1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
2; RUN: opt -passes='print<cost-model>' 2>&1 -disable-output -mtriple=aarch64 < %s | FileCheck %s --check-prefix=RECIP
3; RUN: opt -passes='print<cost-model>' 2>&1 -disable-output -cost-kind=code-size -mtriple=aarch64 < %s | FileCheck %s --check-prefix=SIZE
4
5declare {i64, i1}              @llvm.sadd.with.overflow.i64(i64, i64)
6declare {<2 x i64>, <2 x i1>}  @llvm.sadd.with.overflow.v2i64(<2 x i64>, <2 x i64>)
7declare {<4 x i64>, <4 x i1>}  @llvm.sadd.with.overflow.v4i64(<4 x i64>, <4 x i64>)
8declare {<8 x i64>, <8 x i1>}  @llvm.sadd.with.overflow.v8i64(<8 x i64>, <8 x i64>)
9
10declare {i32, i1}               @llvm.sadd.with.overflow.i32(i32, i32)
11declare {<4 x i32>, <4 x i1>}   @llvm.sadd.with.overflow.v4i32(<4 x i32>, <4 x i32>)
12declare {<8 x i32>, <8 x i1>}   @llvm.sadd.with.overflow.v8i32(<8 x i32>, <8 x i32>)
13declare {<16 x i32>, <16 x i1>} @llvm.sadd.with.overflow.v16i32(<16 x i32>, <16 x i32>)
14
15declare {i16, i1}               @llvm.sadd.with.overflow.i16(i16, i16)
16declare {<8 x i16>,  <8 x i1>}  @llvm.sadd.with.overflow.v8i16(<8 x i16>, <8 x i16>)
17declare {<16 x i16>, <16 x i1>} @llvm.sadd.with.overflow.v16i16(<16 x i16>, <16 x i16>)
18declare {<32 x i16>, <32 x i1>} @llvm.sadd.with.overflow.v32i16(<32 x i16>, <32 x i16>)
19
20declare {i8, i1}                @llvm.sadd.with.overflow.i8(i8, i8)
21declare {<16 x i8>, <16 x i1>}  @llvm.sadd.with.overflow.v16i8(<16 x i8>, <16 x i8>)
22declare {<32 x i8>, <32 x i1>}  @llvm.sadd.with.overflow.v32i8(<32 x i8>, <32 x i8>)
23declare {<64 x i8>, <64 x i1>}  @llvm.sadd.with.overflow.v64i8(<64 x i8>, <64 x i8>)
24
25define i32 @sadd(i32 %arg) {
26; RECIP-LABEL: 'sadd'
27; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
28; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.sadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
29; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
30; RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
31; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 undef, i32 undef)
32; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.sadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
33; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
34; RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
35; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.sadd.with.overflow.i16(i16 undef, i16 undef)
36; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.sadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
37; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
38; RECIP-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
39; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 undef, i8 undef)
40; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.sadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
41; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
42; RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
43; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
44;
45; SIZE-LABEL: 'sadd'
46; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
47; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.sadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
48; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
49; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
50; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 undef, i32 undef)
51; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.sadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
52; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
53; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
54; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.sadd.with.overflow.i16(i16 undef, i16 undef)
55; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.sadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
56; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
57; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
58; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 undef, i8 undef)
59; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.sadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
60; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
61; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
62; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
63;
64  %I64 = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
65  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.sadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
66  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
67  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
68
69  %I32 = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 undef, i32 undef)
70  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.sadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
71  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
72  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
73
74  %I16 = call {i16, i1} @llvm.sadd.with.overflow.i16(i16 undef, i16 undef)
75  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.sadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
76  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
77  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
78
79  %I8 = call {i8, i1} @llvm.sadd.with.overflow.i8(i8 undef, i8 undef)
80  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.sadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
81  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
82  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
83
84  ret i32 undef
85}
86
87declare {i64, i1}              @llvm.uadd.with.overflow.i64(i64, i64)
88declare {<2 x i64>, <2 x i1>}  @llvm.uadd.with.overflow.v2i64(<2 x i64>, <2 x i64>)
89declare {<4 x i64>, <4 x i1>}  @llvm.uadd.with.overflow.v4i64(<4 x i64>, <4 x i64>)
90declare {<8 x i64>, <8 x i1>}  @llvm.uadd.with.overflow.v8i64(<8 x i64>, <8 x i64>)
91
