xref: /llvm-project/llvm/test/Transforms/SLPVectorizer/X86/arith-sub-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.ssub.sat.i64(i64, i64)
24declare i32 @llvm.ssub.sat.i32(i32, i32)
25declare i16 @llvm.ssub.sat.i16(i16, i16)
26declare i8  @llvm.ssub.sat.i8 (i8 , i8 )
27
28define void @sub_v8i64() {
29; SSE-LABEL: @sub_v8i64(
30; SSE-NEXT:    [[A0:%.*]] = load i64, ptr @a64, align 8
31; SSE-NEXT:    [[A1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
32; SSE-NEXT:    [[A2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
33; SSE-NEXT:    [[A3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
34; SSE-NEXT:    [[A4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
35; SSE-NEXT:    [[A5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
36; SSE-NEXT:    [[A6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
37; SSE-NEXT:    [[A7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
38; SSE-NEXT:    [[B0:%.*]] = load i64, ptr @b64, align 8
39; SSE-NEXT:    [[B1:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
40; SSE-NEXT:    [[B2:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
41; SSE-NEXT:    [[B3:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
42; SSE-NEXT:    [[B4:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
43; SSE-NEXT:    [[B5:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
44; SSE-NEXT:    [[B6:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
45; SSE-NEXT:    [[B7:%.*]] = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
46; SSE-NEXT:    [[R0:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A0]], i64 [[B0]])
47; SSE-NEXT:    [[R1:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A1]], i64 [[B1]])
48; SSE-NEXT:    [[R2:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A2]], i64 [[B2]])
49; SSE-NEXT:    [[R3:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A3]], i64 [[B3]])
50; SSE-NEXT:    [[R4:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A4]], i64 [[B4]])
51; SSE-NEXT:    [[R5:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A5]], i64 [[B5]])
52; SSE-NEXT:    [[R6:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A6]], i64 [[B6]])
53; SSE-NEXT:    [[R7:%.*]] = call i64 @llvm.ssub.sat.i64(i64 [[A7]], i64 [[B7]])
54; SSE-NEXT:    store i64 [[R0]], ptr @c64, align 8
55; SSE-NEXT:    store i64 [[R1]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 1), align 8
56; SSE-NEXT:    store i64 [[R2]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 2), align 8
57; SSE-NEXT:    store i64 [[R3]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 3), align 8
58; SSE-NEXT:    store i64 [[R4]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
59; SSE-NEXT:    store i64 [[R5]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 5), align 8
60; SSE-NEXT:    store i64 [[R6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 6), align 8
61; SSE-NEXT:    store i64 [[R7]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 7), align 8
62; SSE-NEXT:    ret void
63;
64; AVX-LABEL: @sub_v8i64(
65; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
66; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, ptr @b64, align 8
67; AVX-NEXT:    [[TMP3:%.*]] = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP2]])
68; AVX-NEXT:    store <4 x i64> [[TMP3]], ptr @c64, align 8
69; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
70; AVX-NEXT:    [[TMP5:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
71; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> [[TMP4]], <4 x i64> [[TMP5]])
72; AVX-NEXT:    store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
73; AVX-NEXT:    ret void
74;
75; AVX512-LABEL: @sub_v8i64(
76; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, ptr @a64, align 8
77; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, ptr @b64, align 8
78; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
79; AVX512-NEXT:    store <8 x i64> [[TMP3]], ptr @c64, align 8
80; AVX512-NEXT:    ret void
81;
82; AVX256BW-LABEL: @sub_v8i64(
83; AVX256BW-NEXT:    [[TMP1:%.*]] = load <4 x i64>, ptr @a64, align 8
84; AVX256BW-NEXT:    [[TMP2:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
85; AVX256BW-NEXT:    [[TMP3:%.*]] = load <4 x i64>, ptr @b64, align 8
86; AVX256BW-NEXT:    [[TMP4:%.*]] = load <4 x i64>, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
87; AVX256BW-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
88; AVX256BW-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
89; AVX256BW-NEXT:    store <4 x i64> [[TMP5]], ptr @c64, align 8
90; AVX256BW-NEXT:    store <4 x i64> [[TMP6]], ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
91; AVX256BW-NEXT:    ret void
92  %a0 = load i64, ptr @a64, align 8
93  %a1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 1), align 8
94  %a2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 2), align 8
95  %a3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 3), align 8
96  %a4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 4), align 8
97  %a5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 5), align 8
98  %a6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 6), align 8
99  %a7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @a64, i32 0, i64 7), align 8
100  %b0 = load i64, ptr @b64, align 8
101  %b1 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 1), align 8
102  %b2 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 2), align 8
103  %b3 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 3), align 8
104  %b4 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 4), align 8
105  %b5 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 5), align 8
106  %b6 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 6), align 8
107  %b7 = load i64, ptr getelementptr inbounds ([8 x i64], ptr @b64, i32 0, i64 7), align 8
108  %r0 = call i64 @llvm.ssub.sat.i64(i64 %a0, i64 %b0)
109  %r1 = call i64 @llvm.ssub.sat.i64(i64 %a1, i64 %b1)
110  %r2 = call i64 @llvm.ssub.sat.i64(i64 %a2, i64 %b2)
