1 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
2 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mattr=+sse2 -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=SSE2
3 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mattr=+sse4.2 -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=SSE42
4 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mattr=+avx -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=AVX,AVX1
5 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mattr=+avx2 -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=AVX,AVX2
7 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mcpu=skylake -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=AVX,SKL
8 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mcpu=knl -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=AVX512,KNL
9 ; RUN: opt < %s -S -mtriple=x86_64-apple-darwin -mcpu=skx -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=AVX512,SKX
11 define i32 @masked_load() {
12 ; SSE2-LABEL: 'masked_load'
13 ; SSE2-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
14 ; SSE2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
15 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
16 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
17 ; SSE2-NEXT: Cost Model: Found an estimated cost of 75 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
18 ; SSE2-NEXT: Cost Model: Found an estimated cost of 37 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
19 ; SSE2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
20 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
21 ; SSE2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
22 ; SSE2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
23 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
24 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
25 ; SSE2-NEXT: Cost Model: Found an estimated cost of 91 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
26 ; SSE2-NEXT: Cost Model: Found an estimated cost of 45 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
27 ; SSE2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
28 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
29 ; SSE2-NEXT: Cost Model: Found an estimated cost of 158 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
30 ; SSE2-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
31 ; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
32 ; SSE2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
33 ; SSE2-NEXT: Cost Model: Found an estimated cost of 376 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
34 ; SSE2-NEXT: Cost Model: Found an estimated cost of 188 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
35 ; SSE2-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
36 ; SSE2-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
37 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
39 ; SSE42-LABEL: 'masked_load'
40 ; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
41 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
42 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
43 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
44 ; SSE42-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
45 ; SSE42-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
46 ; SSE42-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
47 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
48 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
49 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
50 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
51 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
52 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
53 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
54 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
55 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
56 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
57 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
58 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
59 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
60 ; SSE42-NEXT: Cost Model: Found an estimated cost of 256 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
61 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
62 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
63 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
64 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
66 ; AVX-LABEL: 'masked_load'
67 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
68 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
69 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
70 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
71 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
72 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
73 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
74 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
75 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
76 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
77 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
78 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
79 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
80 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
81 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
82 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
83 ; AVX-NEXT: Cost Model: Found an estimated cost of 133 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
84 ; AVX-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
85 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
86 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
87 ; AVX-NEXT: Cost Model: Found an estimated cost of 262 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
88 ; AVX-NEXT: Cost Model: Found an estimated cost of 131 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
89 ; AVX-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
90 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
91 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
93 ; KNL-LABEL: 'masked_load'
94 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
95 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
96 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
97 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
98 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
99 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
100 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
101 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
102 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
103 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
104 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
105 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
106 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
107 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
108 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
109 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
110 ; KNL-NEXT: Cost Model: Found an estimated cost of 133 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
111 ; KNL-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
112 ; KNL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
113 ; KNL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
114 ; KNL-NEXT: Cost Model: Found an estimated cost of 263 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
115 ; KNL-NEXT: Cost Model: Found an estimated cost of 131 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
116 ; KNL-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
117 ; KNL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
118 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
120 ; SKX-LABEL: 'masked_load'
121 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
122 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
123 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
124 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
125 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
126 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
127 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
128 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
129 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
130 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
131 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
132 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
133 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
134 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
135 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
136 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
