xref: /llvm-project/llvm/test/Analysis/CostModel/ARM/mve-minmax.ll (revision 4178e33470763b406f614b646c8b01d24309e20b) 
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=thumbv8.1m.main-none-eabi -mattr=+mve < %s | FileCheck %s --check-prefixes=MVE-RECIP,MVEI-RECIP
3; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve < %s | FileCheck %s --check-prefixes=MVE-SIZE,MVEI-SIZE
4; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=thumbv8.1m.-none-eabimain -mattr=+mve.fp < %s | FileCheck %s --check-prefixes=MVE-RECIP,MVEF-RECIP
5; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp < %s | FileCheck %s --check-prefixes=MVE-SIZE,MVEF-SIZE
6
7target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
8
9declare i64        @llvm.smin.i64(i64, i64)
10declare <2 x i64>  @llvm.smin.v2i64(<2 x i64>, <2 x i64>)
11declare <4 x i64>  @llvm.smin.v4i64(<4 x i64>, <4 x i64>)
12declare <8 x i64>  @llvm.smin.v8i64(<8 x i64>, <8 x i64>)
13
14declare i32        @llvm.smin.i32(i32, i32)
15declare <2 x i32>  @llvm.smin.v2i32(<2 x i32>, <2 x i32>)
16declare <4 x i32>  @llvm.smin.v4i32(<4 x i32>, <4 x i32>)
17declare <8 x i32>  @llvm.smin.v8i32(<8 x i32>, <8 x i32>)
18declare <16 x i32> @llvm.smin.v16i32(<16 x i32>, <16 x i32>)
19
20declare i16        @llvm.smin.i16(i16, i16)
21declare <2 x i16>  @llvm.smin.v2i16(<2 x i16>, <2 x i16>)
22declare <4 x i16>  @llvm.smin.v4i16(<4 x i16>, <4 x i16>)
23declare <8 x i16>  @llvm.smin.v8i16(<8 x i16>, <8 x i16>)
24declare <16 x i16> @llvm.smin.v16i16(<16 x i16>, <16 x i16>)
25declare <32 x i16> @llvm.smin.v32i16(<32 x i16>, <32 x i16>)
26
27declare i8         @llvm.smin.i8(i8,  i8)
28declare <2 x i8>   @llvm.smin.v2i8(<2 x i8>, <2 x i8>)
29declare <4 x i8>   @llvm.smin.v4i8(<4 x i8>, <4 x i8>)
30declare <8 x i8>   @llvm.smin.v8i8(<8 x i8>, <8 x i8>)
31declare <16 x i8>  @llvm.smin.v16i8(<16 x i8>, <16 x i8>)
32declare <32 x i8>  @llvm.smin.v32i8(<32 x i8>, <32 x i8>)
33declare <64 x i8>  @llvm.smin.v64i8(<64 x i8>, <64 x i8>)
34
35define i32 @smin(i32 %arg) {
36; MVE-RECIP-LABEL: 'smin'
37; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
38; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
39; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
40; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
41; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
42; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <2 x i32> @llvm.smin.v2i32(<2 x i32> undef, <2 x i32> undef)
43; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
44; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
45; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
46; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smin.i16(i16 undef, i16 undef)
47; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I16 = call <2 x i16> @llvm.smin.v2i16(<2 x i16> undef, <2 x i16> undef)
48; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.smin.v4i16(<4 x i16> undef, <4 x i16> undef)
49; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.smin.v8i16(<8 x i16> undef, <8 x i16> undef)
50; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.smin.v16i16(<16 x i16> undef, <16 x i16> undef)
51; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.smin.v32i16(<32 x i16> undef, <32 x i16> undef)
52; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smin.i8(i8 undef, i8 undef)
53; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I8 = call <2 x i8> @llvm.smin.v2i8(<2 x i8> undef, <2 x i8> undef)
54; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I8 = call <4 x i8> @llvm.smin.v4i8(<4 x i8> undef, <4 x i8> undef)
55; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.smin.v8i8(<8 x i8> undef, <8 x i8> undef)
56; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
57; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
58; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
59; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
60;
61; MVE-SIZE-LABEL: 'smin'
62; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
63; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
64; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
65; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
66; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
67; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.smin.v2i32(<2 x i32> undef, <2 x i32> undef)
68; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
69; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
70; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
71; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.smin.i16(i16 undef, i16 undef)
72; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I16 = call <2 x i16> @llvm.smin.v2i16(<2 x i16> undef, <2 x i16> undef)
73; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = call <4 x i16> @llvm.smin.v4i16(<4 x i16> undef, <4 x i16> undef)
74; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smin.v8i16(<8 x i16> undef, <8 x i16> undef)
75; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smin.v16i16(<16 x i16> undef, <16 x i16> undef)
76; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smin.v32i16(<32 x i16> undef, <32 x i16> undef)
77; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.smin.i8(i8 undef, i8 undef)
78; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I8 = call <2 x i8> @llvm.smin.v2i8(<2 x i8> undef, <2 x i8> undef)
79; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = call <4 x i8> @llvm.smin.v4i8(<4 x i8> undef, <4 x i8> undef)
80; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = call <8 x i8> @llvm.smin.v8i8(<8 x i8> undef, <8 x i8> undef)
81; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
82; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
83; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
84; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
85;
86  %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
87  %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
88  %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
89  %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
90
91  %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
92  %V2I32  = call <2 x i32>  @llvm.smin.v2i32(<2 x i32> undef, <2 x i32> undef)
93  %V4I32  = call <4 x i32>  @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
94  %V8I32  = call <8 x i32>  @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
95  %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
96
97  %I16 = call i16 @llvm.smin.i16(i16 undef, i16 undef)
98  %V2I16  = call <2 x i16>  @llvm.smin.v2i16(<2 x i16> undef, <2 x i16> undef)
99  %V4I16  = call <4 x i16>  @llvm.smin.v4i16(<4 x i16> undef, <4 x i16> undef)
100  %V8I16  = call <8 x i16>  @llvm.smin.v8i16(<8 x i16> undef, <8 x i16> undef)
101  %V16I16 = call <16 x i16> @llvm.smin.v16i16(<16 x i16> undef, <16 x i16> undef)
102  %V32I16 = call <32 x i16> @llvm.smin.v32i16(<32 x i16> undef, <32 x i16> undef)
103
104  %I8 = call i8 @llvm.smin.i8(i8 undef, i8 undef)
105  %V2I8  = call <2 x i8>  @llvm.smin.v2i8(<2 x i8> undef, <2 x i8> undef)
106  %V4I8  = call <4 x i8>  @llvm.smin.v4i8(<4 x i8> undef, <4 x i8> undef)
107  %V8I8  = call <8 x i8>  @llvm.smin.v8i8(<8 x i8> undef, <8 x i8> undef)
108  %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
109  %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
110  %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
111
112  ret i32 undef
113}
114
115
116declare i64        @llvm.smax.i64(i64, i64)
117declare <2 x i64>  @llvm.smax.v2i64(<2 x i64>, <2 x i64>)
118declare <4 x i64>  @llvm.smax.v4i64(<4 x i64>, <4 x i64>)
119declare <8 x i64>  @llvm.smax.v8i64(<8 x i64>, <8 x i64>)
120
121declare i32        @llvm.smax.i32(i32, i32)
122declare <2 x i32>  @llvm.smax.v2i32(<2 x i32>, <2 x i32>)
123declare <4 x i32>  @llvm.smax.v4i32(<4 x i32>, <4 x i32>)
124declare <8 x i32>  @llvm.smax.v8i32(<8 x i32>, <8 x i32>)
125declare <16 x i32> @llvm.smax.v16i32(<16 x i32>, <16 x i32>)
126
127declare i16        @llvm.smax.i16(i16, i16)
128declare <2 x i16>  @llvm.smax.v2i16(<2 x i16>, <2 x i16>)
129declare <4 x i16>  @llvm.smax.v4i16(<4 x i16>, <4 x i16>)
130declare <8 x i16>  @llvm.smax.v8i16(<8 x i16>, <8 x i16>)
131declare <16 x i16> @llvm.smax.v16i16(<16 x i16>, <16 x i16>)
132declare <32 x i16> @llvm.smax.v32i16(<32 x i16>, <32 x i16>)
133
134declare i8         @llvm.smax.i8(i8,  i8)
135declare <2 x i8>   @llvm.smax.v2i8(<2 x i8>, <2 x i8>)
136declare <4 x i8>   @llvm.smax.v4i8(<4 x i8>, <4 x i8>)
137declare <8 x i8>   @llvm.smax.v8i8(<8 x i8>, <8 x i8>)
138declare <16 x i8>  @llvm.smax.v16i8(<16 x i8>, <16 x i8>)
139declare <32 x i8>  @llvm.smax.v32i8(<32 x i8>, <32 x i8>)
140declare <64 x i8>  @llvm.smax.v64i8(<64 x i8>, <64 x i8>)
141
142define i32 @smax(i32 %arg) {
143; MVE-RECIP-LABEL: 'smax'
144; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
145; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
146; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
147; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
148; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
149; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <2 x i32> @llvm.smax.v2i32(<2 x i32> undef, <2 x i32> undef)
150; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
151; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
152; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
153; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smax.i16(i16 undef, i16 undef)
154; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I16 = call <2 x i16> @llvm.smax.v2i16(<2 x i16> undef, <2 x i16> undef)
155; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.smax.v4i16(<4 x i16> undef, <4 x i16> undef)
156; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.smax.v8i16(<8 x i16> undef, <8 x i16> undef)
157; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.smax.v16i16(<16 x i16> undef, <16 x i16> undef)
158; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.smax.v32i16(<32 x i16> undef, <32 x i16> undef)
159; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smax.i8(i8 undef, i8 undef)
160; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I8 = call <2 x i8> @llvm.smax.v2i8(<2 x i8> undef, <2 x i8> undef)
161; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I8 = call <4 x i8> @llvm.smax.v4i8(<4 x i8> undef, <4 x i8> undef)
162; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.smax.v8i8(<8 x i8> undef, <8 x i8> undef)
163; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
164; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
165; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
166; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
167;
168; MVE-SIZE-LABEL: 'smax'
169; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
170; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
171; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
172; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
173; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
174; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.smax.v2i32(<2 x i32> undef, <2 x i32> undef)
175; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
176; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
177; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
178; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.smax.i16(i16 undef, i16 undef)
179; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I16 = call <2 x i16> @llvm.smax.v2i16(<2 x i16> undef, <2 x i16> undef)
180; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = call <4 x i16> @llvm.smax.v4i16(<4 x i16> undef, <4 x i16> undef)
181; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smax.v8i16(<8 x i16> undef, <8 x i16> undef)
182; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smax.v16i16(<16 x i16> undef, <16 x i16> undef)
183; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smax.v32i16(<32 x i16> undef, <32 x i16> undef)
184; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.smax.i8(i8 undef, i8 undef)
185; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I8 = call <2 x i8> @llvm.smax.v2i8(<2 x i8> undef, <2 x i8> undef)
186; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = call <4 x i8> @llvm.smax.v4i8(<4 x i8> undef, <4 x i8> undef)
187; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = call <8 x i8> @llvm.smax.v8i8(<8 x i8> undef, <8 x i8> undef)
188; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
189; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
190; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
191; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
192;
193  %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
194  %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
195  %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
196  %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
197
198  %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
199  %V2I32  = call <2 x i32>  @llvm.smax.v2i32(<2 x i32> undef, <2 x i32> undef)
200  %V4I32  = call <4 x i32>  @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
201  %V8I32  = call <8 x i32>  @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
202  %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
203
204  %I16 = call i16 @llvm.smax.i16(i16 undef, i16 undef)
205  %V2I16  = call <2 x i16>  @llvm.smax.v2i16(<2 x i16> undef, <2 x i16> undef)
206  %V4I16  = call <4 x i16>  @llvm.smax.v4i16(<4 x i16> undef, <4 x i16> undef)
207  %V8I16  = call <8 x i16>  @llvm.smax.v8i16(<8 x i16> undef, <8 x i16> undef)
208  %V16I16 = call <16 x i16> @llvm.smax.v16i16(<16 x i16> undef, <16 x i16> undef)
209  %V32I16 = call <32 x i16> @llvm.smax.v32i16(<32 x i16> undef, <32 x i16> undef)
210
211  %I8 = call i8 @llvm.smax.i8(i8 undef, i8 undef)
212  %V2I8  = call <2 x i8>  @llvm.smax.v2i8(<2 x i8> undef, <2 x i8> undef)
213  %V4I8  = call <4 x i8>  @llvm.smax.v4i8(<4 x i8> undef, <4 x i8> undef)
214  %V8I8  = call <8 x i8>  @llvm.smax.v8i8(<8 x i8> undef, <8 x i8> undef)
215  %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
216  %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
217  %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
218
219  ret i32 undef
220}
221
222
223
224declare i64        @llvm.umin.i64(i64, i64)
225declare <2 x i64>  @llvm.umin.v2i64(<2 x i64>, <2 x i64>)
226declare <4 x i64>  @llvm.umin.v4i64(<4 x i64>, <4 x i64>)
227declare <8 x i64>  @llvm.umin.v8i64(<8 x i64>, <8 x i64>)
228
229declare i32        @llvm.umin.i32(i32, i32)
230declare <2 x i32>  @llvm.umin.v2i32(<2 x i32>, <2 x i32>)
231declare <4 x i32>  @llvm.umin.v4i32(<4 x i32>, <4 x i32>)
232declare <8 x i32>  @llvm.umin.v8i32(<8 x i32>, <8 x i32>)
233declare <16 x i32> @llvm.umin.v16i32(<16 x i32>, <16 x i32>)
234
235declare i16        @llvm.umin.i16(i16, i16)
236declare <2 x i16>  @llvm.umin.v2i16(<2 x i16>, <2 x i16>)
237declare <4 x i16>  @llvm.umin.v4i16(<4 x i16>, <4 x i16>)
238declare <8 x i16>  @llvm.umin.v8i16(<8 x i16>, <8 x i16>)
239declare <16 x i16> @llvm.umin.v16i16(<16 x i16>, <16 x i16>)
240declare <32 x i16> @llvm.umin.v32i16(<32 x i16>, <32 x i16>)
241
242declare i8         @llvm.umin.i8(i8,  i8)
243declare <2 x i8>   @llvm.umin.v2i8(<2 x i8>, <2 x i8>)
244declare <4 x i8>   @llvm.umin.v4i8(<4 x i8>, <4 x i8>)
245declare <8 x i8>   @llvm.umin.v8i8(<8 x i8>, <8 x i8>)
246declare <16 x i8>  @llvm.umin.v16i8(<16 x i8>, <16 x i8>)
247declare <32 x i8>  @llvm.umin.v32i8(<32 x i8>, <32 x i8>)
248declare <64 x i8>  @llvm.umin.v64i8(<64 x i8>, <64 x i8>)
249
250define i32 @umin(i32 %arg) {
251; MVE-RECIP-LABEL: 'umin'
252; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
253; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
254; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
255; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
256; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
257; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <2 x i32> @llvm.umin.v2i32(<2 x i32> undef, <2 x i32> undef)
258; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
259; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
260; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
261; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umin.i16(i16 undef, i16 undef)
262; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I16 = call <2 x i16> @llvm.umin.v2i16(<2 x i16> undef, <2 x i16> undef)
263; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.umin.v4i16(<4 x i16> undef, <4 x i16> undef)
264; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umin.v8i16(<8 x i16> undef, <8 x i16> undef)
265; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umin.v16i16(<16 x i16> undef, <16 x i16> undef)
266; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.umin.v32i16(<32 x i16> undef, <32 x i16> undef)
267; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.umin.i8(i8 undef, i8 undef)
268; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I8 = call <2 x i8> @llvm.umin.v2i8(<2 x i8> undef, <2 x i8> undef)
269; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I8 = call <4 x i8> @llvm.umin.v4i8(<4 x i8> undef, <4 x i8> undef)
270; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.umin.v8i8(<8 x i8> undef, <8 x i8> undef)
271; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call <16 x i8> @llvm.umin.v16i8(<16 x i8> undef, <16 x i8> undef)
272; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.umin.v32i8(<32 x i8> undef, <32 x i8> undef)
273; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.umin.v64i8(<64 x i8> undef, <64 x i8> undef)
274; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
275;
276; MVE-SIZE-LABEL: 'umin'
277; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
278; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
279; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
280; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
281; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
282; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.umin.v2i32(<2 x i32> undef, <2 x i32> undef)
283; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
284; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
285; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
286; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.umin.i16(i16 undef, i16 undef)
287; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I16 = call <2 x i16> @llvm.umin.v2i16(<2 x i16> undef, <2 x i16> undef)
288; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = call <4 x i16> @llvm.umin.v4i16(<4 x i16> undef, <4 x i16> undef)
289; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.umin.v8i16(<8 x i16> undef, <8 x i16> undef)
290; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.umin.v16i16(<16 x i16> undef, <16 x i16> undef)
291; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.umin.v32i16(<32 x i16> undef, <32 x i16> undef)
292; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.umin.i8(i8 undef, i8 undef)
293; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I8 = call <2 x i8> @llvm.umin.v2i8(<2 x i8> undef, <2 x i8> undef)
294; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = call <4 x i8> @llvm.umin.v4i8(<4 x i8> undef, <4 x i8> undef)
295; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = call <8 x i8> @llvm.umin.v8i8(<8 x i8> undef, <8 x i8> undef)
296; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.umin.v16i8(<16 x i8> undef, <16 x i8> undef)
297; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.umin.v32i8(<32 x i8> undef, <32 x i8> undef)
298; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.umin.v64i8(<64 x i8> undef, <64 x i8> undef)
299; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
300;
301  %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
302  %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
303  %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
304  %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
305
306  %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
307  %V2I32  = call <2 x i32>  @llvm.umin.v2i32(<2 x i32> undef, <2 x i32> undef)
308  %V4I32  = call <4 x i32>  @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
309  %V8I32  = call <8 x i32>  @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
310  %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
311
312  %I16 = call i16 @llvm.umin.i16(i16 undef, i16 undef)
313  %V2I16  = call <2 x i16>  @llvm.umin.v2i16(<2 x i16> undef, <2 x i16> undef)
314  %V4I16  = call <4 x i16>  @llvm.umin.v4i16(<4 x i16> undef, <4 x i16> undef)
315  %V8I16  = call <8 x i16>  @llvm.umin.v8i16(<8 x i16> undef, <8 x i16> undef)
316  %V16I16 = call <16 x i16> @llvm.umin.v16i16(<16 x i16> undef, <16 x i16> undef)
317  %V32I16 = call <32 x i16> @llvm.umin.v32i16(<32 x i16> undef, <32 x i16> undef)
318
319  %I8 = call i8 @llvm.umin.i8(i8 undef, i8 undef)
320  %V2I8  = call <2 x i8>  @llvm.umin.v2i8(<2 x i8> undef, <2 x i8> undef)
321  %V4I8  = call <4 x i8>  @llvm.umin.v4i8(<4 x i8> undef, <4 x i8> undef)
322  %V8I8  = call <8 x i8>  @llvm.umin.v8i8(<8 x i8> undef, <8 x i8> undef)
323  %V16I8 = call <16 x i8> @llvm.umin.v16i8(<16 x i8> undef, <16 x i8> undef)
324  %V32I8 = call <32 x i8> @llvm.umin.v32i8(<32 x i8> undef, <32 x i8> undef)
325  %V64I8 = call <64 x i8> @llvm.umin.v64i8(<64 x i8> undef, <64 x i8> undef)
326
327  ret i32 undef
328}
329
330
331declare i64        @llvm.umax.i64(i64, i64)
332declare <2 x i64>  @llvm.umax.v2i64(<2 x i64>, <2 x i64>)
333declare <4 x i64>  @llvm.umax.v4i64(<4 x i64>, <4 x i64>)
334declare <8 x i64>  @llvm.umax.v8i64(<8 x i64>, <8 x i64>)
335
336declare i32        @llvm.umax.i32(i32, i32)
337declare <2 x i32>  @llvm.umax.v2i32(<2 x i32>, <2 x i32>)
338declare <4 x i32>  @llvm.umax.v4i32(<4 x i32>, <4 x i32>)