111  %r3 = call i64 @llvm.ssub.sat.i64(i64 %a3, i64 %b3)
112  %r4 = call i64 @llvm.ssub.sat.i64(i64 %a4, i64 %b4)
113  %r5 = call i64 @llvm.ssub.sat.i64(i64 %a5, i64 %b5)
114  %r6 = call i64 @llvm.ssub.sat.i64(i64 %a6, i64 %b6)
115  %r7 = call i64 @llvm.ssub.sat.i64(i64 %a7, i64 %b7)
116  store i64 %r0, ptr @c64, align 8
117  store i64 %r1, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 1), align 8
118  store i64 %r2, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 2), align 8
119  store i64 %r3, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 3), align 8
120  store i64 %r4, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 4), align 8
121  store i64 %r5, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 5), align 8
122  store i64 %r6, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 6), align 8
123  store i64 %r7, ptr getelementptr inbounds ([8 x i64], ptr @c64, i32 0, i64 7), align 8
124  ret void
125}
126
127define void @sub_v16i32() {
128; SSE-LABEL: @sub_v16i32(
129; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, ptr @a32, align 4
130; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, ptr @b32, align 4
131; SSE-NEXT:    [[TMP3:%.*]] = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP2]])
132; SSE-NEXT:    store <4 x i32> [[TMP3]], ptr @c32, align 4
133; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4), align 4
134; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4), align 4
135; SSE-NEXT:    [[TMP6:%.*]] = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP5]])
136; SSE-NEXT:    store <4 x i32> [[TMP6]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 4), align 4
137; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
138; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
139; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> [[TMP7]], <4 x i32> [[TMP8]])
140; SSE-NEXT:    store <4 x i32> [[TMP9]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8), align 4
141; SSE-NEXT:    [[TMP10:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
142; SSE-NEXT:    [[TMP11:%.*]] = load <4 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
143; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> [[TMP10]], <4 x i32> [[TMP11]])
144; SSE-NEXT:    store <4 x i32> [[TMP12]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 12), align 4
145; SSE-NEXT:    ret void
146;
147; AVX-LABEL: @sub_v16i32(
148; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, ptr @a32, align 4
149; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, ptr @b32, align 4
150; AVX-NEXT:    [[TMP3:%.*]] = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP2]])
151; AVX-NEXT:    store <8 x i32> [[TMP3]], ptr @c32, align 4
152; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8), align 4
153; AVX-NEXT:    [[TMP5:%.*]] = load <8 x i32>, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8), align 4
154; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> [[TMP4]], <8 x i32> [[TMP5]])
155; AVX-NEXT:    store <8 x i32> [[TMP6]], ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8), align 4
156; AVX-NEXT:    ret void
157;
158; AVX512-LABEL: @sub_v16i32(
159; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, ptr @a32, align 4
160; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, ptr @b32, align 4
161; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
162; AVX512-NEXT:    store <16 x i32> [[TMP3]], ptr @c32, align 4
163; AVX512-NEXT:    ret void
164;
165  %a0  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 0 ), align 4
166  %a1  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 1 ), align 4
167  %a2  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 2 ), align 4
168  %a3  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 3 ), align 4
169  %a4  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 4 ), align 4
170  %a5  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 5 ), align 4
171  %a6  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 6 ), align 4
172  %a7  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 7 ), align 4
173  %a8  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 8 ), align 4
174  %a9  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 9 ), align 4
175  %a10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 10), align 4
176  %a11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 11), align 4
177  %a12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 12), align 4
178  %a13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 13), align 4
179  %a14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 14), align 4
180  %a15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @a32, i32 0, i64 15), align 4
181  %b0  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 0 ), align 4
182  %b1  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 1 ), align 4
183  %b2  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 2 ), align 4
184  %b3  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 3 ), align 4
185  %b4  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 4 ), align 4
186  %b5  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 5 ), align 4
187  %b6  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 6 ), align 4
188  %b7  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 7 ), align 4
189  %b8  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 8 ), align 4
190  %b9  = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 9 ), align 4
191  %b10 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 10), align 4
192  %b11 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 11), align 4
193  %b12 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 12), align 4
194  %b13 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 13), align 4
195  %b14 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 14), align 4
196  %b15 = load i32, ptr getelementptr inbounds ([16 x i32], ptr @b32, i32 0, i64 15), align 4
197  %r0  = call i32 @llvm.ssub.sat.i32(i32 %a0 , i32 %b0 )