137 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
138 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
139 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
140 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
141 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
142 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
143 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
144 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
145 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
147 %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
148 %V4F64 = call <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>* undef, i32 1, <4 x i1> undef, <4 x double> undef)
149 %V2F64 = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* undef, i32 1, <2 x i1> undef, <2 x double> undef)
150 %V1F64 = call <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>* undef, i32 1, <1 x i1> undef, <1 x double> undef)
152 %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
153 %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
154 %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
155 %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
157 %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
158 %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
159 %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
160 %V1I64 = call <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>* undef, i32 1, <1 x i1> undef, <1 x i64> undef)
162 %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
163 %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
164 %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
165 %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
167 %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
168 %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
169 %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
170 %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
172 %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
173 %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
174 %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
175 %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
180 define i32 @masked_store() {
181 ; SSE2-LABEL: 'masked_store'
182 ; SSE2-NEXT: Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
183 ; SSE2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
184 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
185 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
186 ; SSE2-NEXT: Cost Model: Found an estimated cost of 75 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
187 ; SSE2-NEXT: Cost Model: Found an estimated cost of 37 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
188 ; SSE2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
189 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
190 ; SSE2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
191 ; SSE2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
192 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
193 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
194 ; SSE2-NEXT: Cost Model: Found an estimated cost of 91 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
195 ; SSE2-NEXT: Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
196 ; SSE2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
197 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
198 ; SSE2-NEXT: Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
199 ; SSE2-NEXT: Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
200 ; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
201 ; SSE2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
202 ; SSE2-NEXT: Cost Model: Found an estimated cost of 376 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
203 ; SSE2-NEXT: Cost Model: Found an estimated cost of 188 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
204 ; SSE2-NEXT: Cost Model: Found an estimated cost of 94 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
205 ; SSE2-NEXT: Cost Model: Found an estimated cost of 46 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
206 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
208 ; SSE42-LABEL: 'masked_store'
209 ; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
210 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
211 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
212 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
213 ; SSE42-NEXT: Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
214 ; SSE42-NEXT: Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
215 ; SSE42-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
216 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
217 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
218 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
219 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
220 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
221 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
222 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
223 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
224 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
225 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
226 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
227 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
228 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
229 ; SSE42-NEXT: Cost Model: Found an estimated cost of 256 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
230 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
231 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
232 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
233 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
235 ; AVX-LABEL: 'masked_store'
236 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
237 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
238 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
239 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
240 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
241 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
242 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
243 ; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
244 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
245 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
246 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
247 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
248 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
249 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
250 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
251 ; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
252 ; AVX-NEXT: Cost Model: Found an estimated cost of 131 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
253 ; AVX-NEXT: Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
254 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
255 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
256 ; AVX-NEXT: Cost Model: Found an estimated cost of 260 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
257 ; AVX-NEXT: Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
258 ; AVX-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
259 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
260 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
262 ; KNL-LABEL: 'masked_store'
263 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
264 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
265 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
266 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
267 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
268 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
269 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
270 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
271 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
272 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
273 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
274 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
275 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
276 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
277 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
278 ; KNL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
279 ; KNL-NEXT: Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
280 ; KNL-NEXT: Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
281 ; KNL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
282 ; KNL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
283 ; KNL-NEXT: Cost Model: Found an estimated cost of 262 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