339declare <8 x i32>  @llvm.umax.v8i32(<8 x i32>, <8 x i32>)
340declare <16 x i32> @llvm.umax.v16i32(<16 x i32>, <16 x i32>)
341
342declare i16        @llvm.umax.i16(i16, i16)
343declare <2 x i16>  @llvm.umax.v2i16(<2 x i16>, <2 x i16>)
344declare <4 x i16>  @llvm.umax.v4i16(<4 x i16>, <4 x i16>)
345declare <8 x i16>  @llvm.umax.v8i16(<8 x i16>, <8 x i16>)
346declare <16 x i16> @llvm.umax.v16i16(<16 x i16>, <16 x i16>)
347declare <32 x i16> @llvm.umax.v32i16(<32 x i16>, <32 x i16>)
348
349declare i8         @llvm.umax.i8(i8,  i8)
350declare <2 x i8>   @llvm.umax.v2i8(<2 x i8>, <2 x i8>)
351declare <4 x i8>   @llvm.umax.v4i8(<4 x i8>, <4 x i8>)
352declare <8 x i8>   @llvm.umax.v8i8(<8 x i8>, <8 x i8>)
353declare <16 x i8>  @llvm.umax.v16i8(<16 x i8>, <16 x i8>)
354declare <32 x i8>  @llvm.umax.v32i8(<32 x i8>, <32 x i8>)
355declare <64 x i8>  @llvm.umax.v64i8(<64 x i8>, <64 x i8>)
356
357define i32 @sub(i32 %arg) {
358; MVE-RECIP-LABEL: 'sub'
359; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
360; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
361; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
362; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
363; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
364; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <2 x i32> @llvm.umax.v2i32(<2 x i32> undef, <2 x i32> undef)
365; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
366; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
367; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
368; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umax.i16(i16 undef, i16 undef)
369; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I16 = call <2 x i16> @llvm.umax.v2i16(<2 x i16> undef, <2 x i16> undef)
370; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.umax.v4i16(<4 x i16> undef, <4 x i16> undef)
371; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umax.v8i16(<8 x i16> undef, <8 x i16> undef)
372; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umax.v16i16(<16 x i16> undef, <16 x i16> undef)
373; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call <32 x i16> @llvm.umax.v32i16(<32 x i16> undef, <32 x i16> undef)
374; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.umax.i8(i8 undef, i8 undef)
375; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I8 = call <2 x i8> @llvm.umax.v2i8(<2 x i8> undef, <2 x i8> undef)
376; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I8 = call <4 x i8> @llvm.umax.v4i8(<4 x i8> undef, <4 x i8> undef)
377; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.umax.v8i8(<8 x i8> undef, <8 x i8> undef)
378; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call <16 x i8> @llvm.umax.v16i8(<16 x i8> undef, <16 x i8> undef)
379; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call <32 x i8> @llvm.umax.v32i8(<32 x i8> undef, <32 x i8> undef)
380; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call <64 x i8> @llvm.umax.v64i8(<64 x i8> undef, <64 x i8> undef)
381; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
382;
383; MVE-SIZE-LABEL: 'sub'
384; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
385; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
386; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
387; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
388; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
389; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.umax.v2i32(<2 x i32> undef, <2 x i32> undef)
390; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
391; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
392; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
393; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.umax.i16(i16 undef, i16 undef)
394; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I16 = call <2 x i16> @llvm.umax.v2i16(<2 x i16> undef, <2 x i16> undef)
395; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I16 = call <4 x i16> @llvm.umax.v4i16(<4 x i16> undef, <4 x i16> undef)
396; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.umax.v8i16(<8 x i16> undef, <8 x i16> undef)
397; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.umax.v16i16(<16 x i16> undef, <16 x i16> undef)
398; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.umax.v32i16(<32 x i16> undef, <32 x i16> undef)
399; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.umax.i8(i8 undef, i8 undef)
400; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I8 = call <2 x i8> @llvm.umax.v2i8(<2 x i8> undef, <2 x i8> undef)
401; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I8 = call <4 x i8> @llvm.umax.v4i8(<4 x i8> undef, <4 x i8> undef)
402; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I8 = call <8 x i8> @llvm.umax.v8i8(<8 x i8> undef, <8 x i8> undef)
403; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.umax.v16i8(<16 x i8> undef, <16 x i8> undef)
404; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call <32 x i8> @llvm.umax.v32i8(<32 x i8> undef, <32 x i8> undef)
405; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call <64 x i8> @llvm.umax.v64i8(<64 x i8> undef, <64 x i8> undef)
406; MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
407;
408  %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
409  %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
410  %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
411  %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