198  %r1  = call i32 @llvm.ssub.sat.i32(i32 %a1 , i32 %b1 )
199  %r2  = call i32 @llvm.ssub.sat.i32(i32 %a2 , i32 %b2 )
200  %r3  = call i32 @llvm.ssub.sat.i32(i32 %a3 , i32 %b3 )
201  %r4  = call i32 @llvm.ssub.sat.i32(i32 %a4 , i32 %b4 )
202  %r5  = call i32 @llvm.ssub.sat.i32(i32 %a5 , i32 %b5 )
203  %r6  = call i32 @llvm.ssub.sat.i32(i32 %a6 , i32 %b6 )
204  %r7  = call i32 @llvm.ssub.sat.i32(i32 %a7 , i32 %b7 )
205  %r8  = call i32 @llvm.ssub.sat.i32(i32 %a8 , i32 %b8 )
206  %r9  = call i32 @llvm.ssub.sat.i32(i32 %a9 , i32 %b9 )
207  %r10 = call i32 @llvm.ssub.sat.i32(i32 %a10, i32 %b10)
208  %r11 = call i32 @llvm.ssub.sat.i32(i32 %a11, i32 %b11)
209  %r12 = call i32 @llvm.ssub.sat.i32(i32 %a12, i32 %b12)
210  %r13 = call i32 @llvm.ssub.sat.i32(i32 %a13, i32 %b13)
211  %r14 = call i32 @llvm.ssub.sat.i32(i32 %a14, i32 %b14)
212  %r15 = call i32 @llvm.ssub.sat.i32(i32 %a15, i32 %b15)
213  store i32 %r0 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 0 ), align 4
214  store i32 %r1 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 1 ), align 4
215  store i32 %r2 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 2 ), align 4
216  store i32 %r3 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 3 ), align 4
217  store i32 %r4 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 4 ), align 4
218  store i32 %r5 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 5 ), align 4
219  store i32 %r6 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 6 ), align 4
220  store i32 %r7 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 7 ), align 4
221  store i32 %r8 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 8 ), align 4
222  store i32 %r9 , ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 9 ), align 4
223  store i32 %r10, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 10), align 4
224  store i32 %r11, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 11), align 4
225  store i32 %r12, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 12), align 4
226  store i32 %r13, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 13), align 4
227  store i32 %r14, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 14), align 4
228  store i32 %r15, ptr getelementptr inbounds ([16 x i32], ptr @c32, i32 0, i64 15), align 4
229  ret void
230}
231
232define void @sub_v32i16() {
233; SSE-LABEL: @sub_v32i16(
234; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, ptr @a16, align 2
235; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, ptr @b16, align 2
236; SSE-NEXT:    [[TMP3:%.*]] = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP2]])
237; SSE-NEXT:    store <8 x i16> [[TMP3]], ptr @c16, align 2
238; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 8), align 2
239; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 8), align 2
240; SSE-NEXT:    [[TMP6:%.*]] = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP5]])
241; SSE-NEXT:    store <8 x i16> [[TMP6]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 8), align 2
242; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
243; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
244; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> [[TMP7]], <8 x i16> [[TMP8]])
245; SSE-NEXT:    store <8 x i16> [[TMP9]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
246; SSE-NEXT:    [[TMP10:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 24), align 2
247; SSE-NEXT:    [[TMP11:%.*]] = load <8 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 24), align 2
248; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> [[TMP10]], <8 x i16> [[TMP11]])
249; SSE-NEXT:    store <8 x i16> [[TMP12]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 24), align 2
250; SSE-NEXT:    ret void
251;
252; AVX-LABEL: @sub_v32i16(
253; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, ptr @a16, align 2
254; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, ptr @b16, align 2
255; AVX-NEXT:    [[TMP3:%.*]] = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP2]])
256; AVX-NEXT:    store <16 x i16> [[TMP3]], ptr @c16, align 2
257; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
258; AVX-NEXT:    [[TMP5:%.*]] = load <16 x i16>, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
259; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> [[TMP4]], <16 x i16> [[TMP5]])
260; AVX-NEXT:    store <16 x i16> [[TMP6]], ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
261; AVX-NEXT:    ret void
262;
263; AVX512-LABEL: @sub_v32i16(
264; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, ptr @a16, align 2
265; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, ptr @b16, align 2
266; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.ssub.sat.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
267; AVX512-NEXT:    store <32 x i16> [[TMP3]], ptr @c16, align 2
268; AVX512-NEXT:    ret void
269;
270  %a0  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 0 ), align 2
271  %a1  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 1 ), align 2
272  %a2  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 2 ), align 2
273  %a3  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 3 ), align 2
274  %a4  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 4 ), align 2
275  %a5  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 5 ), align 2
276  %a6  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 6 ), align 2
277  %a7  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 7 ), align 2
278  %a8  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 8 ), align 2
279  %a9  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 9 ), align 2
280  %a10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 10), align 2
281  %a11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 11), align 2
282  %a12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 12), align 2
283  %a13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 13), align 2
284  %a14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 14), align 2
