align 2
373 %b2 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 2 ), align 2
374 %b3 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 3 ), align 2
375 %b4 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 4 ), align 2
376 %b5 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 5 ), align 2
377 %b6 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 6 ), align 2
378 %b7 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 7 ), align 2
379 %b8 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 8 ), align 2
380 %b9 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 9 ), align 2
381 %b10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 10), align 2
382 %b11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 11), align 2
383 %b12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 12), align 2
384 %b13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 13), align 2
385 %b14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 14), align 2
386 %b15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 15), align 2
387 %b16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
388 %b17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 17), align 2
389 %b18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 18), align 2
390 %b19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 19), align 2
391 %b20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 20), align 2
392 %b21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 21), align 2
393 %b22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 22), align 2
394 %b23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 23), align 2
395 %b24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 24), align 2
396 %b25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 25), align 2
397 %b26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 26), align 2
398 %b27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 27), align 2
399 %b28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 28), align 2
400 %b29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 29), align 2
401 %b30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 30), align 2
402 %b31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 31), align 2
403 %r0 = call i16 @llvm.fshl.i16(i16 %a0 , i16 %a0 , i16 %b0 )
404 %r1 = call i16 @llvm.fshl.i16(i16 %a1 , i16 %a1 , i16 %b1 )
405 %r2 = call i16 @llvm.fshl.i16(i16 %a2 , i16 %a2 , i16 %b2 )
406 %r3 = call i16 @llvm.fshl.i16(i16 %a3 , i16 %a3 , i16 %b3 )
407 %r4 = call i16 @llvm.fshl.i16(i16 %a4 , i16 %a4 , i16 %b4 )
408 %r5 = call i16 @llvm.fshl.i16(i16 %a5 , i16 %a5 , i16 %b5 )
409 %r6 = call i16 @llvm.fshl.i16(i16 %a6 , i16 %a6 , i16 %b6 )
410 %r7 = call i16 @llvm.fshl.i16(i16 %a7 , i16 %a7 , i16 %b7 )
411 %r8 = call i16 @llvm.fshl.i16(i16 %a8 , i16 %a8 , i16 %b8 )
412 %r9 = call i16 @llvm.fshl.i16(i16 %a9 , i16 %a9 , i16 %b9 )
413 %r10 = call i16 @llvm.fshl.i16(i16 %a10, i16 %a10, i16 %b10)
414 %r11 = call i16 @llvm.fshl.i16(i16 %a11, i16 %a11, i16 %b11)
415 %r12 = call i16 @llvm.fshl.i16(i16 %a12, i16 %a12, i16 %b12)
416 %r13 = call i16 @llvm.fshl.i16(i16 %a13, i16 %a13, i16 %b13)
417 %r14 = call i16 @llvm.fshl.i16(i16 %a14, i16 %a14, i16 %b14)
418 %r15 = call i16 @llvm.fshl.i16(i16 %a15, i16 %a15, i16 %b15)
419 %r16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 %b16)
420 %r17 = call i16 @llvm.fshl.i16(i16 %a17, i16 %a17, i16 %b17)
421 %r18 = call i16 @llvm.fshl.i16(i16 %a18, i16 %a18, i16 %b18)
422 %r19 = call i16 @llvm.fshl.i16(i16 %a19, i16 %a19, i16 %b19)
423 %r20 = call i16 @llvm.fshl.i16(i16 %a20, i16 %a20, i16 %b20)
424 %r21 = call i16 @llvm.fshl.i16(i16 %a21, i16 %a21, i16 %b21)
425 %r22 = call i16 @llvm.fshl.i16(i16 %a22, i16 %a22, i16 %b22)
426 %r23 = call i16 @llvm.fshl.i16(i16 %a23, i16 %a23, i16 %b23)
427 %r24 = call i16 @llvm.fshl.i16(i16 %a24, i16 %a24, i16 %b24)
428 %r25 = call i16 @llvm.fshl.i16(i16 %a25, i16 %a25, i16 %b25)
429 %r26 = call i16 @llvm.fshl.i16(i16 %a26, i16 %a26, i16 %b26)
430 %r27 = call i16 @llvm.fshl.i16(i16 %a27, i16 %a27, i16 %b27)
431 %r28 = call i16 @llvm.fshl.i16(i16 %a28, i16 %a28, i16 %b28)
432 %r29 = call i16 @llvm.fshl.i16(i16 %a29, i16 %a29, i16 %b29)
433 %r30 = call i16 @llvm.fshl.i16(i16 %a30, i16 %a30, i16 %b30)
434 %r31 = call i16 @llvm.fshl.i16(i16 %a31, i16 %a31, i16 %b31)
435 store i16 %r0 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 0 ), align 2
436 store i16 %r1 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 1 ), align 2
437 store i16 %r2 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 2 ), align 2
438 store i16 %r3 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 3 ), align 2
439 store i16 %r4 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 4 ), align 2
440 store i16 %r5 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 5 ), align 2
441 store i16 %r6 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 6 ), align 2