284 ; KNL-NEXT: Cost Model: Found an estimated cost of 130 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
285 ; KNL-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
286 ; KNL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
287 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
289 ; SKX-LABEL: 'masked_store'
290 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
291 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
292 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
293 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
294 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
295 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
296 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
297 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
298 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
299 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
300 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
301 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
302 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
303 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
304 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
305 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
306 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
307 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
308 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
309 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
310 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
311 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
312 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
313 ; SKX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
314 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
316 call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
317 call void @llvm.masked.store.v4f64.p0v4f64(<4 x double> undef, <4 x double>* undef, i32 1, <4 x i1> undef)
318 call void @llvm.masked.store.v2f64.p0v2f64(<2 x double> undef, <2 x double>* undef, i32 1, <2 x i1> undef)
319 call void @llvm.masked.store.v1f64.p0v1f64(<1 x double> undef, <1 x double>* undef, i32 1, <1 x i1> undef)
321 call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
322 call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
323 call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
324 call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
326 call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
327 call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
328 call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
329 call void @llvm.masked.store.v1i64.p0v1i64(<1 x i64> undef, <1 x i64>* undef, i32 1, <1 x i1> undef)
331 call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
332 call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
333 call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
334 call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
336 call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
337 call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
338 call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
339 call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
341 call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
342 call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
343 call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
344 call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
349 define i32 @masked_gather() {
350 ; SSE2-LABEL: 'masked_gather'
351 ; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
352 ; SSE2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
353 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
354 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
355 ; SSE2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
356 ; SSE2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
357 ; SSE2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
358 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
359 ; SSE2-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
360 ; SSE2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
361 ; SSE2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
362 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
363 ; SSE2-NEXT: Cost Model: Found an estimated cost of 76 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
364 ; SSE2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
365 ; SSE2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
366 ; SSE2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
367 ; SSE2-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
368 ; SSE2-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
369 ; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
370 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
371 ; SSE2-NEXT: Cost Model: Found an estimated cost of 316 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
372 ; SSE2-NEXT: Cost Model: Found an estimated cost of 158 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
373 ; SSE2-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
374 ; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
375 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
377 ; SSE42-LABEL: 'masked_gather'
378 ; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
379 ; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
380 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
381 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
382 ; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
383 ; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
384 ; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
385 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
386 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
387 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
388 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
389 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
390 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
391 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
392 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
393 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
394 ; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
395 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
396 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
397 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
398 ; SSE42-NEXT: Cost Model: Found an estimated cost of 192 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
399 ; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
400 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
401 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
402 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
404 ; AVX1-LABEL: 'masked_gather'
405 ; AVX1-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
406 ; AVX1-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
407 ; AVX1-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
408 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
409 ; AVX1-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
410 ; AVX1-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
411 ; AVX1-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
412 ; AVX1-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
413 ; AVX1-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
414 ; AVX1-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
415 ; AVX1-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
416 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
417 ; AVX1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
418 ; AVX1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
419 ; AVX1-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
420 ; AVX1-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
421 ; AVX1-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
422 ; AVX1-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
423 ; AVX1-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
424 ; AVX1-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
425 ; AVX1-NEXT: Cost Model: Found an estimated cost of 212 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
426 ; AVX1-NEXT: Cost Model: Found an estimated cost of 106 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
427 ; AVX1-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
428 ; AVX1-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
429 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
431 ; AVX2-LABEL: 'masked_gather'
432 ; AVX2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