285  %a15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 15), align 2
286  %a16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 16), align 2
287  %a17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 17), align 2
288  %a18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 18), align 2
289  %a19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 19), align 2
290  %a20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 20), align 2
291  %a21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 21), align 2
292  %a22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 22), align 2
293  %a23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 23), align 2
294  %a24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 24), align 2
295  %a25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 25), align 2
296  %a26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 26), align 2
297  %a27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 27), align 2
298  %a28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 28), align 2
299  %a29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 29), align 2
300  %a30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 30), align 2
301  %a31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @a16, i32 0, i64 31), align 2
302  %b0  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 0 ), align 2
303  %b1  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 1 ), align 2
304  %b2  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 2 ), align 2
305  %b3  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 3 ), align 2
306  %b4  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 4 ), align 2
307  %b5  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 5 ), align 2
308  %b6  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 6 ), align 2
309  %b7  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 7 ), align 2
310  %b8  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 8 ), align 2
311  %b9  = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 9 ), align 2
312  %b10 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 10), align 2
313  %b11 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 11), align 2
314  %b12 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 12), align 2
315  %b13 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 13), align 2
316  %b14 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 14), align 2
317  %b15 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 15), align 2
318  %b16 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 16), align 2
319  %b17 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 17), align 2
320  %b18 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 18), align 2
321  %b19 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 19), align 2
322  %b20 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 20), align 2
323  %b21 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 21), align 2
324  %b22 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 22), align 2
325  %b23 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 23), align 2
326  %b24 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 24), align 2
327  %b25 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 25), align 2
328  %b26 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 26), align 2
329  %b27 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 27), align 2
330  %b28 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 28), align 2
331  %b29 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 29), align 2
332  %b30 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 30), align 2
333  %b31 = load i16, ptr getelementptr inbounds ([32 x i16], ptr @b16, i32 0, i64 31), align 2
334  %r0  = call i16 @llvm.ssub.sat.i16(i16 %a0 , i16 %b0 )
335  %r1  = call i16 @llvm.ssub.sat.i16(i16 %a1 , i16 %b1 )
336  %r2  = call i16 @llvm.ssub.sat.i16(i16 %a2 , i16 %b2 )
337  %r3  = call i16 @llvm.ssub.sat.i16(i16 %a3 , i16 %b3 )
338  %r4  = call i16 @llvm.ssub.sat.i16(i16 %a4 , i16 %b4 )
339  %r5  = call i16 @llvm.ssub.sat.i16(i16 %a5 , i16 %b5 )
340  %r6  = call i16 @llvm.ssub.sat.i16(i16 %a6 , i16 %b6 )
341  %r7  = call i16 @llvm.ssub.sat.i16(i16 %a7 , i16 %b7 )
342  %r8  = call i16 @llvm.ssub.sat.i16(i16 %a8 , i16 %b8 )
343  %r9  = call i16 @llvm.ssub.sat.i16(i16 %a9 , i16 %b9 )
344  %r10 = call i16 @llvm.ssub.sat.i16(i16 %a10, i16 %b10)
345  %r11 = call i16 @llvm.ssub.sat.i16(i16 %a11, i16 %b11)
346  %r12 = call i16 @llvm.ssub.sat.i16(i16 %a12, i16 %b12)
347  %r13 = call i16 @llvm.ssub.sat.i16(i16 %a13, i16 %b13)
348  %r14 = call i16 @llvm.ssub.sat.i16(i16 %a14, i16 %b14)
349  %r15 = call i16 @llvm.ssub.sat.i16(i16 %a15, i16 %b15)
350  %r16 = call i16 @llvm.ssub.sat.i16(i16 %a16, i16 %b16)
351  %r17 = call i16 @llvm.ssub.sat.i16(i16 %a17, i16 %b17)
352  %r18 = call i16 @llvm.ssub.sat.i16(i16 %a18, i16 %b18)
353  %r19 = call i16 @llvm.ssub.sat.i16(i16 %a19, i16 %b19)
354  %r20 = call i16 @llvm.ssub.sat.i16(i16 %a20, i16 %b20)
355  %r21 = call i16 @llvm.ssub.sat.i16(i16 %a21, i16 %b21)
356  %r22 = call i16 @llvm.ssub.sat.i16(i16 %a22, i16 %b22)
357  %r23 = call i16 @llvm.ssub.sat.i16(i16 %a23, i16 %b23)
358  %r24 = call i16 @llvm.ssub.sat.i16(i16 %a24, i16 %b24)
359  %r25 = call i16 @llvm.ssub.sat.i16(i16 %a25, i16 %b25)
360  %r26 = call i16 @llvm.ssub.sat.i16(i16 %a26, i16 %b26)
361  %r27 = call i16 @llvm.ssub.sat.i16(i16 %a27, i16 %b27)
362  %r28 = call i16 @llvm.ssub.sat.i16(i16 %a28, i16 %b28)
363  %r29 = call i16 @llvm.ssub.sat.i16(i16 %a29, i16 %b29)
364  %r30 = call i16 @llvm.ssub.sat.i16(i16 %a30, i16 %b30)
365  %r31 = call i16 @llvm.ssub.sat.i16(i16 %a31, i16 %b31)
366  store i16 %r0 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 0 ), align 2
367  store i16 %r1 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 1 ), align 2
368  store i16 %r2 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 2 ), align 2
369  store i16 %r3 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 3 ), align 2