442 store i16 %r7 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 7 ), align 2
443 store i16 %r8 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 8 ), align 2
444 store i16 %r9 , ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 9 ), align 2
445 store i16 %r10, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 10), align 2
446 store i16 %r11, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 11), align 2
447 store i16 %r12, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 12), align 2
448 store i16 %r13, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 13), align 2
449 store i16 %r14, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 14), align 2
450 store i16 %r15, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 15), align 2
451 store i16 %r16, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 16), align 2
452 store i16 %r17, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 17), align 2
453 store i16 %r18, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 18), align 2
454 store i16 %r19, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 19), align 2
455 store i16 %r20, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 20), align 2
456 store i16 %r21, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 21), align 2
457 store i16 %r22, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 22), align 2
458 store i16 %r23, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 23), align 2
459 store i16 %r24, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 24), align 2
460 store i16 %r25, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 25), align 2
461 store i16 %r26, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 26), align 2
462 store i16 %r27, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 27), align 2
463 store i16 %r28, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 28), align 2
464 store i16 %r29, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 29), align 2
465 store i16 %r30, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 30), align 2
466 store i16 %r31, ptr getelementptr inbounds ([32 x i16], ptr @d16, i32 0, i64 31), align 2
470 define void @fshl_v64i8() {
471 ; SSE-LABEL: @fshl_v64i8(
472 ; SSE-NEXT: [[TMP1:%.*]] = load <16 x i8>, ptr @a8, align 1
473 ; SSE-NEXT: [[TMP2:%.*]] = load <16 x i8>, ptr @b8, align 1
474 ; SSE-NEXT: [[TMP3:%.*]] = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP1]], <16 x i8> [[TMP2]])
475 ; SSE-NEXT: store <16 x i8> [[TMP3]], ptr @d8, align 1
476 ; SSE-NEXT: [[TMP4:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
477 ; SSE-NEXT: [[TMP5:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
478 ; SSE-NEXT: [[TMP6:%.*]] = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP4]], <16 x i8> [[TMP5]])
479 ; SSE-NEXT: store <16 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 16), align 1
480 ; SSE-NEXT: [[TMP7:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
481 ; SSE-NEXT: [[TMP8:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
482 ; SSE-NEXT: [[TMP9:%.*]] = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> [[TMP7]], <16 x i8> [[TMP7]], <16 x i8> [[TMP8]])
483 ; SSE-NEXT: store <16 x i8> [[TMP9]], ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 32), align 1
484 ; SSE-NEXT: [[TMP10:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
485 ; SSE-NEXT: [[TMP11:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
486 ; SSE-NEXT: [[TMP12:%.*]] = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> [[TMP10]], <16 x i8> [[TMP10]], <16 x i8> [[TMP11]])
487 ; SSE-NEXT: store <16 x i8> [[TMP12]], ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 48), align 1
490 ; AVX-LABEL: @fshl_v64i8(
491 ; AVX-NEXT: [[TMP1:%.*]] = load <32 x i8>, ptr @a8, align 1
492 ; AVX-NEXT: [[TMP2:%.*]] = load <32 x i8>, ptr @b8, align 1
493 ; AVX-NEXT: [[TMP3:%.*]] = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP1]], <32 x i8> [[TMP2]])
494 ; AVX-NEXT: store <32 x i8> [[TMP3]], ptr @d8, align 1
495 ; AVX-NEXT: [[TMP4:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
496 ; AVX-NEXT: [[TMP5:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
497 ; AVX-NEXT: [[TMP6:%.*]] = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> [[TMP4]], <32 x i8> [[TMP4]], <32 x i8> [[TMP5]])
498 ; AVX-NEXT: store <32 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 32), align 1
501 ; AVX512-LABEL: @fshl_v64i8(
502 ; AVX512-NEXT: [[TMP1:%.*]] = load <64 x i8>, ptr @a8, align 1
503 ; AVX512-NEXT: [[TMP2:%.*]] = load <64 x i8>, ptr @b8, align 1
504 ; AVX512-NEXT: [[TMP3:%.*]] = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
505 ; AVX512-NEXT: store <64 x i8> [[TMP3]], ptr @d8, align 1
506 ; AVX512-NEXT: ret void
508 %a0 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 0 ), align 1
509 %a1 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 1 ), align 1