433 ; AVX2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
434 ; AVX2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
435 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
436 ; AVX2-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
437 ; AVX2-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
438 ; AVX2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
439 ; AVX2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
440 ; AVX2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
441 ; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
442 ; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
443 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
444 ; AVX2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
445 ; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
446 ; AVX2-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
447 ; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
448 ; AVX2-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
449 ; AVX2-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
450 ; AVX2-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
451 ; AVX2-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
452 ; AVX2-NEXT: Cost Model: Found an estimated cost of 212 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
453 ; AVX2-NEXT: Cost Model: Found an estimated cost of 106 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
454 ; AVX2-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
455 ; AVX2-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
456 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
458 ; SKL-LABEL: 'masked_gather'
459 ; SKL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
460 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
461 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
462 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
463 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
464 ; SKL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
465 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
466 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
467 ; SKL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
468 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
469 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
470 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
471 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
472 ; SKL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
473 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
474 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
475 ; SKL-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
476 ; SKL-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
477 ; SKL-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
478 ; SKL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
479 ; SKL-NEXT: Cost Model: Found an estimated cost of 212 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
480 ; SKL-NEXT: Cost Model: Found an estimated cost of 106 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
481 ; SKL-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
482 ; SKL-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
483 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
485 ; KNL-LABEL: 'masked_gather'
486 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
487 ; KNL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
488 ; KNL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
489 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
490 ; KNL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
491 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
492 ; KNL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
493 ; KNL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
494 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
495 ; KNL-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
496 ; KNL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
497 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
498 ; KNL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
499 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
500 ; KNL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
501 ; KNL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
502 ; KNL-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
503 ; KNL-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
504 ; KNL-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
505 ; KNL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
506 ; KNL-NEXT: Cost Model: Found an estimated cost of 220 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
507 ; KNL-NEXT: Cost Model: Found an estimated cost of 110 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
508 ; KNL-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
509 ; KNL-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
510 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
512 ; SKX-LABEL: 'masked_gather'
513 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
514 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
515 ; SKX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
516 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
517 ; SKX-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
518 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
519 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
520 ; SKX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
521 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
522 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
523 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
524 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
525 ; SKX-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
526 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
527 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
528 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
529 ; SKX-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
530 ; SKX-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
531 ; SKX-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
532 ; SKX-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
533 ; SKX-NEXT: Cost Model: Found an estimated cost of 220 for instruction: %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
534 ; SKX-NEXT: Cost Model: Found an estimated cost of 110 for instruction: %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
535 ; SKX-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
536 ; SKX-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
537 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
539 %V8F64 = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> undef, i32 1, <8 x i1> undef, <8 x double> undef)
540 %V4F64 = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*> undef, i32 1, <4 x i1> undef, <4 x double> undef)
541 %V2F64 = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> undef, i32 1, <2 x i1> undef, <2 x double> undef)
542 %V1F64 = call <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*> undef, i32 1, <1 x i1> undef, <1 x double> undef)
544 %V16F32 = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> undef, i32 1, <16 x i1> undef, <16 x float> undef)
545 %V8F32 = call <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*> undef, i32 1, <8 x i1> undef, <8 x float> undef)
546 %V4F32 = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> undef, i32 1, <4 x i1> undef, <4 x float> undef)
547 %V2F32 = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> undef, i32 1, <2 x i1> undef, <2 x float> undef)
549 %V8I64 = call <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*> undef, i32 1, <8 x i1> undef, <8 x i64> undef)
550 %V4I64 = call <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*> undef, i32 1, <4 x i1> undef, <4 x i64> undef)
551 %V2I64 = call <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*> undef, i32 1, <2 x i1> undef, <2 x i64> undef)
552 %V1I64 = call <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*> undef, i32 1, <1 x i1> undef, <1 x i64> undef)
554 %V16I32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 1, <16 x i1> undef, <16 x i32> undef)
555 %V8I32 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*> undef, i32 1, <8 x i1> undef, <8 x i32> undef)
556 %V4I32 = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> undef, i32 1, <4 x i1> undef, <4 x i32> undef)
557 %V2I32 = call <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*> undef, i32 1, <2 x i1> undef, <2 x i32> undef)
559 %V32I16 = call <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*> undef, i32 1, <32 x i1> undef, <32 x i16> undef)
560 %V16I16 = call <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*> undef, i32 1, <16 x i1> undef, <16 x i16> undef)