370  store i16 %r4 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 4 ), align 2
371  store i16 %r5 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 5 ), align 2
372  store i16 %r6 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 6 ), align 2
373  store i16 %r7 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 7 ), align 2
374  store i16 %r8 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 8 ), align 2
375  store i16 %r9 , ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 9 ), align 2
376  store i16 %r10, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 10), align 2
377  store i16 %r11, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 11), align 2
378  store i16 %r12, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 12), align 2
379  store i16 %r13, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 13), align 2
380  store i16 %r14, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 14), align 2
381  store i16 %r15, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 15), align 2
382  store i16 %r16, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 16), align 2
383  store i16 %r17, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 17), align 2
384  store i16 %r18, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 18), align 2
385  store i16 %r19, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 19), align 2
386  store i16 %r20, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 20), align 2
387  store i16 %r21, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 21), align 2
388  store i16 %r22, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 22), align 2
389  store i16 %r23, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 23), align 2
390  store i16 %r24, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 24), align 2
391  store i16 %r25, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 25), align 2
392  store i16 %r26, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 26), align 2
393  store i16 %r27, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 27), align 2
394  store i16 %r28, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 28), align 2
395  store i16 %r29, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 29), align 2
396  store i16 %r30, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 30), align 2
397  store i16 %r31, ptr getelementptr inbounds ([32 x i16], ptr @c16, i32 0, i64 31), align 2
398  ret void
399}
400
401define void @sub_v64i8() {
402; SSE-LABEL: @sub_v64i8(
403; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, ptr @a8, align 1
404; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, ptr @b8, align 1
405; SSE-NEXT:    [[TMP3:%.*]] = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]])
406; SSE-NEXT:    store <16 x i8> [[TMP3]], ptr @c8, align 1
407; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
408; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
409; SSE-NEXT:    [[TMP6:%.*]] = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP5]])
410; SSE-NEXT:    store <16 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 16), align 1
411; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
412; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
413; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> [[TMP7]], <16 x i8> [[TMP8]])
414; SSE-NEXT:    store <16 x i8> [[TMP9]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
415; SSE-NEXT:    [[TMP10:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
416; SSE-NEXT:    [[TMP11:%.*]] = load <16 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
417; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> [[TMP10]], <16 x i8> [[TMP11]])
418; SSE-NEXT:    store <16 x i8> [[TMP12]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 48), align 1
419; SSE-NEXT:    ret void
420;
421; AVX-LABEL: @sub_v64i8(
422; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, ptr @a8, align 1
423; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, ptr @b8, align 1
424; AVX-NEXT:    [[TMP3:%.*]] = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP2]])
425; AVX-NEXT:    store <32 x i8> [[TMP3]], ptr @c8, align 1
426; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
427; AVX-NEXT:    [[TMP5:%.*]] = load <32 x i8>, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
428; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> [[TMP4]], <32 x i8> [[TMP5]])
429; AVX-NEXT:    store <32 x i8> [[TMP6]], ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
430; AVX-NEXT:    ret void
431;
432; AVX512-LABEL: @sub_v64i8(
433; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, ptr @a8, align 1
434; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, ptr @b8, align 1
435; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.ssub.sat.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
436; AVX512-NEXT:    store <64 x i8> [[TMP3]], ptr @c8, align 1
437; AVX512-NEXT:    ret void
438;
439  %a0  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 0 ), align 1
440  %a1  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 1 ), align 1
441  %a2  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 2 ), align 1
442  %a3  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 3 ), align 1
443  %a4  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 4 ), align 1
444  %a5  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 5 ), align 1
445  %a6  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 6 ), align 1
446  %a7  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 7 ), align 1
447  %a8  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 8 ), align 1
448  %a9  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 9 ), align 1
449  %a10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 10), align 1
450  %a11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 11), align 1
451  %a12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 12), align 1