510 %a2 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 2 ), align 1
511 %a3 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 3 ), align 1
512 %a4 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 4 ), align 1
513 %a5 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 5 ), align 1
514 %a6 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 6 ), align 1
515 %a7 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 7 ), align 1
516 %a8 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 8 ), align 1
517 %a9 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 9 ), align 1
518 %a10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 10), align 1
519 %a11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 11), align 1
520 %a12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 12), align 1
521 %a13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 13), align 1
522 %a14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 14), align 1
523 %a15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 15), align 1
524 %a16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
525 %a17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 17), align 1
526 %a18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 18), align 1
527 %a19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 19), align 1
528 %a20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 20), align 1
529 %a21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 21), align 1
530 %a22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 22), align 1
531 %a23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 23), align 1
532 %a24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 24), align 1
533 %a25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 25), align 1
534 %a26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 26), align 1
535 %a27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 27), align 1
536 %a28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 28), align 1
537 %a29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 29), align 1
538 %a30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 30), align 1
539 %a31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 31), align 1
540 %a32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
541 %a33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 33), align 1
542 %a34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 34), align 1
543 %a35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 35), align 1
544 %a36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 36), align 1
545 %a37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 37), align 1
546 %a38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 38), align 1
547 %a39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 39), align 1
548 %a40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 40), align 1
549 %a41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 41), align 1
550 %a42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 42), align 1
551 %a43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 43), align 1
552 %a44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 44), align 1
553 %a45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 45), align 1
554 %a46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 46), align 1
555 %a47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 47), align 1
556 %a48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
557 %a49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 49), align 1
558 %a50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 50), align 1
559 %a51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 51), align 1
560 %a52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 52), align 1
561 %a53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 53), align 1
562 %a54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 54), align 1
563 %a55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 55), align 1
564 %a56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 56), align 1
565 %a57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 57), align 1
566 %a58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 58), align 1
567 %a59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 59), align 1
568 %a60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 60), align 1
569 %a61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 61), align 1
570 %a62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 62), align 1
571 %a63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 63), align 1
572 %b0 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 0 ), align 1
573 %b1 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 1 ), align 1
574 %b2 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 2 ), align 1
575 %b3 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 3 ), align 1