561 %V8I16 = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> undef, i32 1, <8 x i1> undef, <8 x i16> undef)
562 %V4I16 = call <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*> undef, i32 1, <4 x i1> undef, <4 x i16> undef)
564 %V64I8 = call <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*> undef, i32 1, <64 x i1> undef, <64 x i8> undef)
565 %V32I8 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*> undef, i32 1, <32 x i1> undef, <32 x i8> undef)
566 %V16I8 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*> undef, i32 1, <16 x i1> undef, <16 x i8> undef)
567 %V8I8 = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> undef, i32 1, <8 x i1> undef, <8 x i8> undef)
572 define i32 @masked_scatter() {
573 ; SSE2-LABEL: 'masked_scatter'
574 ; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
575 ; SSE2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
576 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
577 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
578 ; SSE2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
579 ; SSE2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
580 ; SSE2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
581 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
582 ; SSE2-NEXT: Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
583 ; SSE2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
584 ; SSE2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
585 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
586 ; SSE2-NEXT: Cost Model: Found an estimated cost of 76 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
587 ; SSE2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
588 ; SSE2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
589 ; SSE2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
590 ; SSE2-NEXT: Cost Model: Found an estimated cost of 128 for instruction: call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
591 ; SSE2-NEXT: Cost Model: Found an estimated cost of 64 for instruction: call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
592 ; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
593 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
594 ; SSE2-NEXT: Cost Model: Found an estimated cost of 316 for instruction: call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
595 ; SSE2-NEXT: Cost Model: Found an estimated cost of 158 for instruction: call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
596 ; SSE2-NEXT: Cost Model: Found an estimated cost of 79 for instruction: call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
597 ; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
598 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
600 ; SSE42-LABEL: 'masked_scatter'
601 ; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
602 ; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
603 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
604 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
605 ; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
606 ; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
607 ; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
608 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
609 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
610 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
611 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
612 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
613 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
614 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
615 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
616 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
617 ; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
618 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
619 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
620 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
621 ; SSE42-NEXT: Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
622 ; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
623 ; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
624 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
625 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
627 ; AVX-LABEL: 'masked_scatter'
628 ; AVX-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
629 ; AVX-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
630 ; AVX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
631 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
632 ; AVX-NEXT: Cost Model: Found an estimated cost of 50 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
633 ; AVX-NEXT: Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
634 ; AVX-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
635 ; AVX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
636 ; AVX-NEXT: Cost Model: Found an estimated cost of 28 for instruction: call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
637 ; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
638 ; AVX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
639 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
640 ; AVX-NEXT: Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
641 ; AVX-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
642 ; AVX-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
643 ; AVX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
644 ; AVX-NEXT: Cost Model: Found an estimated cost of 106 for instruction: call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
645 ; AVX-NEXT: Cost Model: Found an estimated cost of 53 for instruction: call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
646 ; AVX-NEXT: Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
647 ; AVX-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
648 ; AVX-NEXT: Cost Model: Found an estimated cost of 210 for instruction: call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
649 ; AVX-NEXT: Cost Model: Found an estimated cost of 105 for instruction: call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
650 ; AVX-NEXT: Cost Model: Found an estimated cost of 52 for instruction: call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
651 ; AVX-NEXT: Cost Model: Found an estimated cost of 26 for instruction: call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
652 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
654 ; KNL-LABEL: 'masked_scatter'
655 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
656 ; KNL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
657 ; KNL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
658 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
659 ; KNL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
660 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
661 ; KNL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
662 ; KNL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
663 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
664 ; KNL-NEXT: Cost Model: Found an estimated cost of 14 for instruction: call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
665 ; KNL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
666 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
667 ; KNL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
668 ; KNL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
669 ; KNL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
670 ; KNL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
671 ; KNL-NEXT: Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
672 ; KNL-NEXT: Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
673 ; KNL-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
674 ; KNL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
675 ; KNL-NEXT: Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
676 ; KNL-NEXT: Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
677 ; KNL-NEXT: Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
678 ; KNL-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
679 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
681 ; SKX-LABEL: 'masked_scatter'
682 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
683 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
684 ; SKX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
685 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
686 ; SKX-NEXT: Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
687 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
688 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
689 ; SKX-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
690 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