452  %a13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 13), align 1
453  %a14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 14), align 1
454  %a15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 15), align 1
455  %a16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 16), align 1
456  %a17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 17), align 1
457  %a18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 18), align 1
458  %a19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 19), align 1
459  %a20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 20), align 1
460  %a21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 21), align 1
461  %a22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 22), align 1
462  %a23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 23), align 1
463  %a24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 24), align 1
464  %a25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 25), align 1
465  %a26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 26), align 1
466  %a27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 27), align 1
467  %a28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 28), align 1
468  %a29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 29), align 1
469  %a30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 30), align 1
470  %a31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 31), align 1
471  %a32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 32), align 1
472  %a33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 33), align 1
473  %a34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 34), align 1
474  %a35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 35), align 1
475  %a36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 36), align 1
476  %a37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 37), align 1
477  %a38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 38), align 1
478  %a39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 39), align 1
479  %a40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 40), align 1
480  %a41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 41), align 1
481  %a42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 42), align 1
482  %a43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 43), align 1
483  %a44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 44), align 1
484  %a45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 45), align 1
485  %a46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 46), align 1
486  %a47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 47), align 1
487  %a48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 48), align 1
488  %a49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 49), align 1
489  %a50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 50), align 1
490  %a51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 51), align 1
491  %a52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 52), align 1
492  %a53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 53), align 1
493  %a54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 54), align 1
494  %a55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 55), align 1
495  %a56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 56), align 1
496  %a57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 57), align 1
497  %a58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 58), align 1
498  %a59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 59), align 1
499  %a60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 60), align 1
500  %a61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 61), align 1
501  %a62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 62), align 1
502  %a63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @a8, i32 0, i64 63), align 1
503  %b0  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 0 ), align 1
504  %b1  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 1 ), align 1
505  %b2  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 2 ), align 1
506  %b3  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 3 ), align 1
507  %b4  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 4 ), align 1
508  %b5  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 5 ), align 1
509  %b6  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 6 ), align 1
510  %b7  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 7 ), align 1
511  %b8  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 8 ), align 1
512  %b9  = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 9 ), align 1
513  %b10 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 10), align 1
514  %b11 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 11), align 1
515  %b12 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 12), align 1
516  %b13 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 13), align 1
517  %b14 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 14), align 1
518  %b15 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 15), align 1
519  %b16 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 16), align 1
520  %b17 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 17), align 1
521  %b18 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 18), align 1
522  %b19 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 19), align 1
523  %b20 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 20), align 1
524  %b21 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 21), align 1
525  %b22 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 22), align 1
526  %b23 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 23), align 1
527  %b24 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 24), align 1
528  %b25 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 25), align 1
529  %b26 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 26), align 1
530  %b27 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 27), align 1
