• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
2; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,SSE2
3; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+ssse3 | FileCheck %s --check-prefixes=SSE,SSSE3
4; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+sse4.2 | FileCheck %s --check-prefixes=SSE,SSE42
5; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1
6; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX2
7; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+avx512f | FileCheck %s --check-prefixes=AVX512,AVX512F
8; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=AVX512,AVX512BW
9; RUN: opt < %s -cost-model -mtriple=x86_64-apple-darwin -analyze -mattr=+avx512f,+avx512dq | FileCheck %s --check-prefixes=AVX512,AVX512DQ
10
11define i32 @reduce_i64(i32 %arg) {
12; SSE-LABEL: 'reduce_i64'
13; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.xor.v1i64(<1 x i64> undef)
14; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> undef)
15; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4 = call i64 @llvm.vector.reduce.xor.v4i64(<4 x i64> undef)
16; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8 = call i64 @llvm.vector.reduce.xor.v8i64(<8 x i64> undef)
17; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16 = call i64 @llvm.vector.reduce.xor.v16i64(<16 x i64> undef)
18; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
19;
20; AVX-LABEL: 'reduce_i64'
21; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.xor.v1i64(<1 x i64> undef)
22; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> undef)
23; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.vector.reduce.xor.v4i64(<4 x i64> undef)
24; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8 = call i64 @llvm.vector.reduce.xor.v8i64(<8 x i64> undef)
25; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i64 @llvm.vector.reduce.xor.v16i64(<16 x i64> undef)
26; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
27;
28; AVX512-LABEL: 'reduce_i64'
29; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.xor.v1i64(<1 x i64> undef)
30; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> undef)
31; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.vector.reduce.xor.v4i64(<4 x i64> undef)
32; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.vector.reduce.xor.v8i64(<8 x i64> undef)
33; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i64 @llvm.vector.reduce.xor.v16i64(<16 x i64> undef)
34; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
35;
36  %V1  = call i64 @llvm.vector.reduce.xor.v1i64(<1 x i64> undef)
37  %V2  = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> undef)
38  %V4  = call i64 @llvm.vector.reduce.xor.v4i64(<4 x i64> undef)
39  %V8  = call i64 @llvm.vector.reduce.xor.v8i64(<8 x i64> undef)
40  %V16 = call i64 @llvm.vector.reduce.xor.v16i64(<16 x i64> undef)
41  ret i32 undef
42}
43
44define i32 @reduce_i32(i32 %arg) {
45; SSE-LABEL: 'reduce_i32'
46; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.vector.reduce.xor.v2i32(<2 x i32> undef)
47; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> undef)
48; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8 = call i32 @llvm.vector.reduce.xor.v8i32(<8 x i32> undef)
49; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i32 @llvm.vector.reduce.xor.v16i32(<16 x i32> undef)
50; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32 = call i32 @llvm.vector.reduce.xor.v32i32(<32 x i32> undef)
51; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
52;
53; AVX-LABEL: 'reduce_i32'
54; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.vector.reduce.xor.v2i32(<2 x i32> undef)
55; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> undef)
56; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.vector.reduce.xor.v8i32(<8 x i32> undef)
57; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i32 @llvm.vector.reduce.xor.v16i32(<16 x i32> undef)
58; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i32 @llvm.vector.reduce.xor.v32i32(<32 x i32> undef)
59; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
60;
61; AVX512-LABEL: 'reduce_i32'
62; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.vector.reduce.xor.v2i32(<2 x i32> undef)
63; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> undef)
64; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.vector.reduce.xor.v8i32(<8 x i32> undef)
65; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.vector.reduce.xor.v16i32(<16 x i32> undef)
66; AVX512-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i32 @llvm.vector.reduce.xor.v32i32(<32 x i32> undef)
67; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
68;
69  %V2  = call i32 @llvm.vector.reduce.xor.v2i32(<2 x i32> undef)
70  %V4  = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> undef)
71  %V8  = call i32 @llvm.vector.reduce.xor.v8i32(<8 x i32> undef)
72  %V16 = call i32 @llvm.vector.reduce.xor.v16i32(<16 x i32> undef)
73  %V32 = call i32 @llvm.vector.reduce.xor.v32i32(<32 x i32> undef)
74  ret i32 undef
75}
76
77define i32 @reduce_i16(i32 %arg) {
78; SSE-LABEL: 'reduce_i16'
79; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i16 @llvm.vector.reduce.xor.v2i16(<2 x i16> undef)
80; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.vector.reduce.xor.v4i16(<4 x i16> undef)
81; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.vector.reduce.xor.v8i16(<8 x i16> undef)
82; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i16 @llvm.vector.reduce.xor.v16i16(<16 x i16> undef)
83; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i16 @llvm.vector.reduce.xor.v32i16(<32 x i16> undef)
84; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V64 = call i16 @llvm.vector.reduce.xor.v64i16(<64 x i16> undef)
85; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
86;
87; AVX-LABEL: 'reduce_i16'
88; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i16 @llvm.vector.reduce.xor.v2i16(<2 x i16> undef)
89; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.vector.reduce.xor.v4i16(<4 x i16> undef)
90; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.vector.reduce.xor.v8i16(<8 x i16> undef)
91; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.vector.reduce.xor.v16i16(<16 x i16> undef)
92; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i16 @llvm.vector.reduce.xor.v32i16(<32 x i16> undef)
93; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i16 @llvm.vector.reduce.xor.v64i16(<64 x i16> undef)
94; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
95;
96; AVX512-LABEL: 'reduce_i16'
97; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i16 @llvm.vector.reduce.xor.v2i16(<2 x i16> undef)
98; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.vector.reduce.xor.v4i16(<4 x i16> undef)
99; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.vector.reduce.xor.v8i16(<8 x i16> undef)
100; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.vector.reduce.xor.v16i16(<16 x i16> undef)
101; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.xor.v32i16(<32 x i16> undef)
102; AVX512-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i16 @llvm.vector.reduce.xor.v64i16(<64 x i16> undef)
103; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
104;
105  %V2  = call i16 @llvm.vector.reduce.xor.v2i16(<2 x i16> undef)
106  %V4  = call i16 @llvm.vector.reduce.xor.v4i16(<4 x i16> undef)
107  %V8  = call i16 @llvm.vector.reduce.xor.v8i16(<8 x i16> undef)
108  %V16 = call i16 @llvm.vector.reduce.xor.v16i16(<16 x i16> undef)
109  %V32 = call i16 @llvm.vector.reduce.xor.v32i16(<32 x i16> undef)
110  %V64 = call i16 @llvm.vector.reduce.xor.v64i16(<64 x i16> undef)
111  ret i32 undef
112}
113
114define i32 @reduce_i8(i32 %arg) {
115; SSE-LABEL: 'reduce_i8'
116; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i8 @llvm.vector.reduce.xor.v2i8(<2 x i8> undef)
117; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i8 @llvm.vector.reduce.xor.v4i8(<4 x i8> undef)
118; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.vector.reduce.xor.v8i8(<8 x i8> undef)
119; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.vector.reduce.xor.v16i8(<16 x i8> undef)
120; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i8 @llvm.vector.reduce.xor.v32i8(<32 x i8> undef)
121; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i8 @llvm.vector.reduce.xor.v64i8(<64 x i8> undef)
122; SSE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V128 = call i8 @llvm.vector.reduce.xor.v128i8(<128 x i8> undef)
123; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
124;
125; AVX-LABEL: 'reduce_i8'
126; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i8 @llvm.vector.reduce.xor.v2i8(<2 x i8> undef)
127; AVX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i8 @llvm.vector.reduce.xor.v4i8(<4 x i8> undef)
128; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.vector.reduce.xor.v8i8(<8 x i8> undef)
129; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.vector.reduce.xor.v16i8(<16 x i8> undef)
130; AVX-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.vector.reduce.xor.v32i8(<32 x i8> undef)
131; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i8 @llvm.vector.reduce.xor.v64i8(<64 x i8> undef)
132; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V128 = call i8 @llvm.vector.reduce.xor.v128i8(<128 x i8> undef)
133; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
134;
135; AVX512-LABEL: 'reduce_i8'
136; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i8 @llvm.vector.reduce.xor.v2i8(<2 x i8> undef)
137; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i8 @llvm.vector.reduce.xor.v4i8(<4 x i8> undef)
138; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.vector.reduce.xor.v8i8(<8 x i8> undef)
139; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.vector.reduce.xor.v16i8(<16 x i8> undef)
140; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.vector.reduce.xor.v32i8(<32 x i8> undef)
141; AVX512-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.xor.v64i8(<64 x i8> undef)
142; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V128 = call i8 @llvm.vector.reduce.xor.v128i8(<128 x i8> undef)
143; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
144;
145  %V2   = call i8 @llvm.vector.reduce.xor.v2i8(<2 x i8> undef)
146  %V4   = call i8 @llvm.vector.reduce.xor.v4i8(<4 x i8> undef)
147  %V8   = call i8 @llvm.vector.reduce.xor.v8i8(<8 x i8> undef)
148  %V16  = call i8 @llvm.vector.reduce.xor.v16i8(<16 x i8> undef)
149  %V32  = call i8 @llvm.vector.reduce.xor.v32i8(<32 x i8> undef)
150  %V64  = call i8 @llvm.vector.reduce.xor.v64i8(<64 x i8> undef)
151  %V128 = call i8 @llvm.vector.reduce.xor.v128i8(<128 x i8> undef)
152  ret i32 undef
153}
154
155define i32 @reduce_i1(i32 %arg) {
156; SSE2-LABEL: 'reduce_i1'
157; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
158; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
159; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
160; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
161; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
162; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
163; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
164; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
165; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
166;
167; SSSE3-LABEL: 'reduce_i1'
168; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
169; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
170; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
171; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
172; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
173; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
174; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
175; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
176; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
177;
178; SSE42-LABEL: 'reduce_i1'
179; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
180; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
181; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
182; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
183; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
184; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
185; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
186; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
187; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
188;
189; AVX1-LABEL: 'reduce_i1'
190; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
191; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
192; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
193; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
194; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
195; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
196; AVX1-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
197; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
198; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
199;
200; AVX2-LABEL: 'reduce_i1'
201; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
202; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
203; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
204; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
205; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
206; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
207; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
208; AVX2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
209; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
210;
211; AVX512F-LABEL: 'reduce_i1'
212; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
213; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
214; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
215; AVX512F-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
216; AVX512F-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
217; AVX512F-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
218; AVX512F-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
219; AVX512F-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
220; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
221;
222; AVX512BW-LABEL: 'reduce_i1'
223; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
224; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
225; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
226; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
227; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
228; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
229; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 775 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
230; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 776 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
231; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
232;
233; AVX512DQ-LABEL: 'reduce_i1'
234; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
235; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
236; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
237; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
238; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
239; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V32 = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
240; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64 = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
241; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
242; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
243;
244  %V1   = call i1 @llvm.vector.reduce.xor.v1i1(<1 x i1> undef)
245  %V2   = call i1 @llvm.vector.reduce.xor.v2i1(<2 x i1> undef)
246  %V4   = call i1 @llvm.vector.reduce.xor.v4i1(<4 x i1> undef)
247  %V8   = call i1 @llvm.vector.reduce.xor.v8i1(<8 x i1> undef)
248  %V16  = call i1 @llvm.vector.reduce.xor.v16i1(<16 x i1> undef)
249  %V32  = call i1 @llvm.vector.reduce.xor.v32i1(<32 x i1> undef)
250  %V64  = call i1 @llvm.vector.reduce.xor.v64i1(<64 x i1> undef)
251  %V128 = call i1 @llvm.vector.reduce.xor.v128i1(<128 x i1> undef)
252  ret i32 undef
253}
254
255declare i64 @llvm.vector.reduce.xor.v1i64(<1 x i64>)
256declare i64 @llvm.vector.reduce.xor.v2i64(<2 x i64>)
257declare i64 @llvm.vector.reduce.xor.v4i64(<4 x i64>)
258declare i64 @llvm.vector.reduce.xor.v8i64(<8 x i64>)
259declare i64 @llvm.vector.reduce.xor.v16i64(<16 x i64>)
260
261declare i32 @llvm.vector.reduce.xor.v2i32(<2 x i32>)
262declare i32 @llvm.vector.reduce.xor.v4i32(<4 x i32>)
263declare i32 @llvm.vector.reduce.xor.v8i32(<8 x i32>)
264declare i32 @llvm.vector.reduce.xor.v16i32(<16 x i32>)
265declare i32 @llvm.vector.reduce.xor.v32i32(<32 x i32>)
266
267declare i16 @llvm.vector.reduce.xor.v2i16(<2 x i16>)
268declare i16 @llvm.vector.reduce.xor.v4i16(<4 x i16>)
269declare i16 @llvm.vector.reduce.xor.v8i16(<8 x i16>)
270declare i16 @llvm.vector.reduce.xor.v16i16(<16 x i16>)
271declare i16 @llvm.vector.reduce.xor.v32i16(<32 x i16>)
272declare i16 @llvm.vector.reduce.xor.v64i16(<64 x i16>)
273
274declare i8 @llvm.vector.reduce.xor.v2i8(<2 x i8>)
275declare i8 @llvm.vector.reduce.xor.v4i8(<4 x i8>)
276declare i8 @llvm.vector.reduce.xor.v8i8(<8 x i8>)
277declare i8 @llvm.vector.reduce.xor.v16i8(<16 x i8>)
278declare i8 @llvm.vector.reduce.xor.v32i8(<32 x i8>)
279declare i8 @llvm.vector.reduce.xor.v64i8(<64 x i8>)
280declare i8 @llvm.vector.reduce.xor.v128i8(<128 x i8>)
281
282declare i1 @llvm.vector.reduce.xor.v1i1(<1 x i1>)
283declare i1 @llvm.vector.reduce.xor.v2i1(<2 x i1>)
284declare i1 @llvm.vector.reduce.xor.v4i1(<4 x i1>)
285declare i1 @llvm.vector.reduce.xor.v8i1(<8 x i1>)
286declare i1 @llvm.vector.reduce.xor.v16i1(<16 x i1>)
287declare i1 @llvm.vector.reduce.xor.v32i1(<32 x i1>)
288declare i1 @llvm.vector.reduce.xor.v64i1(<64 x i1>)
289declare i1 @llvm.vector.reduce.xor.v128i1(<128 x i1>)
290