xref: /llvm-project/llvm/test/Transforms/SLPVectorizer/X86/arith-add-ssat.ll (revision b2b68c241a61bbbe0e2f89cb2182f1b1bec564b6) 
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,SSE2
3; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=SSE,SLM
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=knl -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX512
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=skx -mattr=+prefer-256-bit -passes=slp-vectorizer -S | FileCheck %s --check-prefixes=AVX
9
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@a32 = common global [16 x i32] zeroinitializer, align 64
14@b32 = common global [16 x i32] zeroinitializer, align 64
15@c32 = common global [16 x i32] zeroinitializer, align 64
16@a16 = common global [32 x i16] zeroinitializer, align 64
17@b16 = common global [32 x i16] zeroinitializer, align 64
18@c16 = common global [32 x i16] zeroinitializer, align 64
19@a8  = common global [64 x i8] zeroinitializer, align 64
20@b8  = common global [64 x i8] zeroinitializer, align 64
21@c8  = common global [64 x i8] zeroinitializer, align 64
22
23declare i64 @llvm.sadd.sat.i64(i64, i64)
24declare i32 @llvm.sadd.sat.i32(i32, i32)
25declare i16 @llvm.sadd.sat.i16(i16, i16)
26declare i8  @llvm.sadd.sat.i8 (i8 , i8 )
27
28define void @add_v8i64() {
29; SSE2-LABEL: @add_v8i64(
30; SSE2-NEXT:    [[A0:%.*]] = load i64, ptr @a64, align 8
31; SSE2-NEXT:    [[A1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
32; SSE2-NEXT:    [[A2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
33; SSE2-NEXT:    [[A3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
34; SSE2-NEXT:    [[A4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
35; SSE2-NEXT:    [[A5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
36; SSE2-NEXT:    [[A6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
37; SSE2-NEXT:    [[A7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
38; SSE2-NEXT:    [[B0:%.*]] = load i64, ptr @b64, align 8
39; SSE2-NEXT:    [[B1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
40; SSE2-NEXT:    [[B2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
41; SSE2-NEXT:    [[B3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
42; SSE2-NEXT:    [[B4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
43; SSE2-NEXT:    [[B5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
44; SSE2-NEXT:    [[B6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
45; SSE2-NEXT:    [[B7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
46; SSE2-NEXT:    [[R0:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A0]], i64 [[B0]])
47; SSE2-NEXT:    [[R1:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A1]], i64 [[B1]])
48; SSE2-NEXT:    [[R2:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A2]], i64 [[B2]])
49; SSE2-NEXT:    [[R3:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A3]], i64 [[B3]])
50; SSE2-NEXT:    [[R4:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A4]], i64 [[B4]])
51; SSE2-NEXT:    [[R5:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A5]], i64 [[B5]])
52; SSE2-NEXT:    [[R6:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A6]], i64 [[B6]])
53; SSE2-NEXT:    [[R7:%.*]] = call i64 @llvm.sadd.sat.i64(i64 [[A7]], i64 [[B7]])
54; SSE2-NEXT:    store i64 [[R0]], ptr @c64, align 8
55; SSE2-NEXT:    store i64 [[R1]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 1), align 8
56; SSE2-NEXT:    store i64 [[R2]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 2), align 8
57; SSE2-NEXT:    store i64 [[R3]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 3), align 8
58; SSE2-NEXT:    store i64 [[R4]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
59; SSE2-NEXT:    store i64 [[R5]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 5), align 8
60; SSE2-NEXT:    store i64 [[R6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 6), align 8
61; SSE2-NEXT:    store i64 [[R7]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 7), align 8
62; SSE2-NEXT:    ret void
63;
64; SLM-LABEL: @add_v8i64(
65; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, ptr @a64, align 8
66; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, ptr @b64, align 8
67; SLM-NEXT:    [[TMP3:%.*]] = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP2]])
68; SLM-NEXT:    store <2 x i64> [[TMP3]], ptr @c64, align 8
69; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
70; SLM-NEXT:    [[TMP5:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
71; SLM-NEXT:    [[TMP6:%.*]] = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP5]])
72; SLM-NEXT:    store <2 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 2), align 8
73; SLM-NEXT:    [[TMP7:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
74; SLM-NEXT:    [[TMP8:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
75; SLM-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> [[TMP7]], <2 x i64> [[TMP8]])
76; SLM-NEXT:    store <2 x i64> [[TMP9]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
77; SLM-NEXT:    [[TMP10:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
78; SLM-NEXT:    [[TMP11:%.*]] = load <2 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
79; SLM-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> [[TMP10]], <2 x i64> [[TMP11]])
80; SLM-NEXT:    store <2 x i64> [[TMP12]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 6), align 8
81; SLM-NEXT:    ret void
82;
83; AVX-LABEL: @add_v8i64(
84; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
85; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, ptr @b64, align 8
86; AVX-NEXT:    [[TMP3:%.*]] = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP2]])
87; AVX-NEXT:    store <4 x i64> [[TMP3]], ptr @c64, align 8
88; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
89; AVX-NEXT:    [[TMP5:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
90; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> [[TMP4]], <4 x i64> [[TMP5]])
91; AVX-NEXT:    store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
92; AVX-NEXT:    ret void
93;
94; AVX512-LABEL: @add_v8i64(
95; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, ptr @a64, align 8
96; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, ptr @b64, align 8
97; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
98; AVX512-NEXT:    store <8 x i64> [[TMP3]], ptr @c64, align 8
99; AVX512-NEXT:    ret void
100;
101; AVX256BW-LABEL: @add_v8i64(
102; AVX256BW-NEXT:    [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
103; AVX256BW-NEXT:    [[TMP2:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
104; AVX256BW-NEXT:    [[TMP3:%.*]] = load <4 x i64>, ptr @b64, align 8
105; AVX256BW-NEXT:    [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
106; AVX256BW-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
107; AVX256BW-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
108; AVX256BW-NEXT:    store <4 x i64> [[TMP5]], ptr @c64, align 8
109; AVX256BW-NEXT:    store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
110; AVX256BW-NEXT:    ret void
111  %a0 = load i64, ptr @a64, align 8
112  %a1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
113  %a2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
114  %a3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
115  %a4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
116  %a5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
117  %a6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
118  %a7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
119  %b0 = load i64, ptr @b64, align 8
120  %b1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
121  %b2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
122  %b3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
123  %b4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
124  %b5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
125  %b6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
126  %b7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
127  %r0 = call i64 @llvm.sadd.sat.i64(i64 %a0, i64 %b0)
128  %r1 = call i64 @llvm.sadd.sat.i64(i64 %a1, i64 %b1)
129  %r2 = call i64 @llvm.sadd.sat.i64(i64 %a2, i64 %b2)
130  %r3 = call i64 @llvm.sadd.sat.i64(i64 %a3, i64 %b3)
131  %r4 = call i64 @llvm.sadd.sat.i64(i64 %a4, i64 %b4)
132  %r5 = call i64 @llvm.sadd.sat.i64(i64 %a5, i64 %b5)
133  %r6 = call i64 @llvm.sadd.sat.i64(i64 %a6, i64 %b6)
134  %r7 = call i64 @llvm.sadd.sat.i64(i64 %a7, i64 %b7)
135  store i64 %r0, ptr @c64, align 8
136  store i64 %r1, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 1), align 8
137  store i64 %r2, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 2), align 8
138  store i64 %r3, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 3), align 8
139  store i64 %r4, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
140  store i64 %r5, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 5), align 8
141  store i64 %r6, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 6), align 8
142  store i64 %r7, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 7), align 8
143  ret void
144}
145
146define void @add_v16i32() {
147; SSE-LABEL: @add_v16i32(
148; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, ptr @a32, align 4
149; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, ptr @b32, align 4
150; SSE-NEXT:    [[TMP3:%.*]] = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP2]])
151; SSE-NEXT:    store <4 x i32> [[TMP3]], ptr @c32, align 4
152; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4), align 4
153; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4), align 4
154; SSE-NEXT:    [[TMP6:%.*]] = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP5]])
155; SSE-NEXT:    store <4 x i32> [[TMP6]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 4), align 4
156; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
157; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
158; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> [[TMP7]], <4 x i32> [[TMP8]])
159; SSE-NEXT:    store <4 x i32> [[TMP9]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8), align 4
160; SSE-NEXT:    [[TMP10:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
161; SSE-NEXT:    [[TMP11:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
162; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> [[TMP10]], <4 x i32> [[TMP11]])
163; SSE-NEXT:    store <4 x i32> [[TMP12]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 12), align 4
164; SSE-NEXT:    ret void
165;
166; AVX-LABEL: @add_v16i32(
167; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, ptr @a32, align 4
168; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, ptr @b32, align 4
169; AVX-NEXT:    [[TMP3:%.*]] = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP2]])
170; AVX-NEXT:    store <8 x i32> [[TMP3]], ptr @c32, align 4
171; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
172; AVX-NEXT:    [[TMP5:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
173; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> [[TMP4]], <8 x i32> [[TMP5]])
174; AVX-NEXT:    store <8 x i32> [[TMP6]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8), align 4
175; AVX-NEXT:    ret void
176;
177; AVX512-LABEL: @add_v16i32(
178; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, ptr @a32, align 4
179; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, ptr @b32, align 4
180; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
181; AVX512-NEXT:    store <16 x i32> [[TMP3]], ptr @c32, align 4
182; AVX512-NEXT:    ret void
183;
184  %a0  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 0 ), align 4
185  %a1  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1 ), align 4
186  %a2  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 2 ), align 4
187  %a3  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 3 ), align 4
188  %a4  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4 ), align 4
189  %a5  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 5 ), align 4
190  %a6  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 6 ), align 4
191  %a7  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 7 ), align 4
192  %a8  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8 ), align 4
193  %a9  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 9 ), align 4
194  %a10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 10), align 4