92declare {i32, i1}               @llvm.uadd.with.overflow.i32(i32, i32)
93declare {<4 x i32>, <4 x i1>}   @llvm.uadd.with.overflow.v4i32(<4 x i32>, <4 x i32>)
94declare {<8 x i32>, <8 x i1>}   @llvm.uadd.with.overflow.v8i32(<8 x i32>, <8 x i32>)
95declare {<16 x i32>, <16 x i1>} @llvm.uadd.with.overflow.v16i32(<16 x i32>, <16 x i32>)
96
97declare {i16, i1}               @llvm.uadd.with.overflow.i16(i16, i16)
98declare {<8 x i16>,  <8 x i1>}  @llvm.uadd.with.overflow.v8i16(<8 x i16>, <8 x i16>)
99declare {<16 x i16>, <16 x i1>} @llvm.uadd.with.overflow.v16i16(<16 x i16>, <16 x i16>)
100declare {<32 x i16>, <32 x i1>} @llvm.uadd.with.overflow.v32i16(<32 x i16>, <32 x i16>)
101
102declare {i8, i1}                @llvm.uadd.with.overflow.i8(i8, i8)
103declare {<16 x i8>, <16 x i1>}  @llvm.uadd.with.overflow.v16i8(<16 x i8>, <16 x i8>)
104declare {<32 x i8>, <32 x i1>}  @llvm.uadd.with.overflow.v32i8(<32 x i8>, <32 x i8>)
105declare {<64 x i8>, <64 x i1>}  @llvm.uadd.with.overflow.v64i8(<64 x i8>, <64 x i8>)
106
107define i32 @uadd(i32 %arg) {
108; RECIP-LABEL: 'uadd'
109; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
110; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
111; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
112; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
113; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
114; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
115; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
116; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
117; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
118; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
119; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
120; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
121; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
122; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
123; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
124; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
125; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
126;
127; SIZE-LABEL: 'uadd'
128; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
129; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
130; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
131; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
132; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
133; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
134; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
135; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
136; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
137; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
138; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
139; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
140; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
141; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
142; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
143; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
144; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
145;
146  %I64 = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
147  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
148  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
149  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
150
151  %I32 = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
152  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
153  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
154  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
155
156  %I16 = call {i16, i1} @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
157  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
158  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
159  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
160
161  %I8 = call {i8, i1} @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
162  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
163  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
164  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
165
166  ret i32 undef
167}
168
169declare {i64, i1}              @llvm.ssub.with.overflow.i64(i64, i64)
170declare {<2 x i64>, <2 x i1>}  @llvm.ssub.with.overflow.v2i64(<2 x i64>, <2 x i64>)
171declare {<4 x i64>, <4 x i1>}  @llvm.ssub.with.overflow.v4i64(<4 x i64>, <4 x i64>)
172declare {<8 x i64>, <8 x i1>}  @llvm.ssub.with.overflow.v8i64(<8 x i64>, <8 x i64>)
173
174declare {i32, i1}               @llvm.ssub.with.overflow.i32(i32, i32)
175declare {<4 x i32>, <4 x i1>}   @llvm.ssub.with.overflow.v4i32(<4 x i32>, <4 x i32>)
176declare {<8 x i32>, <8 x i1>}   @llvm.ssub.with.overflow.v8i32(<8 x i32>, <8 x i32>)
177declare {<16 x i32>, <16 x i1>} @llvm.ssub.with.overflow.v16i32(<16 x i32>, <16 x i32>)
178
179declare {i16, i1}               @llvm.ssub.with.overflow.i16(i16, i16)
180declare {<8 x i16>,  <8 x i1>}  @llvm.ssub.with.overflow.v8i16(<8 x i16>, <8 x i16>)
181declare {<16 x i16>, <16 x i1>} @llvm.ssub.with.overflow.v16i16(<16 x i16>, <16 x i16>)
182declare {<32 x i16>, <32 x i1>} @llvm.ssub.with.overflow.v32i16(<32 x i16>, <32 x i16>)
183
184declare {i8, i1}                @llvm.ssub.with.overflow.i8(i8, i8)
185declare {<16 x i8>, <16 x i1>}  @llvm.ssub.with.overflow.v16i8(<16 x i8>, <16 x i8>)
186declare {<32 x i8>, <32 x i1>}  @llvm.ssub.with.overflow.v32i8(<32 x i8>, <32 x i8>)
187declare {<64 x i8>, <64 x i1>}  @llvm.ssub.with.overflow.v64i8(<64 x i8>, <64 x i8>)
188
189define i32 @ssub(i32 %arg) {
190; RECIP-LABEL: 'ssub'
191; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
192; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.ssub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
193; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
194; RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
195; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.ssub.with.overflow.i32(i32 undef, i32 undef)
196; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.ssub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
197; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
198; RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
199; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.ssub.with.overflow.i16(i16 undef, i16 undef)
200; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.ssub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
201; RECIP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
202; RECIP-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
203; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.ssub.with.overflow.i8(i8 undef, i8 undef)
204; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.ssub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
205; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
206; RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
207; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
208;
209; SIZE-LABEL: 'ssub'
210; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
211; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.ssub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
212; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
213; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
214; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.ssub.with.overflow.i32(i32 undef, i32 undef)
215; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.ssub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
216; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
217; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
218; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.ssub.with.overflow.i16(i16 undef, i16 undef)
219; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.ssub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
220; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
221; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
222; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.ssub.with.overflow.i8(i8 undef, i8 undef)
223; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.ssub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
224; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
225; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
226; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
227;
228  %I64 = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
229  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.ssub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
230  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
231  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
232
233  %I32 = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 undef, i32 undef)
234  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.ssub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
235  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
236  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
237
238  %I16 = call {i16, i1} @llvm.ssub.with.overflow.i16(i16 undef, i16 undef)
239  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.ssub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
240  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
241  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
242
243  %I8 = call {i8, i1} @llvm.ssub.with.overflow.i8(i8 undef, i8 undef)
244  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.ssub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
245  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
246  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
247
248  ret i32 undef
249}
250
251declare {i64, i1}              @llvm.usub.with.overflow.i64(i64, i64)
252declare {<2 x i64>, <2 x i1>}  @llvm.usub.with.overflow.v2i64(<2 x i64>, <2 x i64>)
253declare {<4 x i64>, <4 x i1>}  @llvm.usub.with.overflow.v4i64(<4 x i64>, <4 x i64>)
254declare {<8 x i64>, <8 x i1>}  @llvm.usub.with.overflow.v8i64(<8 x i64>, <8 x i64>)
255
256declare {i32, i1}               @llvm.usub.with.overflow.i32(i32, i32)
257declare {<4 x i32>, <4 x i1>}   @llvm.usub.with.overflow.v4i32(<4 x i32>, <4 x i32>)
258declare {<8 x i32>, <8 x i1>}   @llvm.usub.with.overflow.v8i32(<8 x i32>, <8 x i32>)
259declare {<16 x i32>, <16 x i1>} @llvm.usub.with.overflow.v16i32(<16 x i32>, <16 x i32>)
260
261declare {i16, i1}               @llvm.usub.with.overflow.i16(i16, i16)
262declare {<8 x i16>,  <8 x i1>}  @llvm.usub.with.overflow.v8i16(<8 x i16>, <8 x i16>)
263declare {<16 x i16>, <16 x i1>} @llvm.usub.with.overflow.v16i16(<16 x i16>, <16 x i16>)
264declare {<32 x i16>, <32 x i1>} @llvm.usub.with.overflow.v32i16(<32 x i16>, <32 x i16>)
265
266declare {i8, i1}                @llvm.usub.with.overflow.i8(i8, i8)
267declare {<16 x i8>, <16 x i1>}  @llvm.usub.with.overflow.v16i8(<16 x i8>, <16 x i8>)
268declare {<32 x i8>, <32 x i1>}  @llvm.usub.with.overflow.v32i8(<32 x i8>, <32 x i8>)
269declare {<64 x i8>, <64 x i1>}  @llvm.usub.with.overflow.v64i8(<64 x i8>, <64 x i8>)
270
271define i32 @usub(i32 %arg) {
272; RECIP-LABEL: 'usub'
273; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
274; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
275; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
276; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
277; RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
278; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
279; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
280; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
281; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
282; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
283; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
284; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
285; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
286; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
287; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
288; RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
289; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
290;
291; SIZE-LABEL: 'usub'
292; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
293; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
294; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
295; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
296; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
297; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
298; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
299; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
300; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
301; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
302; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
303; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
304; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
305; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
306; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
307; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
308; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
309;
310  %I64 = call {i64, i1} @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
311  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
312  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
313  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
314
315  %I32 = call {i32, i1} @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
316  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
317  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
318  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
319
320  %I16 = call {i16, i1} @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
321  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
322  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
323  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
324
325  %I8 = call {i8, i1} @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
326  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
327  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
328  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
329
330  ret i32 undef
331}
332
333declare {i64, i1}              @llvm.smul.with.overflow.i64(i64, i64)
334declare {<2 x i64>, <2 x i1>}  @llvm.smul.with.overflow.v2i64(<2 x i64>, <2 x i64>)
335declare {<4 x i64>, <4 x i1>}  @llvm.smul.with.overflow.v4i64(<4 x i64>, <4 x i64>)
336declare {<8 x i64>, <8 x i1>}  @llvm.smul.with.overflow.v8i64(<8 x i64>, <8 x i64>)
337
338declare {i32, i1}               @llvm.smul.with.overflow.i32(i32, i32)
339declare {<4 x i32>, <4 x i1>}   @llvm.smul.with.overflow.v4i32(<4 x i32>, <4 x i32>)
340declare {<8 x i32>, <8 x i1>}   @llvm.smul.with.overflow.v8i32(<8 x i32>, <8 x i32>)
341declare {<16 x i32>, <16 x i1>} @llvm.smul.with.overflow.v16i32(<16 x i32>, <16 x i32>)
342
343declare {i16, i1}               @llvm.smul.with.overflow.i16(i16, i16)
344declare {<8 x i16>,  <8 x i1>}  @llvm.smul.with.overflow.v8i16(<8 x i16>, <8 x i16>)
345declare {<16 x i16>, <16 x i1>} @llvm.smul.with.overflow.v16i16(<16 x i16>, <16 x i16>)
346declare {<32 x i16>, <32 x i1>} @llvm.smul.with.overflow.v32i16(<32 x i16>, <32 x i16>)
347
348declare {i8, i1}                @llvm.smul.with.overflow.i8(i8, i8)
349declare {<16 x i8>, <16 x i1>}  @llvm.smul.with.overflow.v16i8(<16 x i8>, <16 x i8>)
350declare {<32 x i8>, <32 x i1>}  @llvm.smul.with.overflow.v32i8(<32 x i8>, <32 x i8>)
351declare {<64 x i8>, <64 x i1>}  @llvm.smul.with.overflow.v64i8(<64 x i8>, <64 x i8>)
352
353define i32 @smul(i32 %arg) {
354; RECIP-LABEL: 'smul'
355; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
356; RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
357; RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
358; RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
359; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
360; RECIP-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
361; RECIP-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
362; RECIP-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
363; RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
364; RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
365; RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
366; RECIP-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
367; RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
368; RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
369; RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
370; RECIP-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
371; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
372;
373; SIZE-LABEL: 'smul'
374; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
375; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
376; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
377; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
378; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
379; SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
380; SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
381; SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
382; SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
383; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
384; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
385; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
386; SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
387; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
388; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
389; SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
390; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
391;
392  %I64 = call {i64, i1} @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
393  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
394  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