576 %b4 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 4 ), align 1
577 %b5 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 5 ), align 1
578 %b6 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 6 ), align 1
579 %b7 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 7 ), align 1
580 %b8 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 8 ), align 1
581 %b9 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 9 ), align 1
582 %b10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 10), align 1
583 %b11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 11), align 1
584 %b12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 12), align 1
585 %b13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 13), align 1
586 %b14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 14), align 1
587 %b15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 15), align 1
588 %b16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
589 %b17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 17), align 1
590 %b18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 18), align 1
591 %b19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 19), align 1
592 %b20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 20), align 1
593 %b21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 21), align 1
594 %b22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 22), align 1
595 %b23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 23), align 1
596 %b24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 24), align 1
597 %b25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 25), align 1
598 %b26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 26), align 1
599 %b27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 27), align 1
600 %b28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 28), align 1
601 %b29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 29), align 1
602 %b30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 30), align 1
603 %b31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 31), align 1
604 %b32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
605 %b33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 33), align 1
606 %b34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 34), align 1
607 %b35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 35), align 1
608 %b36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 36), align 1
609 %b37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 37), align 1
610 %b38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 38), align 1
611 %b39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 39), align 1
612 %b40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 40), align 1
613 %b41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 41), align 1
614 %b42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 42), align 1
615 %b43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 43), align 1
616 %b44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 44), align 1
617 %b45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 45), align 1
618 %b46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 46), align 1
619 %b47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 47), align 1
620 %b48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
621 %b49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 49), align 1
622 %b50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 50), align 1
623 %b51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 51), align 1
624 %b52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 52), align 1
625 %b53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 53), align 1
626 %b54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 54), align 1
627 %b55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 55), align 1
628 %b56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 56), align 1
629 %b57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 57), align 1
630 %b58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 58), align 1
631 %b59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 59), align 1
632 %b60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 60), align 1
633 %b61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 61), align 1
634 %b62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 62), align 1
635 %b63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 63), align 1
636 %r0 = call i8 @llvm.fshl.i8(i8 %a0 , i8 %a0 , i8 %b0 )
637 %r1 = call i8 @llvm.fshl.i8(i8 %a1 , i8 %a1 , i8 %b1 )
638 %r2 = call i8 @llvm.fshl.i8(i8 %a2 , i8 %a2 , i8 %b2 )
639 %r3 = call i8 @llvm.fshl.i8(i8 %a3 , i8 %a3 , i8 %b3 )
640 %r4 = call i8 @llvm.fshl.i8(i8 %a4 , i8 %a4 , i8 %b4 )
641 %r5 = call i8 @llvm.fshl.i8(i8 %a5 , i8 %a5 , i8 %b5 )
642 %r6 = call i8 @llvm.fshl.i8(i8 %a6 , i8 %a6 , i8 %b6 )