195  %a11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 11), align 4
196  %a12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
197  %a13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 13), align 4
198  %a14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 14), align 4
199  %a15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 15), align 4
200  %b0  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 0 ), align 4
201  %b1  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1 ), align 4
202  %b2  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 2 ), align 4
203  %b3  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 3 ), align 4
204  %b4  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4 ), align 4
205  %b5  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 5 ), align 4
206  %b6  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 6 ), align 4
207  %b7  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 7 ), align 4
208  %b8  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8 ), align 4
209  %b9  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 9 ), align 4
210  %b10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 10), align 4
211  %b11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 11), align 4
212  %b12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
213  %b13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 13), align 4
214  %b14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 14), align 4
215  %b15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 15), align 4
216  %r0  = call i32 @llvm.sadd.sat.i32(i32 %a0 , i32 %b0 )
217  %r1  = call i32 @llvm.sadd.sat.i32(i32 %a1 , i32 %b1 )
218  %r2  = call i32 @llvm.sadd.sat.i32(i32 %a2 , i32 %b2 )
219  %r3  = call i32 @llvm.sadd.sat.i32(i32 %a3 , i32 %b3 )
220  %r4  = call i32 @llvm.sadd.sat.i32(i32 %a4 , i32 %b4 )
221  %r5  = call i32 @llvm.sadd.sat.i32(i32 %a5 , i32 %b5 )
222  %r6  = call i32 @llvm.sadd.sat.i32(i32 %a6 , i32 %b6 )
223  %r7  = call i32 @llvm.sadd.sat.i32(i32 %a7 , i32 %b7 )
224  %r8  = call i32 @llvm.sadd.sat.i32(i32 %a8 , i32 %b8 )
225  %r9  = call i32 @llvm.sadd.sat.i32(i32 %a9 , i32 %b9 )
226  %r10 = call i32 @llvm.sadd.sat.i32(i32 %a10, i32 %b10)
227  %r11 = call i32 @llvm.sadd.sat.i32(i32 %a11, i32 %b11)
228  %r12 = call i32 @llvm.sadd.sat.i32(i32 %a12, i32 %b12)
229  %r13 = call i32 @llvm.sadd.sat.i32(i32 %a13, i32 %b13)
230  %r14 = call i32 @llvm.sadd.sat.i32(i32 %a14, i32 %b14)
231  %r15 = call i32 @llvm.sadd.sat.i32(i32 %a15, i32 %b15)
232  store i32 %r0 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 0 ), align 4
233  store i32 %r1 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 1 ), align 4
234  store i32 %r2 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 2 ), align 4
235  store i32 %r3 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 3 ), align 4
236  store i32 %r4 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 4 ), align 4
237  store i32 %r5 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 5 ), align 4
238  store i32 %r6 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 6 ), align 4
239  store i32 %r7 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 7 ), align 4
240  store i32 %r8 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8 ), align 4
241  store i32 %r9 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 9 ), align 4
242  store i32 %r10, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 10), align 4
243  store i32 %r11, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 11), align 4
244  store i32 %r12, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 12), align 4
245  store i32 %r13, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 13), align 4
246  store i32 %r14, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 14), align 4
247  store i32 %r15, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 15), align 4
248  ret void
249}
250
251define void @add_v32i16() {
252; SSE-LABEL: @add_v32i16(
253; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, ptr @a16, align 2
254; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, ptr @b16, align 2
255; SSE-NEXT:    [[TMP3:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP2]])
256; SSE-NEXT:    store <8 x i16> [[TMP3]], ptr @c16, align 2
257; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 8), align 2
258; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 8), align 2
259; SSE-NEXT:    [[TMP6:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP5]])
260; SSE-NEXT:    store <8 x i16> [[TMP6]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 8), align 2
261; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
262; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
263; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[TMP7]], <8 x i16> [[TMP8]])
264; SSE-NEXT:    store <8 x i16> [[TMP9]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
265; SSE-NEXT:    [[TMP10:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 24), align 2
266; SSE-NEXT:    [[TMP11:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 24), align 2
267; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[TMP10]], <8 x i16> [[TMP11]])
268; SSE-NEXT:    store <8 x i16> [[TMP12]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 24), align 2
269; SSE-NEXT:    ret void
270;
271; AVX-LABEL: @add_v32i16(
272; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, ptr @a16, align 2
273; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, ptr @b16, align 2
274; AVX-NEXT:    [[TMP3:%.*]] = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP2]])
275; AVX-NEXT:    store <16 x i16> [[TMP3]], ptr @c16, align 2
276; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
277; AVX-NEXT:    [[TMP5:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
278; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> [[TMP4]], <16 x i16> [[TMP5]])
279; AVX-NEXT:    store <16 x i16> [[TMP6]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
280; AVX-NEXT:    ret void
281;
282; AVX512-LABEL: @add_v32i16(
283; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, ptr @a16, align 2
284; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, ptr @b16, align 2
285; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.sadd.sat.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
286; AVX512-NEXT:    store <32 x i16> [[TMP3]], ptr @c16, align 2
287; AVX512-NEXT:    ret void
288;
289  %a0  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 0 ), align 2
290  %a1  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 1 ), align 2
291  %a2  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 2 ), align 2
292  %a3  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 3 ), align 2
293  %a4  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 4 ), align 2
294  %a5  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 5 ), align 2
295  %a6  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 6 ), align 2
296  %a7  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 7 ), align 2
297  %a8  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 8 ), align 2
298  %a9  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 9 ), align 2
299  %a10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 10), align 2
300  %a11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 11), align 2
301  %a12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 12), align 2
302  %a13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 13), align 2
303  %a14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 14), align 2
304  %a15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 15), align 2
305  %a16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
306  %a17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 17), align 2
307  %a18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 18), align 2
308  %a19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 19), align 2
309  %a20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 20), align 2
310  %a21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 21), align 2
311  %a22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 22), align 2
312  %a23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 23), align 2
313  %a24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 24), align 2
314  %a25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 25), align 2
315  %a26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 26), align 2
316  %a27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 27), align 2
317  %a28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 28), align 2
318  %a29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 29), align 2
319  %a30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 30), align 2
320  %a31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 31), align 2
321  %b0  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 0 ), align 2
322  %b1  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 1 ), align 2
323  %b2  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 2 ), align 2
324  %b3  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 3 ), align 2
325  %b4  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 4 ), align 2
326  %b5  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 5 ), align 2
327  %b6  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 6 ), align 2
328  %b7  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 7 ), align 2
329  %b8  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 8 ), align 2
330  %b9  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 9 ), align 2
331  %b10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 10), align 2
332  %b11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 11), align 2
333  %b12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 12), align 2
334  %b13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 13), align 2
335  %b14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 14), align 2
336  %b15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 15), align 2
337  %b16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
338  %b17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 17), align 2
339  %b18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 18), align 2
340  %b19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 19), align 2
341  %b20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 20), align 2
342  %b21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 21), align 2
343  %b22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 22), align 2
344  %b23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 23), align 2
345  %b24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 24), align 2
346  %b25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 25), align 2
347  %b26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 26), align 2
348  %b27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 27), align 2
349  %b28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 28), align 2
350  %b29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 29), align 2
351  %b30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 30), align 2
352  %b31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 31), align 2
353  %r0  = call i16 @llvm.sadd.sat.i16(i16 %a0 , i16 %b0 )
354  %r1  = call i16 @llvm.sadd.sat.i16(i16 %a1 , i16 %b1 )
355  %r2  = call i16 @llvm.sadd.sat.i16(i16 %a2 , i16 %b2 )
356  %r3  = call i16 @llvm.sadd.sat.i16(i16 %a3 , i16 %b3 )
357  %r4  = call i16 @llvm.sadd.sat.i16(i16 %a4 , i16 %b4 )
358  %r5  = call i16 @llvm.sadd.sat.i16(i16 %a5 , i16 %b5 )
359  %r6  = call i16 @llvm.sadd.sat.i16(i16 %a6 , i16 %b6 )
360  %r7  = call i16 @llvm.sadd.sat.i16(i16 %a7 , i16 %b7 )
361  %r8  = call i16 @llvm.sadd.sat.i16(i16 %a8 , i16 %b8 )
362  %r9  = call i16 @llvm.sadd.sat.i16(i16 %a9 , i16 %b9 )
363  %r10 = call i16 @llvm.sadd.sat.i16(i16 %a10, i16 %b10)
364  %r11 = call i16 @llvm.sadd.sat.i16(i16 %a11, i16 %b11)
365  %r12 = call i16 @llvm.sadd.sat.i16(i16 %a12, i16 %b12)
366  %r13 = call i16 @llvm.sadd.sat.i16(i16 %a13, i16 %b13)
367  %r14 = call i16 @llvm.sadd.sat.i16(i16 %a14, i16 %b14)
368  %r15 = call i16 @llvm.sadd.sat.i16(i16 %a15, i16 %b15)
369  %r16 = call i16 @llvm.sadd.sat.i16(i16 %a16, i16 %b16)
370  %r17 = call i16 @llvm.sadd.sat.i16(i16 %a17, i16 %b17)
371  %r18 = call i16 @llvm.sadd.sat.i16(i16 %a18, i16 %b18)