691 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
692 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
693 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
694 ; SKX-NEXT: Cost Model: Found an estimated cost of 20 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
695 ; SKX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
696 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
697 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
698 ; SKX-NEXT: Cost Model: Found an estimated cost of 111 for instruction: call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
699 ; SKX-NEXT: Cost Model: Found an estimated cost of 55 for instruction: call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
700 ; SKX-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
701 ; SKX-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
702 ; SKX-NEXT: Cost Model: Found an estimated cost of 219 for instruction: call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
703 ; SKX-NEXT: Cost Model: Found an estimated cost of 109 for instruction: call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
704 ; SKX-NEXT: Cost Model: Found an estimated cost of 54 for instruction: call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
705 ; SKX-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
706 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
708 call void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double> undef, <8 x double*> undef, i32 1, <8 x i1> undef)
709 call void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double> undef, <4 x double*> undef, i32 1, <4 x i1> undef)
710 call void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double> undef, <2 x double*> undef, i32 1, <2 x i1> undef)
711 call void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double> undef, <1 x double*> undef, i32 1, <1 x i1> undef)
713 call void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float> undef, <16 x float*> undef, i32 1, <16 x i1> undef)
714 call void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float> undef, <8 x float*> undef, i32 1, <8 x i1> undef)
715 call void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float> undef, <4 x float*> undef, i32 1, <4 x i1> undef)
716 call void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float> undef, <2 x float*> undef, i32 1, <2 x i1> undef)
718 call void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64> undef, <8 x i64*> undef, i32 1, <8 x i1> undef)
719 call void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64> undef, <4 x i64*> undef, i32 1, <4 x i1> undef)
720 call void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64> undef, <2 x i64*> undef, i32 1, <2 x i1> undef)
721 call void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64> undef, <1 x i64*> undef, i32 1, <1 x i1> undef)
723 call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 1, <16 x i1> undef)
724 call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> undef, <8 x i32*> undef, i32 1, <8 x i1> undef)
725 call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> undef, <4 x i32*> undef, i32 1, <4 x i1> undef)
726 call void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32> undef, <2 x i32*> undef, i32 1, <2 x i1> undef)
728 call void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16> undef, <32 x i16*> undef, i32 1, <32 x i1> undef)
729 call void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16> undef, <16 x i16*> undef, i32 1, <16 x i1> undef)
730 call void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16> undef, <8 x i16*> undef, i32 1, <8 x i1> undef)
731 call void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16> undef, <4 x i16*> undef, i32 1, <4 x i1> undef)
733 call void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8> undef, <64 x i8*> undef, i32 1, <64 x i1> undef)
734 call void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8> undef, <32 x i8*> undef, i32 1, <32 x i1> undef)
735 call void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8> undef, <16 x i8*> undef, i32 1, <16 x i1> undef)
736 call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> undef, <8 x i8*> undef, i32 1, <8 x i1> undef)
741 define i32 @masked_expandload() {
742 ; SSE2-LABEL: 'masked_expandload'
743 ; SSE2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(double* undef, <8 x i1> undef, <8 x double> undef)
744 ; SSE2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(double* undef, <4 x i1> undef, <4 x double> undef)
745 ; SSE2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(double* undef, <2 x i1> undef, <2 x double> undef)
746 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(double* undef, <1 x i1> undef, <1 x double> undef)
747 ; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(float* undef, <16 x i1> undef, <16 x float> undef)
748 ; SSE2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(float* undef, <8 x i1> undef, <8 x float> undef)
749 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(float* undef, <4 x i1> undef, <4 x float> undef)
750 ; SSE2-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(float* undef, <2 x i1> undef, <2 x float> undef)
751 ; SSE2-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(i64* undef, <8 x i1> undef, <8 x i64> undef)
752 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(i64* undef, <4 x i1> undef, <4 x i64> undef)
753 ; SSE2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(i64* undef, <2 x i1> undef, <2 x i64> undef)
754 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(i64* undef, <1 x i1> undef, <1 x i64> undef)
755 ; SSE2-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(i32* undef, <16 x i1> undef, <16 x i32> undef)
756 ; SSE2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(i32* undef, <8 x i1> undef, <8 x i32> undef)
757 ; SSE2-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(i32* undef, <4 x i1> undef, <4 x i32> undef)
758 ; SSE2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(i32* undef, <2 x i1> undef, <2 x i32> undef)
759 ; SSE2-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(i16* undef, <32 x i1> undef, <32 x i16> undef)
760 ; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(i16* undef, <16 x i1> undef, <16 x i16> undef)
761 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(i16* undef, <8 x i1> undef, <8 x i16> undef)
762 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(i16* undef, <4 x i1> undef, <4 x i16> undef)
763 ; SSE2-NEXT: Cost Model: Found an estimated cost of 188 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(i8* undef, <64 x i1> undef, <64 x i8> undef)
764 ; SSE2-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(i8* undef, <32 x i1> undef, <32 x i8> undef)
765 ; SSE2-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(i8* undef, <16 x i1> undef, <16 x i8> undef)
766 ; SSE2-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(i8* undef, <8 x i1> undef, <8 x i8> undef)
767 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
769 ; SSE42-LABEL: 'masked_expandload'
770 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(double* undef, <8 x i1> undef, <8 x double> undef)
771 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(double* undef, <4 x i1> undef, <4 x double> undef)
772 ; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(double* undef, <2 x i1> undef, <2 x double> undef)
773 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(double* undef, <1 x i1> undef, <1 x double> undef)
774 ; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(float* undef, <16 x i1> undef, <16 x float> undef)
775 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(float* undef, <8 x i1> undef, <8 x float> undef)
776 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(float* undef, <4 x i1> undef, <4 x float> undef)
777 ; SSE42-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(float* undef, <2 x i1> undef, <2 x float> undef)
778 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(i64* undef, <8 x i1> undef, <8 x i64> undef)
779 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(i64* undef, <4 x i1> undef, <4 x i64> undef)
780 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(i64* undef, <2 x i1> undef, <2 x i64> undef)
781 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(i64* undef, <1 x i1> undef, <1 x i64> undef)
782 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(i32* undef, <16 x i1> undef, <16 x i32> undef)
783 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(i32* undef, <8 x i1> undef, <8 x i32> undef)
784 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(i32* undef, <4 x i1> undef, <4 x i32> undef)