643 %r7 = call i8 @llvm.fshl.i8(i8 %a7 , i8 %a7 , i8 %b7 )
644 %r8 = call i8 @llvm.fshl.i8(i8 %a8 , i8 %a8 , i8 %b8 )
645 %r9 = call i8 @llvm.fshl.i8(i8 %a9 , i8 %a9 , i8 %b9 )
646 %r10 = call i8 @llvm.fshl.i8(i8 %a10, i8 %a10, i8 %b10)
647 %r11 = call i8 @llvm.fshl.i8(i8 %a11, i8 %a11, i8 %b11)
648 %r12 = call i8 @llvm.fshl.i8(i8 %a12, i8 %a12, i8 %b12)
649 %r13 = call i8 @llvm.fshl.i8(i8 %a13, i8 %a13, i8 %b13)
650 %r14 = call i8 @llvm.fshl.i8(i8 %a14, i8 %a14, i8 %b14)
651 %r15 = call i8 @llvm.fshl.i8(i8 %a15, i8 %a15, i8 %b15)
652 %r16 = call i8 @llvm.fshl.i8(i8 %a16, i8 %a16, i8 %b16)
653 %r17 = call i8 @llvm.fshl.i8(i8 %a17, i8 %a17, i8 %b17)
654 %r18 = call i8 @llvm.fshl.i8(i8 %a18, i8 %a18, i8 %b18)
655 %r19 = call i8 @llvm.fshl.i8(i8 %a19, i8 %a19, i8 %b19)
656 %r20 = call i8 @llvm.fshl.i8(i8 %a20, i8 %a20, i8 %b20)
657 %r21 = call i8 @llvm.fshl.i8(i8 %a21, i8 %a21, i8 %b21)
658 %r22 = call i8 @llvm.fshl.i8(i8 %a22, i8 %a22, i8 %b22)
659 %r23 = call i8 @llvm.fshl.i8(i8 %a23, i8 %a23, i8 %b23)
660 %r24 = call i8 @llvm.fshl.i8(i8 %a24, i8 %a24, i8 %b24)
661 %r25 = call i8 @llvm.fshl.i8(i8 %a25, i8 %a25, i8 %b25)
662 %r26 = call i8 @llvm.fshl.i8(i8 %a26, i8 %a26, i8 %b26)
663 %r27 = call i8 @llvm.fshl.i8(i8 %a27, i8 %a27, i8 %b27)
664 %r28 = call i8 @llvm.fshl.i8(i8 %a28, i8 %a28, i8 %b28)
665 %r29 = call i8 @llvm.fshl.i8(i8 %a29, i8 %a29, i8 %b29)
666 %r30 = call i8 @llvm.fshl.i8(i8 %a30, i8 %a30, i8 %b30)
667 %r31 = call i8 @llvm.fshl.i8(i8 %a31, i8 %a31, i8 %b31)
668 %r32 = call i8 @llvm.fshl.i8(i8 %a32, i8 %a32, i8 %b32)
669 %r33 = call i8 @llvm.fshl.i8(i8 %a33, i8 %a33, i8 %b33)
670 %r34 = call i8 @llvm.fshl.i8(i8 %a34, i8 %a34, i8 %b34)
671 %r35 = call i8 @llvm.fshl.i8(i8 %a35, i8 %a35, i8 %b35)
672 %r36 = call i8 @llvm.fshl.i8(i8 %a36, i8 %a36, i8 %b36)
673 %r37 = call i8 @llvm.fshl.i8(i8 %a37, i8 %a37, i8 %b37)
674 %r38 = call i8 @llvm.fshl.i8(i8 %a38, i8 %a38, i8 %b38)
675 %r39 = call i8 @llvm.fshl.i8(i8 %a39, i8 %a39, i8 %b39)
676 %r40 = call i8 @llvm.fshl.i8(i8 %a40, i8 %a40, i8 %b40)
677 %r41 = call i8 @llvm.fshl.i8(i8 %a41, i8 %a41, i8 %b41)
678 %r42 = call i8 @llvm.fshl.i8(i8 %a42, i8 %a42, i8 %b42)
679 %r43 = call i8 @llvm.fshl.i8(i8 %a43, i8 %a43, i8 %b43)
680 %r44 = call i8 @llvm.fshl.i8(i8 %a44, i8 %a44, i8 %b44)
681 %r45 = call i8 @llvm.fshl.i8(i8 %a45, i8 %a45, i8 %b45)
682 %r46 = call i8 @llvm.fshl.i8(i8 %a46, i8 %a46, i8 %b46)
683 %r47 = call i8 @llvm.fshl.i8(i8 %a47, i8 %a47, i8 %b47)
684 %r48 = call i8 @llvm.fshl.i8(i8 %a48, i8 %a48, i8 %b48)
685 %r49 = call i8 @llvm.fshl.i8(i8 %a49, i8 %a49, i8 %b49)
686 %r50 = call i8 @llvm.fshl.i8(i8 %a50, i8 %a50, i8 %b50)
687 %r51 = call i8 @llvm.fshl.i8(i8 %a51, i8 %a51, i8 %b51)
688 %r52 = call i8 @llvm.fshl.i8(i8 %a52, i8 %a52, i8 %b52)
689 %r53 = call i8 @llvm.fshl.i8(i8 %a53, i8 %a53, i8 %b53)
690 %r54 = call i8 @llvm.fshl.i8(i8 %a54, i8 %a54, i8 %b54)
691 %r55 = call i8 @llvm.fshl.i8(i8 %a55, i8 %a55, i8 %b55)
692 %r56 = call i8 @llvm.fshl.i8(i8 %a56, i8 %a56, i8 %b56)
693 %r57 = call i8 @llvm.fshl.i8(i8 %a57, i8 %a57, i8 %b57)
694 %r58 = call i8 @llvm.fshl.i8(i8 %a58, i8 %a58, i8 %b58)
695 %r59 = call i8 @llvm.fshl.i8(i8 %a59, i8 %a59, i8 %b59)
696 %r60 = call i8 @llvm.fshl.i8(i8 %a60, i8 %a60, i8 %b60)
697 %r61 = call i8 @llvm.fshl.i8(i8 %a61, i8 %a61, i8 %b61)
698 %r62 = call i8 @llvm.fshl.i8(i8 %a62, i8 %a62, i8 %b62)
699 %r63 = call i8 @llvm.fshl.i8(i8 %a63, i8 %a63, i8 %b63)
700 store i8 %r0 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 0 ), align 1
701 store i8 %r1 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 1 ), align 1
702 store i8 %r2 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 2 ), align 1
703 store i8 %r3 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 3 ), align 1
704 store i8 %r4 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 4 ), align 1
705 store i8 %r5 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 5 ), align 1
706 store i8 %r6 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 6 ), align 1
707 store i8 %r7 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 7 ), align 1
708 store i8 %r8 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 8 ), align 1
709 store i8 %r9 , ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 9 ), align 1
710 store i8 %r10, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 10), align 1
711 store i8 %r11, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 11), align 1
712 store i8 %r12, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 12), align 1
713 store i8 %r13, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 13), align 1
714 store i8 %r14, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 14), align 1
715 store i8 %r15, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 15), align 1
716 store i8 %r16, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 16), align 1