372  %r19 = call i16 @llvm.sadd.sat.i16(i16 %a19, i16 %b19)
373  %r20 = call i16 @llvm.sadd.sat.i16(i16 %a20, i16 %b20)
374  %r21 = call i16 @llvm.sadd.sat.i16(i16 %a21, i16 %b21)
375  %r22 = call i16 @llvm.sadd.sat.i16(i16 %a22, i16 %b22)
376  %r23 = call i16 @llvm.sadd.sat.i16(i16 %a23, i16 %b23)
377  %r24 = call i16 @llvm.sadd.sat.i16(i16 %a24, i16 %b24)
378  %r25 = call i16 @llvm.sadd.sat.i16(i16 %a25, i16 %b25)
379  %r26 = call i16 @llvm.sadd.sat.i16(i16 %a26, i16 %b26)
380  %r27 = call i16 @llvm.sadd.sat.i16(i16 %a27, i16 %b27)
381  %r28 = call i16 @llvm.sadd.sat.i16(i16 %a28, i16 %b28)
382  %r29 = call i16 @llvm.sadd.sat.i16(i16 %a29, i16 %b29)
383  %r30 = call i16 @llvm.sadd.sat.i16(i16 %a30, i16 %b30)
384  %r31 = call i16 @llvm.sadd.sat.i16(i16 %a31, i16 %b31)
385  store i16 %r0 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 0 ), align 2
386  store i16 %r1 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 1 ), align 2
387  store i16 %r2 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 2 ), align 2
388  store i16 %r3 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 3 ), align 2
389  store i16 %r4 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 4 ), align 2
390  store i16 %r5 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 5 ), align 2
391  store i16 %r6 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 6 ), align 2
392  store i16 %r7 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 7 ), align 2
393  store i16 %r8 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 8 ), align 2
394  store i16 %r9 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 9 ), align 2
395  store i16 %r10, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 10), align 2
396  store i16 %r11, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 11), align 2
397  store i16 %r12, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 12), align 2
398  store i16 %r13, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 13), align 2
399  store i16 %r14, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 14), align 2
400  store i16 %r15, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 15), align 2
401  store i16 %r16, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
402  store i16 %r17, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 17), align 2
403  store i16 %r18, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 18), align 2
404  store i16 %r19, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 19), align 2
405  store i16 %r20, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 20), align 2
406  store i16 %r21, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 21), align 2
407  store i16 %r22, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 22), align 2
408  store i16 %r23, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 23), align 2
409  store i16 %r24, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 24), align 2
410  store i16 %r25, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 25), align 2
411  store i16 %r26, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 26), align 2
412  store i16 %r27, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 27), align 2
413  store i16 %r28, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 28), align 2
414  store i16 %r29, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 29), align 2
415  store i16 %r30, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 30), align 2
416  store i16 %r31, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 31), align 2
417  ret void
418}
419
420define void @add_v64i8() {
421; SSE-LABEL: @add_v64i8(
422; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, ptr @a8, align 1
423; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, ptr @b8, align 1
424; SSE-NEXT:    [[TMP3:%.*]] = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]])
425; SSE-NEXT:    store <16 x i8> [[TMP3]], ptr @c8, align 1
426; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
427; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
428; SSE-NEXT:    [[TMP6:%.*]] = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP5]])
429; SSE-NEXT:    store <16 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 16), align 1
430; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
431; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
432; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> [[TMP7]], <16 x i8> [[TMP8]])
433; SSE-NEXT:    store <16 x i8> [[TMP9]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
434; SSE-NEXT:    [[TMP10:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
435; SSE-NEXT:    [[TMP11:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
436; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> [[TMP10]], <16 x i8> [[TMP11]])
437; SSE-NEXT:    store <16 x i8> [[TMP12]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 48), align 1
438; SSE-NEXT:    ret void
439;
440; AVX-LABEL: @add_v64i8(
441; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, ptr @a8, align 1
442; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, ptr @b8, align 1
443; AVX-NEXT:    [[TMP3:%.*]] = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP2]])
444; AVX-NEXT:    store <32 x i8> [[TMP3]], ptr @c8, align 1
445; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
446; AVX-NEXT:    [[TMP5:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
447; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> [[TMP4]], <32 x i8> [[TMP5]])
448; AVX-NEXT:    store <32 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
449; AVX-NEXT:    ret void
450;
451; AVX512-LABEL: @add_v64i8(
452; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, ptr @a8, align 1
453; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, ptr @b8, align 1
454; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.sadd.sat.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
455; AVX512-NEXT:    store <64 x i8> [[TMP3]], ptr @c8, align 1
456; AVX512-NEXT:    ret void
457;
458  %a0  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 0 ), align 1
459  %a1  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 1 ), align 1
460  %a2  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 2 ), align 1