785 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(i32* undef, <2 x i1> undef, <2 x i32> undef)
786 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(i16* undef, <32 x i1> undef, <32 x i16> undef)
787 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(i16* undef, <16 x i1> undef, <16 x i16> undef)
788 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(i16* undef, <8 x i1> undef, <8 x i16> undef)
789 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(i16* undef, <4 x i1> undef, <4 x i16> undef)
790 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(i8* undef, <64 x i1> undef, <64 x i8> undef)
791 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(i8* undef, <32 x i1> undef, <32 x i8> undef)
792 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(i8* undef, <16 x i1> undef, <16 x i8> undef)
793 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(i8* undef, <8 x i1> undef, <8 x i8> undef)
794 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
796 ; AVX-LABEL: 'masked_expandload'
797 ; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(double* undef, <8 x i1> undef, <8 x double> undef)
798 ; AVX-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(double* undef, <4 x i1> undef, <4 x double> undef)
799 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(double* undef, <2 x i1> undef, <2 x double> undef)
800 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(double* undef, <1 x i1> undef, <1 x double> undef)
801 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(float* undef, <16 x i1> undef, <16 x float> undef)
802 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(float* undef, <8 x i1> undef, <8 x float> undef)
803 ; AVX-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(float* undef, <4 x i1> undef, <4 x float> undef)
804 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(float* undef, <2 x i1> undef, <2 x float> undef)
805 ; AVX-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(i64* undef, <8 x i1> undef, <8 x i64> undef)
806 ; AVX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(i64* undef, <4 x i1> undef, <4 x i64> undef)
807 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(i64* undef, <2 x i1> undef, <2 x i64> undef)
808 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(i64* undef, <1 x i1> undef, <1 x i64> undef)
809 ; AVX-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(i32* undef, <16 x i1> undef, <16 x i32> undef)
810 ; AVX-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(i32* undef, <8 x i1> undef, <8 x i32> undef)
811 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(i32* undef, <4 x i1> undef, <4 x i32> undef)
812 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(i32* undef, <2 x i1> undef, <2 x i32> undef)
813 ; AVX-NEXT: Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(i16* undef, <32 x i1> undef, <32 x i16> undef)
814 ; AVX-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(i16* undef, <16 x i1> undef, <16 x i16> undef)
815 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(i16* undef, <8 x i1> undef, <8 x i16> undef)
816 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(i16* undef, <4 x i1> undef, <4 x i16> undef)
817 ; AVX-NEXT: Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(i8* undef, <64 x i1> undef, <64 x i8> undef)
818 ; AVX-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(i8* undef, <32 x i1> undef, <32 x i8> undef)
819 ; AVX-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(i8* undef, <16 x i1> undef, <16 x i8> undef)
820 ; AVX-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(i8* undef, <8 x i1> undef, <8 x i8> undef)
821 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
823 ; AVX512-LABEL: 'masked_expandload'
824 ; AVX512-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(double* undef, <8 x i1> undef, <8 x double> undef)
825 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(double* undef, <4 x i1> undef, <4 x double> undef)
826 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(double* undef, <2 x i1> undef, <2 x double> undef)
827 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(double* undef, <1 x i1> undef, <1 x double> undef)
828 ; AVX512-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(float* undef, <16 x i1> undef, <16 x float> undef)
829 ; AVX512-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(float* undef, <8 x i1> undef, <8 x float> undef)
830 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(float* undef, <4 x i1> undef, <4 x float> undef)
831 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(float* undef, <2 x i1> undef, <2 x float> undef)
832 ; AVX512-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(i64* undef, <8 x i1> undef, <8 x i64> undef)
833 ; AVX512-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(i64* undef, <4 x i1> undef, <4 x i64> undef)
834 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(i64* undef, <2 x i1> undef, <2 x i64> undef)
835 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(i64* undef, <1 x i1> undef, <1 x i64> undef)
836 ; AVX512-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(i32* undef, <16 x i1> undef, <16 x i32> undef)
837 ; AVX512-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(i32* undef, <8 x i1> undef, <8 x i32> undef)
838 ; AVX512-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(i32* undef, <4 x i1> undef, <4 x i32> undef)
839 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(i32* undef, <2 x i1> undef, <2 x i32> undef)
840 ; AVX512-NEXT: Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(i16* undef, <32 x i1> undef, <32 x i16> undef)
841 ; AVX512-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(i16* undef, <16 x i1> undef, <16 x i16> undef)
842 ; AVX512-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(i16* undef, <8 x i1> undef, <8 x i16> undef)
843 ; AVX512-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(i16* undef, <4 x i1> undef, <4 x i16> undef)
844 ; AVX512-NEXT: Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(i8* undef, <64 x i1> undef, <64 x i8> undef)
845 ; AVX512-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(i8* undef, <32 x i1> undef, <32 x i8> undef)
846 ; AVX512-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(i8* undef, <16 x i1> undef, <16 x i8> undef)
847 ; AVX512-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(i8* undef, <8 x i1> undef, <8 x i8> undef)
848 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
850 %V8F64 = call <8 x double> @llvm.masked.expandload.v8f64(double* undef, <8 x i1> undef, <8 x double> undef)
851 %V4F64 = call <4 x double> @llvm.masked.expandload.v4f64(double* undef, <4 x i1> undef, <4 x double> undef)
852 %V2F64 = call <2 x double> @llvm.masked.expandload.v2f64(double* undef, <2 x i1> undef, <2 x double> undef)
853 %V1F64 = call <1 x double> @llvm.masked.expandload.v1f64(double* undef, <1 x i1> undef, <1 x double> undef)
855 %V16F32 = call <16 x float> @llvm.masked.expandload.v16f32(float* undef, <16 x i1> undef, <16 x float> undef)
856 %V8F32 = call <8 x float> @llvm.masked.expandload.v8f32(float* undef, <8 x i1> undef, <8 x float> undef)
857 %V4F32 = call <4 x float> @llvm.masked.expandload.v4f32(float* undef, <4 x i1> undef, <4 x float> undef)
858 %V2F32 = call <2 x float> @llvm.masked.expandload.v2f32(float* undef, <2 x i1> undef, <2 x float> undef)
860 %V8I64 = call <8 x i64> @llvm.masked.expandload.v8i64(i64* undef, <8 x i1> undef, <8 x i64> undef)
861 %V4I64 = call <4 x i64> @llvm.masked.expandload.v4i64(i64* undef, <4 x i1> undef, <4 x i64> undef)
862 %V2I64 = call <2 x i64> @llvm.masked.expandload.v2i64(i64* undef, <2 x i1> undef, <2 x i64> undef)
863 %V1I64 = call <1 x i64> @llvm.masked.expandload.v1i64(i64* undef, <1 x i1> undef, <1 x i64> undef)
865 %V16I32 = call <16 x i32> @llvm.masked.expandload.v16i32(i32* undef, <16 x i1> undef, <16 x i32> undef)
866 %V8I32 = call <8 x i32> @llvm.masked.expandload.v8i32(i32* undef, <8 x i1> undef, <8 x i32> undef)
867 %V4I32 = call <4 x i32> @llvm.masked.expandload.v4i32(i32* undef, <4 x i1> undef, <4 x i32> undef)
868 %V2I32 = call <2 x i32> @llvm.masked.expandload.v2i32(i32* undef, <2 x i1> undef, <2 x i32> undef)
870 %V32I16 = call <32 x i16> @llvm.masked.expandload.v32i16(i16* undef, <32 x i1> undef, <32 x i16> undef)
871 %V16I16 = call <16 x i16> @llvm.masked.expandload.v16i16(i16* undef, <16 x i1> undef, <16 x i16> undef)
872 %V8I16 = call <8 x i16> @llvm.masked.expandload.v8i16(i16* undef, <8 x i1> undef, <8 x i16> undef)
873 %V4I16 = call <4 x i16> @llvm.masked.expandload.v4i16(i16* undef, <4 x i1> undef, <4 x i16> undef)
875 %V64I8 = call <64 x i8> @llvm.masked.expandload.v64i8(i8* undef, <64 x i1> undef, <64 x i8> undef)