531  %b28 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 28), align 1
532  %b29 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 29), align 1
533  %b30 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 30), align 1
534  %b31 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 31), align 1
535  %b32 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 32), align 1
536  %b33 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 33), align 1
537  %b34 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 34), align 1
538  %b35 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 35), align 1
539  %b36 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 36), align 1
540  %b37 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 37), align 1
541  %b38 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 38), align 1
542  %b39 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 39), align 1
543  %b40 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 40), align 1
544  %b41 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 41), align 1
545  %b42 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 42), align 1
546  %b43 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 43), align 1
547  %b44 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 44), align 1
548  %b45 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 45), align 1
549  %b46 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 46), align 1
550  %b47 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 47), align 1
551  %b48 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 48), align 1
552  %b49 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 49), align 1
553  %b50 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 50), align 1
554  %b51 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 51), align 1
555  %b52 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 52), align 1
556  %b53 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 53), align 1
557  %b54 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 54), align 1
558  %b55 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 55), align 1
559  %b56 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 56), align 1
560  %b57 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 57), align 1
561  %b58 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 58), align 1
562  %b59 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 59), align 1
563  %b60 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 60), align 1
564  %b61 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 61), align 1
565  %b62 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 62), align 1
566  %b63 = load i8, ptr getelementptr inbounds ([64 x i8], ptr @b8, i32 0, i64 63), align 1
567  %r0  = call i8 @llvm.ssub.sat.i8(i8 %a0 , i8 %b0 )
568  %r1  = call i8 @llvm.ssub.sat.i8(i8 %a1 , i8 %b1 )
569  %r2  = call i8 @llvm.ssub.sat.i8(i8 %a2 , i8 %b2 )
570  %r3  = call i8 @llvm.ssub.sat.i8(i8 %a3 , i8 %b3 )
571  %r4  = call i8 @llvm.ssub.sat.i8(i8 %a4 , i8 %b4 )
572  %r5  = call i8 @llvm.ssub.sat.i8(i8 %a5 , i8 %b5 )
573  %r6  = call i8 @llvm.ssub.sat.i8(i8 %a6 , i8 %b6 )
574  %r7  = call i8 @llvm.ssub.sat.i8(i8 %a7 , i8 %b7 )
575  %r8  = call i8 @llvm.ssub.sat.i8(i8 %a8 , i8 %b8 )
576  %r9  = call i8 @llvm.ssub.sat.i8(i8 %a9 , i8 %b9 )
577  %r10 = call i8 @llvm.ssub.sat.i8(i8 %a10, i8 %b10)
578  %r11 = call i8 @llvm.ssub.sat.i8(i8 %a11, i8 %b11)
579  %r12 = call i8 @llvm.ssub.sat.i8(i8 %a12, i8 %b12)
580  %r13 = call i8 @llvm.ssub.sat.i8(i8 %a13, i8 %b13)
581  %r14 = call i8 @llvm.ssub.sat.i8(i8 %a14, i8 %b14)
582  %r15 = call i8 @llvm.ssub.sat.i8(i8 %a15, i8 %b15)
583  %r16 = call i8 @llvm.ssub.sat.i8(i8 %a16, i8 %b16)
584  %r17 = call i8 @llvm.ssub.sat.i8(i8 %a17, i8 %b17)
585  %r18 = call i8 @llvm.ssub.sat.i8(i8 %a18, i8 %b18)
586  %r19 = call i8 @llvm.ssub.sat.i8(i8 %a19, i8 %b19)
587  %r20 = call i8 @llvm.ssub.sat.i8(i8 %a20, i8 %b20)
588  %r21 = call i8 @llvm.ssub.sat.i8(i8 %a21, i8 %b21)
589  %r22 = call i8 @llvm.ssub.sat.i8(i8 %a22, i8 %b22)
590  %r23 = call i8 @llvm.ssub.sat.i8(i8 %a23, i8 %b23)
591  %r24 = call i8 @llvm.ssub.sat.i8(i8 %a24, i8 %b24)
592  %r25 = call i8 @llvm.ssub.sat.i8(i8 %a25, i8 %b25)
593  %r26 = call i8 @llvm.ssub.sat.i8(i8 %a26, i8 %b26)
594  %r27 = call i8 @llvm.ssub.sat.i8(i8 %a27, i8 %b27)
595  %r28 = call i8 @llvm.ssub.sat.i8(i8 %a28, i8 %b28)
596  %r29 = call i8 @llvm.ssub.sat.i8(i8 %a29, i8 %b29)
597  %r30 = call i8 @llvm.ssub.sat.i8(i8 %a30, i8 %b30)
598  %r31 = call i8 @llvm.ssub.sat.i8(i8 %a31, i8 %b31)
599  %r32 = call i8 @llvm.ssub.sat.i8(i8 %a32, i8 %b32)
600  %r33 = call i8 @llvm.ssub.sat.i8(i8 %a33, i8 %b33)
601  %r34 = call i8 @llvm.ssub.sat.i8(i8 %a34, i8 %b34)
602  %r35 = call i8 @llvm.ssub.sat.i8(i8 %a35, i8 %b35)
603  %r36 = call i8 @llvm.ssub.sat.i8(i8 %a36, i8 %b36)
604  %r37 = call i8 @llvm.ssub.sat.i8(i8 %a37, i8 %b37)
605  %r38 = call i8 @llvm.ssub.sat.i8(i8 %a38, i8 %b38)
606  %r39 = call i8 @llvm.ssub.sat.i8(i8 %a39, i8 %b39)
607  %r40 = call i8 @llvm.ssub.sat.i8(i8 %a40, i8 %b40)
608  %r41 = call i8 @llvm.ssub.sat.i8(i8 %a41, i8 %b41)
609  %r42 = call i8 @llvm.ssub.sat.i8(i8 %a42, i8 %b42)
610  %r43 = call i8 @llvm.ssub.sat.i8(i8 %a43, i8 %b43)
611  %r44 = call i8 @llvm.ssub.sat.i8(i8 %a44, i8 %b44)
612  %r45 = call i8 @llvm.ssub.sat.i8(i8 %a45, i8 %b45)
613  %r46 = call i8 @llvm.ssub.sat.i8(i8 %a46, i8 %b46)
614  %r47 = call i8 @llvm.ssub.sat.i8(i8 %a47, i8 %b47)
615  %r48 = call i8 @llvm.ssub.sat.i8(i8 %a48, i8 %b48)
616  %r49 = call i8 @llvm.ssub.sat.i8(i8 %a49, i8 %b49)
617  %r50 = call i8 @llvm.ssub.sat.i8(i8 %a50, i8 %b50)
618  %r51 = call i8 @llvm.ssub.sat.i8(i8 %a51, i8 %b51)
619  %r52 = call i8 @llvm.ssub.sat.i8(i8 %a52, i8 %b52)
620  %r53 = call i8 @llvm.ssub.sat.i8(i8 %a53, i8 %b53)
621  %r54 = call i8 @llvm.ssub.sat.i8(i8 %a54, i8 %b54)
622  %r55 = call i8 @llvm.ssub.sat.i8(i8 %a55, i8 %b55)
623  %r56 = call i8 @llvm.ssub.sat.i8(i8 %a56, i8 %b56)
624  %r57 = call i8 @llvm.ssub.sat.i8(i8 %a57, i8 %b57)