461  %a3  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 3 ), align 1
462  %a4  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 4 ), align 1
463  %a5  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 5 ), align 1
464  %a6  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 6 ), align 1
465  %a7  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 7 ), align 1
466  %a8  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 8 ), align 1
467  %a9  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 9 ), align 1
468  %a10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 10), align 1
469  %a11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 11), align 1
470  %a12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 12), align 1
471  %a13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 13), align 1
472  %a14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 14), align 1
473  %a15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 15), align 1
474  %a16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
475  %a17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 17), align 1
476  %a18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 18), align 1
477  %a19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 19), align 1
478  %a20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 20), align 1
479  %a21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 21), align 1
480  %a22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 22), align 1
481  %a23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 23), align 1
482  %a24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 24), align 1
483  %a25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 25), align 1
484  %a26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 26), align 1
485  %a27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 27), align 1
486  %a28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 28), align 1
487  %a29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 29), align 1
488  %a30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 30), align 1
489  %a31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 31), align 1
490  %a32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
491  %a33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 33), align 1
492  %a34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 34), align 1
493  %a35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 35), align 1
494  %a36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 36), align 1
495  %a37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 37), align 1
496  %a38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 38), align 1
497  %a39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 39), align 1
498  %a40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 40), align 1
499  %a41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 41), align 1
500  %a42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 42), align 1
501  %a43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 43), align 1
502  %a44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 44), align 1
503  %a45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 45), align 1
504  %a46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 46), align 1
505  %a47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 47), align 1
506  %a48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
507  %a49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 49), align 1
508  %a50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 50), align 1
509  %a51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 51), align 1
510  %a52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 52), align 1
511  %a53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 53), align 1
512  %a54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 54), align 1
513  %a55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 55), align 1
514  %a56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 56), align 1
515  %a57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 57), align 1
516  %a58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 58), align 1
517  %a59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 59), align 1
518  %a60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 60), align 1
519  %a61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 61), align 1
520  %a62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 62), align 1
521  %a63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 63), align 1
522  %b0  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 0 ), align 1
523  %b1  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 1 ), align 1
524  %b2  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 2 ), align 1
525  %b3  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 3 ), align 1
526  %b4  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 4 ), align 1
527  %b5  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 5 ), align 1
528  %b6  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 6 ), align 1
529  %b7  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 7 ), align 1
530  %b8  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 8 ), align 1
531  %b9  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 9 ), align 1
532  %b10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 10), align 1
533  %b11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 11), align 1
534  %b12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 12), align 1
535  %b13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 13), align 1
536  %b14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 14), align 1
537  %b15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 15), align 1
538  %b16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
539  %b17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 17), align 1
540  %b18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 18), align 1
541  %b19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 19), align 1
542  %b20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 20), align 1