395  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
396
397  %I32 = call {i32, i1} @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
398  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
399  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
400  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
401
402  %I16 = call {i16, i1} @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
403  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
404  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
405  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
406
407  %I8 = call {i8, i1} @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
408  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
409  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
410  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
411
412  ret i32 undef
413}
414
415declare {i64, i1}              @llvm.umul.with.overflow.i64(i64, i64)
416declare {<2 x i64>, <2 x i1>}  @llvm.umul.with.overflow.v2i64(<2 x i64>, <2 x i64>)
417declare {<4 x i64>, <4 x i1>}  @llvm.umul.with.overflow.v4i64(<4 x i64>, <4 x i64>)
418declare {<8 x i64>, <8 x i1>}  @llvm.umul.with.overflow.v8i64(<8 x i64>, <8 x i64>)
419
420declare {i32, i1}               @llvm.umul.with.overflow.i32(i32, i32)
421declare {<4 x i32>, <4 x i1>}   @llvm.umul.with.overflow.v4i32(<4 x i32>, <4 x i32>)
422declare {<8 x i32>, <8 x i1>}   @llvm.umul.with.overflow.v8i32(<8 x i32>, <8 x i32>)
423declare {<16 x i32>, <16 x i1>} @llvm.umul.with.overflow.v16i32(<16 x i32>, <16 x i32>)
424
425declare {i16, i1}               @llvm.umul.with.overflow.i16(i16, i16)
426declare {<8 x i16>,  <8 x i1>}  @llvm.umul.with.overflow.v8i16(<8 x i16>, <8 x i16>)
427declare {<16 x i16>, <16 x i1>} @llvm.umul.with.overflow.v16i16(<16 x i16>, <16 x i16>)
428declare {<32 x i16>, <32 x i1>} @llvm.umul.with.overflow.v32i16(<32 x i16>, <32 x i16>)
429
430declare {i8, i1}                @llvm.umul.with.overflow.i8(i8, i8)
431declare {<16 x i8>, <16 x i1>}  @llvm.umul.with.overflow.v16i8(<16 x i8>, <16 x i8>)
432declare {<32 x i8>, <32 x i1>}  @llvm.umul.with.overflow.v32i8(<32 x i8>, <32 x i8>)
433declare {<64 x i8>, <64 x i1>}  @llvm.umul.with.overflow.v64i8(<64 x i8>, <64 x i8>)
434
435define i32 @umul(i32 %arg) {
436; RECIP-LABEL: 'umul'
437; RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
438; RECIP-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
439; RECIP-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
440; RECIP-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
441; RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
442; RECIP-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
443; RECIP-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
444; RECIP-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
445; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
446; RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
447; RECIP-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
448; RECIP-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
449; RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
450; RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
451; RECIP-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
452; RECIP-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
453; RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
454;
455; SIZE-LABEL: 'umul'
456; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
457; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
458; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
459; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
460; SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
461; SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
462; SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
463; SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
464; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
465; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
466; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
467; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
468; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
469; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
470; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
471; SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
472; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
473;
474  %I64 = call {i64, i1} @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
475  %V2I64 = call {<2 x i64>, <2 x i1>} @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
476  %V4I64 = call {<4 x i64>, <4 x i1>} @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
477  %V8I64 = call {<8 x i64>, <8 x i1>} @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
478
479  %I32 = call {i32, i1} @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
480  %V4I32  = call {<4 x i32>, <4 x i1>}  @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
481  %V8I32  = call {<8 x i32>, <8 x i1>}  @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
482  %V16I32 = call {<16 x i32>, <16 x i1>} @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
483
484  %I16 = call {i16, i1} @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
485  %V8I16  = call {<8 x i16>, <8 x i1>}  @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
486  %V16I16 = call {<16 x i16>, <16 x i1>} @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
487  %V32I16 = call {<32 x i16>, <32 x i1>} @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
488
489  %I8 = call {i8, i1} @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
490  %V16I8 = call {<16 x i8>, <16 x i1>} @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
491  %V32I8 = call {<32 x i8>, <32 x i1>} @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
492  %V64I8 = call {<64 x i8>, <64 x i1>} @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
493
494  ret i32 undef
495}
496