625  %r58 = call i8 @llvm.ssub.sat.i8(i8 %a58, i8 %b58)
626  %r59 = call i8 @llvm.ssub.sat.i8(i8 %a59, i8 %b59)
627  %r60 = call i8 @llvm.ssub.sat.i8(i8 %a60, i8 %b60)
628  %r61 = call i8 @llvm.ssub.sat.i8(i8 %a61, i8 %b61)
629  %r62 = call i8 @llvm.ssub.sat.i8(i8 %a62, i8 %b62)
630  %r63 = call i8 @llvm.ssub.sat.i8(i8 %a63, i8 %b63)
631  store i8 %r0 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 0 ), align 1
632  store i8 %r1 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 1 ), align 1
633  store i8 %r2 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 2 ), align 1
634  store i8 %r3 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 3 ), align 1
635  store i8 %r4 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 4 ), align 1
636  store i8 %r5 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 5 ), align 1
637  store i8 %r6 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 6 ), align 1
638  store i8 %r7 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 7 ), align 1
639  store i8 %r8 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 8 ), align 1
640  store i8 %r9 , ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 9 ), align 1
641  store i8 %r10, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 10), align 1
642  store i8 %r11, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 11), align 1
643  store i8 %r12, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 12), align 1
644  store i8 %r13, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 13), align 1
645  store i8 %r14, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 14), align 1
646  store i8 %r15, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 15), align 1
647  store i8 %r16, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 16), align 1
648  store i8 %r17, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 17), align 1
649  store i8 %r18, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 18), align 1
650  store i8 %r19, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 19), align 1
651  store i8 %r20, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 20), align 1
652  store i8 %r21, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 21), align 1
653  store i8 %r22, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 22), align 1
654  store i8 %r23, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 23), align 1
655  store i8 %r24, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 24), align 1
656  store i8 %r25, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 25), align 1
657  store i8 %r26, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 26), align 1
658  store i8 %r27, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 27), align 1
659  store i8 %r28, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 28), align 1
660  store i8 %r29, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 29), align 1
661  store i8 %r30, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 30), align 1
662  store i8 %r31, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 31), align 1
663  store i8 %r32, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 32), align 1
664  store i8 %r33, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 33), align 1
665  store i8 %r34, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 34), align 1
666  store i8 %r35, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 35), align 1
667  store i8 %r36, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 36), align 1
668  store i8 %r37, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 37), align 1
669  store i8 %r38, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 38), align 1
670  store i8 %r39, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 39), align 1
671  store i8 %r40, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 40), align 1
672  store i8 %r41, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 41), align 1
673  store i8 %r42, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 42), align 1
674  store i8 %r43, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 43), align 1
675  store i8 %r44, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 44), align 1
676  store i8 %r45, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 45), align 1
677  store i8 %r46, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 46), align 1
678  store i8 %r47, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 47), align 1
679  store i8 %r48, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 48), align 1
680  store i8 %r49, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 49), align 1
681  store i8 %r50, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 50), align 1
682  store i8 %r51, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 51), align 1
683  store i8 %r52, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 52), align 1
684  store i8 %r53, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 53), align 1
685  store i8 %r54, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 54), align 1
686  store i8 %r55, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 55), align 1
687  store i8 %r56, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 56), align 1
688  store i8 %r57, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 57), align 1
689  store i8 %r58, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 58), align 1
690  store i8 %r59, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 59), align 1
691  store i8 %r60, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 60), align 1
692  store i8 %r61, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 61), align 1
693  store i8 %r62, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 62), align 1
694  store i8 %r63, ptr getelementptr inbounds ([64 x i8], ptr @c8, i32 0, i64 63), align 1
695  ret void
696}
697;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
698; AVX1: {{.*}}
699; AVX2: {{.*}}
700; SLM: {{.*}}
701; SSE2: {{.*}}
702