543  %b21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 21), align 1
544  %b22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 22), align 1
545  %b23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 23), align 1
546  %b24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 24), align 1
547  %b25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 25), align 1
548  %b26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 26), align 1
549  %b27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 27), align 1
550  %b28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 28), align 1
551  %b29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 29), align 1
552  %b30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 30), align 1
553  %b31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 31), align 1
554  %b32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
555  %b33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 33), align 1
556  %b34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 34), align 1
557  %b35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 35), align 1
558  %b36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 36), align 1
559  %b37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 37), align 1
560  %b38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 38), align 1
561  %b39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 39), align 1
562  %b40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 40), align 1
563  %b41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 41), align 1
564  %b42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 42), align 1
565  %b43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 43), align 1
566  %b44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 44), align 1
567  %b45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 45), align 1
568  %b46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 46), align 1
569  %b47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 47), align 1
570  %b48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
571  %b49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 49), align 1
572  %b50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 50), align 1
573  %b51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 51), align 1
574  %b52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 52), align 1
575  %b53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 53), align 1
576  %b54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 54), align 1
577  %b55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 55), align 1
578  %b56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 56), align 1
579  %b57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 57), align 1
580  %b58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 58), align 1
581  %b59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 59), align 1
582  %b60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 60), align 1
583  %b61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 61), align 1
584  %b62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 62), align 1
585  %b63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 63), align 1
586  %r0  = call i8 @llvm.sadd.sat.i8(i8 %a0 , i8 %b0 )
587  %r1  = call i8 @llvm.sadd.sat.i8(i8 %a1 , i8 %b1 )
588  %r2  = call i8 @llvm.sadd.sat.i8(i8 %a2 , i8 %b2 )
589  %r3  = call i8 @llvm.sadd.sat.i8(i8 %a3 , i8 %b3 )
590  %r4  = call i8 @llvm.sadd.sat.i8(i8 %a4 , i8 %b4 )
591  %r5  = call i8 @llvm.sadd.sat.i8(i8 %a5 , i8 %b5 )
592  %r6  = call i8 @llvm.sadd.sat.i8(i8 %a6 , i8 %b6 )
593  %r7  = call i8 @llvm.sadd.sat.i8(i8 %a7 , i8 %b7 )
594  %r8  = call i8 @llvm.sadd.sat.i8(i8 %a8 , i8 %b8 )
595  %r9  = call i8 @llvm.sadd.sat.i8(i8 %a9 , i8 %b9 )
596  %r10 = call i8 @llvm.sadd.sat.i8(i8 %a10, i8 %b10)
597  %r11 = call i8 @llvm.sadd.sat.i8(i8 %a11, i8 %b11)
598  %r12 = call i8 @llvm.sadd.sat.i8(i8 %a12, i8 %b12)
599  %r13 = call i8 @llvm.sadd.sat.i8(i8 %a13, i8 %b13)
600  %r14 = call i8 @llvm.sadd.sat.i8(i8 %a14, i8 %b14)
601  %r15 = call i8 @llvm.sadd.sat.i8(i8 %a15, i8 %b15)
602  %r16 = call i8 @llvm.sadd.sat.i8(i8 %a16, i8 %b16)
603  %r17 = call i8 @llvm.sadd.sat.i8(i8 %a17, i8 %b17)
604  %r18 = call i8 @llvm.sadd.sat.i8(i8 %a18, i8 %b18)
605  %r19 = call i8 @llvm.sadd.sat.i8(i8 %a19, i8 %b19)
606  %r20 = call i8 @llvm.sadd.sat.i8(i8 %a20, i8 %b20)
607  %r21 = call i8 @llvm.sadd.sat.i8(i8 %a21, i8 %b21)
608  %r22 = call i8 @llvm.sadd.sat.i8(i8 %a22, i8 %b22)
609  %r23 = call i8 @llvm.sadd.sat.i8(i8 %a23, i8 %b23)
610  %r24 = call i8 @llvm.sadd.sat.i8(i8 %a24, i8 %b24)
611  %r25 = call i8 @llvm.sadd.sat.i8(i8 %a25, i8 %b25)
612  %r26 = call i8 @llvm.sadd.sat.i8(i8 %a26, i8 %b26)
613  %r27 = call i8 @llvm.sadd.sat.i8(i8 %a27, i8 %b27)
614  %r28 = call i8 @llvm.sadd.sat.i8(i8 %a28, i8 %b28)
615  %r29 = call i8 @llvm.sadd.sat.i8(i8 %a29, i8 %b29)
616  %r30 = call i8 @llvm.sadd.sat.i8(i8 %a30, i8 %b30)
617  %r31 = call i8 @llvm.sadd.sat.i8(i8 %a31, i8 %b31)
618  %r32 = call i8 @llvm.sadd.sat.i8(i8 %a32, i8 %b32)
619  %r33 = call i8 @llvm.sadd.sat.i8(i8 %a33, i8 %b33)
620  %r34 = call i8 @llvm.sadd.sat.i8(i8 %a34, i8 %b34)
621  %r35 = call i8 @llvm.sadd.sat.i8(i8 %a35, i8 %b35)
622  %r36 = call i8 @llvm.sadd.sat.i8(i8 %a36, i8 %b36)
623  %r37 = call i8 @llvm.sadd.sat.i8(i8 %a37, i8 %b37)
624  %r38 = call i8 @llvm.sadd.sat.i8(i8 %a38, i8 %b38)
625  %r39 = call i8 @llvm.sadd.sat.i8(i8 %a39, i8 %b39)
626  %r40 = call i8 @llvm.sadd.sat.i8(i8 %a40, i8 %b40)
627  %r41 = call i8 @llvm.sadd.sat.i8(i8 %a41, i8 %b41)
628  %r42 = call i8 @llvm.sadd.sat.i8(i8 %a42, i8 %b42)
629  %r43 = call i8 @llvm.sadd.sat.i8(i8 %a43, i8 %b43)
630  %r44 = call i8 @llvm.sadd.sat.i8(i8 %a44, i8 %b44)
631  %r45 = call i8 @llvm.sadd.sat.i8(i8 %a45, i8 %b45)
632  %r46 = call i8 @llvm.sadd.sat.i8(i8 %a46, i8 %b46)
633  %r47 = call i8 @llvm.sadd.sat.i8(i8 %a47, i8 %b47)
634  %r48 = call i8 @llvm.sadd.sat.i8(i8 %a48, i8 %b48)
635  %r49 = call i8 @llvm.sadd.sat.i8(i8 %a49, i8 %b49)
636  %r50 = call i8 @llvm.sadd.sat.i8(i8 %a50, i8 %b50)
637  %r51 = call i8 @llvm.sadd.sat.i8(i8 %a51, i8 %b51)
638  %r52 = call i8 @llvm.sadd.sat.i8(i8 %a52, i8 %b52)