412
413  %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
414  %V2I32  = call <2 x i32>  @llvm.umax.v2i32(<2 x i32> undef, <2 x i32> undef)
415  %V4I32  = call <4 x i32>  @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
416  %V8I32  = call <8 x i32>  @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
417  %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
418
419  %I16 = call i16 @llvm.umax.i16(i16 undef, i16 undef)
420  %V2I16  = call <2 x i16>  @llvm.umax.v2i16(<2 x i16> undef, <2 x i16> undef)
421  %V4I16  = call <4 x i16>  @llvm.umax.v4i16(<4 x i16> undef, <4 x i16> undef)
422  %V8I16  = call <8 x i16>  @llvm.umax.v8i16(<8 x i16> undef, <8 x i16> undef)
423  %V16I16 = call <16 x i16> @llvm.umax.v16i16(<16 x i16> undef, <16 x i16> undef)
424  %V32I16 = call <32 x i16> @llvm.umax.v32i16(<32 x i16> undef, <32 x i16> undef)
425
426  %I8 = call i8 @llvm.umax.i8(i8 undef, i8 undef)
427  %V2I8  = call <2 x i8>  @llvm.umax.v2i8(<2 x i8> undef, <2 x i8> undef)
428  %V4I8  = call <4 x i8>  @llvm.umax.v4i8(<4 x i8> undef, <4 x i8> undef)
429  %V8I8  = call <8 x i8>  @llvm.umax.v8i8(<8 x i8> undef, <8 x i8> undef)
430  %V16I8 = call <16 x i8> @llvm.umax.v16i8(<16 x i8> undef, <16 x i8> undef)
431  %V32I8 = call <32 x i8> @llvm.umax.v32i8(<32 x i8> undef, <32 x i8> undef)
432  %V64I8 = call <64 x i8> @llvm.umax.v64i8(<64 x i8> undef, <64 x i8> undef)
433
434  ret i32 undef
435}
436
437
438
439declare double        @llvm.minnum.f64(double, double)
440declare <2 x double>  @llvm.minnum.v2f64(<2 x double>, <2 x double>)
441declare <4 x double>  @llvm.minnum.v4f64(<4 x double>, <4 x double>)
442declare <8 x double>  @llvm.minnum.v8f64(<8 x double>, <8 x double>)
443
444declare float        @llvm.minnum.f32(float, float)
445declare <2 x float>  @llvm.minnum.v2f32(<2 x float>, <2 x float>)
446declare <4 x float>  @llvm.minnum.v4f32(<4 x float>, <4 x float>)
447declare <8 x float>  @llvm.minnum.v8f32(<8 x float>, <8 x float>)
448declare <16 x float> @llvm.minnum.v16f32(<16 x float>, <16 x float>)
449
450declare half        @llvm.minnum.f16(half, half)
451declare <2 x half>  @llvm.minnum.v2f16(<2 x half>, <2 x half>)
452declare <4 x half>  @llvm.minnum.v4f16(<4 x half>, <4 x half>)
453declare <8 x half>  @llvm.minnum.v8f16(<8 x half>, <8 x half>)
454declare <16 x half> @llvm.minnum.v16f16(<16 x half>, <16 x half>)
455declare <32 x half> @llvm.minnum.v32f16(<32 x half>, <32 x half>)
456
457define float @minnum(float %arg) {
458; MVEI-RECIP-LABEL: 'minnum'
459; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.minnum.f64(double undef, double undef)
460; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
461; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
462; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
463; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
464; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
465; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
466; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
467; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
468; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F16 = call half @llvm.minnum.f16(half undef, half undef)
469; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
470; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
471; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
472; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
473; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 352 for instruction: %V32F16 = call <32 x half> @llvm.minnum.v32f16(<32 x half> undef, <32 x half> undef)
474; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float undef
475;
476; MVEI-SIZE-LABEL: 'minnum'
477; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.minnum.f64(double undef, double undef)
478; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
479; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
480; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
481; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
482; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
483; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
484; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
485; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
486; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.minnum.f16(half undef, half undef)
487; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
488; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
489; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
490; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
491; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32F16 = call <32 x half> @llvm.minnum.v32f16(<32 x half> undef, <32 x half> undef)
492; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret float undef
493;
494; MVEF-RECIP-LABEL: 'minnum'
495; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.minnum.f64(double undef, double undef)
496; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
497; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
498; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
499; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
500; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
501; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
502; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
503; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
504; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.minnum.f16(half undef, half undef)
505; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
506; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
507; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
508; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
509; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32F16 = call <32 x half> @llvm.minnum.v32f16(<32 x half> undef, <32 x half> undef)
510; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float undef
511;
512; MVEF-SIZE-LABEL: 'minnum'
513; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.minnum.f64(double undef, double undef)
514; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
515; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
516; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
517; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
518; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
519; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
520; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
521; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
522; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.minnum.f16(half undef, half undef)
523; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
524; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
525; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
526; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
527; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32F16 = call <32 x half> @llvm.minnum.v32f16(<32 x half> undef, <32 x half> undef)
528; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret float undef
529;
530  %F64 = call double @llvm.minnum.f64(double undef, double undef)
531  %V2F64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
532  %V4F64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
533  %V8F64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
534
535  %F32 = call float @llvm.minnum.f32(float undef, float undef)
536  %V2F32 = call <2 x float>  @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
537  %V4F32 = call <4 x float>  @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
538  %V8F32 = call <8 x float>  @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
539  %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
540
541  %F16 = call half @llvm.minnum.f16(half undef, half undef)
542  %V2F16 = call <2 x half>  @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
543  %V4F16 = call <4 x half>  @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
544  %V8F16 = call <8 x half>  @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
545  %V16F16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
546  %V32F16 = call <32 x half> @llvm.minnum.v32f16(<32 x half> undef, <32 x half> undef)
547
548  ret float undef
549}
550
551declare double        @llvm.maxnum.f64(double, double)
552declare <2 x double>  @llvm.maxnum.v2f64(<2 x double>, <2 x double>)
553declare <4 x double>  @llvm.maxnum.v4f64(<4 x double>, <4 x double>)
554declare <8 x double>  @llvm.maxnum.v8f64(<8 x double>, <8 x double>)
555
556declare float        @llvm.maxnum.f32(float, float)
557declare <2 x float>  @llvm.maxnum.v2f32(<2 x float>, <2 x float>)
558declare <4 x float>  @llvm.maxnum.v4f32(<4 x float>, <4 x float>)
559declare <8 x float>  @llvm.maxnum.v8f32(<8 x float>, <8 x float>)
560declare <16 x float> @llvm.maxnum.v16f32(<16 x float>, <16 x float>)
561
562declare half        @llvm.maxnum.f16(half, half)
563declare <2 x half>  @llvm.maxnum.v2f16(<2 x half>, <2 x half>)
564declare <4 x half>  @llvm.maxnum.v4f16(<4 x half>, <4 x half>)
565declare <8 x half>  @llvm.maxnum.v8f16(<8 x half>, <8 x half>)
566declare <16 x half> @llvm.maxnum.v16f16(<16 x half>, <16 x half>)
567declare <32 x half> @llvm.maxnum.v32f16(<32 x half>, <32 x half>)
568
569define float @maxnum(float %arg) {
570; MVEI-RECIP-LABEL: 'maxnum'
571; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.maxnum.f64(double undef, double undef)
572; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
573; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
574; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
575; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
576; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
577; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
578; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
579; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
580; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F16 = call half @llvm.maxnum.f16(half undef, half undef)
581; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F16 = call <2 x half> @llvm.maxnum.v2f16(<2 x half> undef, <2 x half> undef)
582; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F16 = call <4 x half> @llvm.maxnum.v4f16(<4 x half> undef, <4 x half> undef)
583; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F16 = call <8 x half> @llvm.maxnum.v8f16(<8 x half> undef, <8 x half> undef)
584; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F16 = call <16 x half> @llvm.maxnum.v16f16(<16 x half> undef, <16 x half> undef)
585; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 352 for instruction: %V32F16 = call <32 x half> @llvm.maxnum.v32f16(<32 x half> undef, <32 x half> undef)
586; MVEI-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float undef
587;
588; MVEI-SIZE-LABEL: 'maxnum'
589; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.maxnum.f64(double undef, double undef)
590; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
591; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
592; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
593; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
594; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
595; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
596; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
597; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
598; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.maxnum.f16(half undef, half undef)
599; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F16 = call <2 x half> @llvm.maxnum.v2f16(<2 x half> undef, <2 x half> undef)
600; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F16 = call <4 x half> @llvm.maxnum.v4f16(<4 x half> undef, <4 x half> undef)
601; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F16 = call <8 x half> @llvm.maxnum.v8f16(<8 x half> undef, <8 x half> undef)
602; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F16 = call <16 x half> @llvm.maxnum.v16f16(<16 x half> undef, <16 x half> undef)
603; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V32F16 = call <32 x half> @llvm.maxnum.v32f16(<32 x half> undef, <32 x half> undef)
604; MVEI-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret float undef
605;
606; MVEF-RECIP-LABEL: 'maxnum'
607; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.maxnum.f64(double undef, double undef)
608; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
609; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
610; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
611; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
612; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
613; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
614; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
615; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
616; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.maxnum.f16(half undef, half undef)
617; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F16 = call <2 x half> @llvm.maxnum.v2f16(<2 x half> undef, <2 x half> undef)
618; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F16 = call <4 x half> @llvm.maxnum.v4f16(<4 x half> undef, <4 x half> undef)
619; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F16 = call <8 x half> @llvm.maxnum.v8f16(<8 x half> undef, <8 x half> undef)
620; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F16 = call <16 x half> @llvm.maxnum.v16f16(<16 x half> undef, <16 x half> undef)
621; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32F16 = call <32 x half> @llvm.maxnum.v32f16(<32 x half> undef, <32 x half> undef)
622; MVEF-RECIP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float undef
623;
624; MVEF-SIZE-LABEL: 'maxnum'
625; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.maxnum.f64(double undef, double undef)
626; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
627; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
628; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
629; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
630; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
631; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
632; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
633; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
634; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.maxnum.f16(half undef, half undef)
635; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F16 = call <2 x half> @llvm.maxnum.v2f16(<2 x half> undef, <2 x half> undef)
636; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F16 = call <4 x half> @llvm.maxnum.v4f16(<4 x half> undef, <4 x half> undef)
637; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F16 = call <8 x half> @llvm.maxnum.v8f16(<8 x half> undef, <8 x half> undef)
638; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F16 = call <16 x half> @llvm.maxnum.v16f16(<16 x half> undef, <16 x half> undef)
639; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32F16 = call <32 x half> @llvm.maxnum.v32f16(<32 x half> undef, <32 x half> undef)
640; MVEF-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret float undef
641;
642  %F64 = call double @llvm.maxnum.f64(double undef, double undef)
643  %V2F64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
644  %V4F64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
645  %V8F64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
646
647  %F32 = call float @llvm.maxnum.f32(float undef, float undef)
648  %V2F32 = call <2 x float>  @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
649  %V4F32 = call <4 x float>  @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
650  %V8F32 = call <8 x float>  @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
651  %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
652
653  %F16 = call half @llvm.maxnum.f16(half undef, half undef)
654  %V2F16 = call <2 x half>  @llvm.maxnum.v2f16(<2 x half> undef, <2 x half> undef)
655  %V4F16 = call <4 x half>  @llvm.maxnum.v4f16(<4 x half> undef, <4 x half> undef)
656  %V8F16 = call <8 x half>  @llvm.maxnum.v8f16(<8 x half> undef, <8 x half> undef)
657  %V16F16 = call <16 x half> @llvm.maxnum.v16f16(<16 x half> undef, <16 x half> undef)
658  %V32F16 = call <32 x half> @llvm.maxnum.v32f16(<32 x half> undef, <32 x half> undef)
659
660  ret float undef
661}
662