717 store i8 %r17, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 17), align 1
718 store i8 %r18, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 18), align 1
719 store i8 %r19, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 19), align 1
720 store i8 %r20, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 20), align 1
721 store i8 %r21, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 21), align 1
722 store i8 %r22, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 22), align 1
723 store i8 %r23, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 23), align 1
724 store i8 %r24, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 24), align 1
725 store i8 %r25, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 25), align 1
726 store i8 %r26, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 26), align 1
727 store i8 %r27, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 27), align 1
728 store i8 %r28, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 28), align 1
729 store i8 %r29, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 29), align 1
730 store i8 %r30, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 30), align 1
731 store i8 %r31, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 31), align 1
732 store i8 %r32, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 32), align 1
733 store i8 %r33, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 33), align 1
734 store i8 %r34, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 34), align 1
735 store i8 %r35, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 35), align 1
736 store i8 %r36, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 36), align 1
737 store i8 %r37, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 37), align 1
738 store i8 %r38, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 38), align 1
739 store i8 %r39, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 39), align 1
740 store i8 %r40, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 40), align 1
741 store i8 %r41, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 41), align 1
742 store i8 %r42, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 42), align 1
743 store i8 %r43, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 43), align 1
744 store i8 %r44, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 44), align 1
745 store i8 %r45, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 45), align 1
746 store i8 %r46, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 46), align 1
747 store i8 %r47, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 47), align 1
748 store i8 %r48, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 48), align 1
749 store i8 %r49, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 49), align 1
750 store i8 %r50, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 50), align 1
751 store i8 %r51, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 51), align 1
752 store i8 %r52, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 52), align 1
753 store i8 %r53, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 53), align 1
754 store i8 %r54, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 54), align 1
755 store i8 %r55, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 55), align 1
756 store i8 %r56, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 56), align 1
757 store i8 %r57, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 57), align 1
758 store i8 %r58, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 58), align 1
759 store i8 %r59, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 59), align 1
760 store i8 %r60, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 60), align 1
761 store i8 %r61, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 61), align 1
762 store i8 %r62, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 62), align 1
763 store i8 %r63, ptr getelementptr inbounds ([64 x i8], ptr @d8, i32 0, i64 63), align 1
767 define void @fshl_v2i32() {
768 ; SSE-LABEL: @fshl_v2i32(
769 ; SSE-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
770 ; SSE-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
771 ; SSE-NEXT: [[B0:%.*]] = load i32, ptr @b32, align 4
772 ; SSE-NEXT: [[B1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1), align 4
773 ; SSE-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 [[B0]])
774 ; SSE-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 [[B1]])
775 ; SSE-NEXT: store i32 [[R0]], ptr @d32, align 4
776 ; SSE-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
779 ; AVX1-LABEL: @fshl_v2i32(
780 ; AVX1-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
781 ; AVX1-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
782 ; AVX1-NEXT: [[B0:%.*]] = load i32, ptr @b32, align 4
783 ; AVX1-NEXT: [[B1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1), align 4