639  %r53 = call i8 @llvm.sadd.sat.i8(i8 %a53, i8 %b53)
640  %r54 = call i8 @llvm.sadd.sat.i8(i8 %a54, i8 %b54)
641  %r55 = call i8 @llvm.sadd.sat.i8(i8 %a55, i8 %b55)
642  %r56 = call i8 @llvm.sadd.sat.i8(i8 %a56, i8 %b56)
643  %r57 = call i8 @llvm.sadd.sat.i8(i8 %a57, i8 %b57)
644  %r58 = call i8 @llvm.sadd.sat.i8(i8 %a58, i8 %b58)
645  %r59 = call i8 @llvm.sadd.sat.i8(i8 %a59, i8 %b59)
646  %r60 = call i8 @llvm.sadd.sat.i8(i8 %a60, i8 %b60)
647  %r61 = call i8 @llvm.sadd.sat.i8(i8 %a61, i8 %b61)
648  %r62 = call i8 @llvm.sadd.sat.i8(i8 %a62, i8 %b62)
649  %r63 = call i8 @llvm.sadd.sat.i8(i8 %a63, i8 %b63)
650  store i8 %r0 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 0 ), align 1
651  store i8 %r1 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 1 ), align 1
652  store i8 %r2 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 2 ), align 1
653  store i8 %r3 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 3 ), align 1
654  store i8 %r4 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 4 ), align 1
655  store i8 %r5 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 5 ), align 1
656  store i8 %r6 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 6 ), align 1
657  store i8 %r7 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 7 ), align 1
658  store i8 %r8 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 8 ), align 1
659  store i8 %r9 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 9 ), align 1
660  store i8 %r10, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 10), align 1
661  store i8 %r11, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 11), align 1
662  store i8 %r12, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 12), align 1
663  store i8 %r13, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 13), align 1
664  store i8 %r14, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 14), align 1
665  store i8 %r15, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 15), align 1
666  store i8 %r16, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 16), align 1
667  store i8 %r17, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 17), align 1
668  store i8 %r18, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 18), align 1
669  store i8 %r19, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 19), align 1
670  store i8 %r20, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 20), align 1
671  store i8 %r21, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 21), align 1
672  store i8 %r22, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 22), align 1
673  store i8 %r23, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 23), align 1
674  store i8 %r24, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 24), align 1
675  store i8 %r25, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 25), align 1
676  store i8 %r26, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 26), align 1
677  store i8 %r27, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 27), align 1
678  store i8 %r28, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 28), align 1
679  store i8 %r29, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 29), align 1
680  store i8 %r30, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 30), align 1
681  store i8 %r31, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 31), align 1
682  store i8 %r32, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
683  store i8 %r33, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 33), align 1
684  store i8 %r34, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 34), align 1
685  store i8 %r35, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 35), align 1
686  store i8 %r36, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 36), align 1
687  store i8 %r37, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 37), align 1
688  store i8 %r38, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 38), align 1
689  store i8 %r39, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 39), align 1
690  store i8 %r40, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 40), align 1
691  store i8 %r41, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 41), align 1
692  store i8 %r42, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 42), align 1
693  store i8 %r43, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 43), align 1
694  store i8 %r44, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 44), align 1
695  store i8 %r45, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 45), align 1
696  store i8 %r46, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 46), align 1
697  store i8 %r47, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 47), align 1
698  store i8 %r48, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 48), align 1
699  store i8 %r49, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 49), align 1
700  store i8 %r50, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 50), align 1
701  store i8 %r51, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 51), align 1
702  store i8 %r52, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 52), align 1
703  store i8 %r53, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 53), align 1
704  store i8 %r54, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 54), align 1
705  store i8 %r55, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 55), align 1
706  store i8 %r56, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 56), align 1
707  store i8 %r57, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 57), align 1
708  store i8 %r58, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 58), align 1
709  store i8 %r59, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 59), align 1
710  store i8 %r60, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 60), align 1
711  store i8 %r61, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 61), align 1
712  store i8 %r62, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 62), align 1
713  store i8 %r63, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 63), align 1
714  ret void
715}
716;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
717; AVX1: {{.*}}
718; AVX2: {{.*}}
719