784 ; AVX1-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 [[B0]])
785 ; AVX1-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 [[B1]])
786 ; AVX1-NEXT: store i32 [[R0]], ptr @d32, align 4
787 ; AVX1-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
788 ; AVX1-NEXT: ret void
790 ; AVX2-LABEL: @fshl_v2i32(
791 ; AVX2-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
792 ; AVX2-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
793 ; AVX2-NEXT: [[B0:%.*]] = load i32, ptr @b32, align 4
794 ; AVX2-NEXT: [[B1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1), align 4
795 ; AVX2-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 [[B0]])
796 ; AVX2-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 [[B1]])
797 ; AVX2-NEXT: store i32 [[R0]], ptr @d32, align 4
798 ; AVX2-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
799 ; AVX2-NEXT: ret void
801 ; AVX256-LABEL: @fshl_v2i32(
802 ; AVX256-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr @a32, align 4
803 ; AVX256-NEXT: [[TMP2:%.*]] = load <2 x i32>, ptr @b32, align 4
804 ; AVX256-NEXT: [[TMP3:%.*]] = call <2 x i32> @llvm.fshl.v2i32(<2 x i32> [[TMP1]], <2 x i32> [[TMP1]], <2 x i32> [[TMP2]])
805 ; AVX256-NEXT: store <2 x i32> [[TMP3]], ptr @d32, align 4
806 ; AVX256-NEXT: ret void
808 ; AVX512-LABEL: @fshl_v2i32(
809 ; AVX512-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr @a32, align 4
810 ; AVX512-NEXT: [[TMP2:%.*]] = load <2 x i32>, ptr @b32, align 4
811 ; AVX512-NEXT: [[TMP3:%.*]] = call <2 x i32> @llvm.fshl.v2i32(<2 x i32> [[TMP1]], <2 x i32> [[TMP1]], <2 x i32> [[TMP2]])
812 ; AVX512-NEXT: store <2 x i32> [[TMP3]], ptr @d32, align 4
813 ; AVX512-NEXT: ret void
815 %a0 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 0 ), align 4
816 %a1 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1 ), align 4
817 %b0 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 0 ), align 4
818 %b1 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1 ), align 4
819 %r0 = call i32 @llvm.fshl.i32(i32 %a0 , i32 %a0 , i32 %b0 )
820 %r1 = call i32 @llvm.fshl.i32(i32 %a1 , i32 %a1 , i32 %b1 )
821 store i32 %r0 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 0 ), align 4
822 store i32 %r1 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1 ), align 4
827 define void @fshl_v2i32_uniformconst() {
828 ; SSE-LABEL: @fshl_v2i32_uniformconst(
829 ; SSE-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
830 ; SSE-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
831 ; SSE-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 1)
832 ; SSE-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 1)
833 ; SSE-NEXT: store i32 [[R0]], ptr @d32, align 4
834 ; SSE-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
837 ; AVX1-LABEL: @fshl_v2i32_uniformconst(
838 ; AVX1-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
839 ; AVX1-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
840 ; AVX1-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 1)
841 ; AVX1-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 1)
842 ; AVX1-NEXT: store i32 [[R0]], ptr @d32, align 4
843 ; AVX1-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
844 ; AVX1-NEXT: ret void
846 ; AVX2-LABEL: @fshl_v2i32_uniformconst(
847 ; AVX2-NEXT: [[A0:%.*]] = load i32, ptr @a32, align 4
848 ; AVX2-NEXT: [[A1:%.*]] = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1), align 4
849 ; AVX2-NEXT: [[R0:%.*]] = call i32 @llvm.fshl.i32(i32 [[A0]], i32 [[A0]], i32 1)
850 ; AVX2-NEXT: [[R1:%.*]] = call i32 @llvm.fshl.i32(i32 [[A1]], i32 [[A1]], i32 1)
851 ; AVX2-NEXT: store i32 [[R0]], ptr @d32, align 4
852 ; AVX2-NEXT: store i32 [[R1]], ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1), align 4
853 ; AVX2-NEXT: ret void
855 ; AVX256-LABEL: @fshl_v2i32_uniformconst(
856 ; AVX256-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr @a32, align 4
857 ; AVX256-NEXT: [[TMP2:%.*]] = call <2 x i32> @llvm.fshl.v2i32(<2 x i32> [[TMP1]], <2 x i32> [[TMP1]], <2 x i32> splat (i32 1))
858 ; AVX256-NEXT: store <2 x i32> [[TMP2]], ptr @d32, align 4
859 ; AVX256-NEXT: ret void
861 ; AVX512-LABEL: @fshl_v2i32_uniformconst(
862 ; AVX512-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr @a32, align 4
863 ; AVX512-NEXT: [[TMP2:%.*]] = call <2 x i32> @llvm.fshl.v2i32(<2 x i32> [[TMP1]], <2 x i32> [[TMP1]], <2 x i32> splat (i32 1))
864 ; AVX512-NEXT: store <2 x i32> [[TMP2]], ptr @d32, align 4
865 ; AVX512-NEXT: ret void
867 %a0 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 0 ), align 4
868 %a1 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1 ), align 4
869 %r0 = call i32 @llvm.fshl.i32(i32 %a0 , i32 %a0 , i32 1 )
870 %r1 = call i32 @llvm.fshl.i32(i32 %a1 , i32 %a1 , i32 1 )
871 store i32 %r0 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 0 ), align 4
872 store i32 %r1 , ptr getelementptr inbounds ([16 x i32], ptr @d32, i32 0, i64 1 ), align 4