876 %V32I8 = call <32 x i8> @llvm.masked.expandload.v32i8(i8* undef, <32 x i1> undef, <32 x i8> undef)
877 %V16I8 = call <16 x i8> @llvm.masked.expandload.v16i8(i8* undef, <16 x i1> undef, <16 x i8> undef)
878 %V8I8 = call <8 x i8> @llvm.masked.expandload.v8i8(i8* undef, <8 x i1> undef, <8 x i8> undef)
883 define i32 @masked_compressstore() {
884 ; SSE2-LABEL: 'masked_compressstore'
885 ; SSE2-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
886 ; SSE2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
887 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
888 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
889 ; SSE2-NEXT: Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
890 ; SSE2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
891 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
892 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
893 ; SSE2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
894 ; SSE2-NEXT: Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
895 ; SSE2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
896 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
897 ; SSE2-NEXT: Cost Model: Found an estimated cost of 45 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
898 ; SSE2-NEXT: Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
899 ; SSE2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
900 ; SSE2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
901 ; SSE2-NEXT: Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
902 ; SSE2-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
903 ; SSE2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
904 ; SSE2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
905 ; SSE2-NEXT: Cost Model: Found an estimated cost of 192 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
906 ; SSE2-NEXT: Cost Model: Found an estimated cost of 96 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
907 ; SSE2-NEXT: Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
908 ; SSE2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
909 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
911 ; SSE42-LABEL: 'masked_compressstore'
912 ; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
913 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
914 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
915 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
916 ; SSE42-NEXT: Cost Model: Found an estimated cost of 29 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
917 ; SSE42-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
918 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
919 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
920 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
921 ; SSE42-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
922 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
923 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
924 ; SSE42-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
925 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
926 ; SSE42-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
927 ; SSE42-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
928 ; SSE42-NEXT: Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
929 ; SSE42-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
930 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
931 ; SSE42-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
932 ; SSE42-NEXT: Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
933 ; SSE42-NEXT: Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
934 ; SSE42-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
935 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
936 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
938 ; AVX1-LABEL: 'masked_compressstore'
939 ; AVX1-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
940 ; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
941 ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
942 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
943 ; AVX1-NEXT: Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
944 ; AVX1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
945 ; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
946 ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
947 ; AVX1-NEXT: Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
948 ; AVX1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
949 ; AVX1-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
950 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
951 ; AVX1-NEXT: Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
952 ; AVX1-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
953 ; AVX1-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
954 ; AVX1-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
955 ; AVX1-NEXT: Cost Model: Found an estimated cost of 68 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
956 ; AVX1-NEXT: Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
957 ; AVX1-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
958 ; AVX1-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
959 ; AVX1-NEXT: Cost Model: Found an estimated cost of 134 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
960 ; AVX1-NEXT: Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
961 ; AVX1-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
962 ; AVX1-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
963 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
965 ; AVX2-LABEL: 'masked_compressstore'
966 ; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
967 ; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
968 ; AVX2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
969 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
970 ; AVX2-NEXT: Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
971 ; AVX2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
972 ; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
973 ; AVX2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
974 ; AVX2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
975 ; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
976 ; AVX2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
977 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
978 ; AVX2-NEXT: Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
979 ; AVX2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
980 ; AVX2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
981 ; AVX2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
982 ; AVX2-NEXT: Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
983 ; AVX2-NEXT: Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
984 ; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
985 ; AVX2-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
986 ; AVX2-NEXT: Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
987 ; AVX2-NEXT: Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
988 ; AVX2-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
989 ; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
990 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
992 ; SKL-LABEL: 'masked_compressstore'
993 ; SKL-NEXT: Cost Model: Found an estimated cost of 15 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
994 ; SKL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
995 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
996 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
997 ; SKL-NEXT: Cost Model: Found an estimated cost of 31 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
998 ; SKL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
999 ; SKL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
1000 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
1001 ; SKL-NEXT: Cost Model: Found an estimated cost of 19 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
1002 ; SKL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
1003 ; SKL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
1004 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
1005 ; SKL-NEXT: Cost Model: Found an estimated cost of 35 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
1006 ; SKL-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
1007 ; SKL-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
1008 ; SKL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
1009 ; SKL-NEXT: Cost Model: Found an estimated cost of 67 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
1010 ; SKL-NEXT: Cost Model: Found an estimated cost of 34 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
1011 ; SKL-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
1012 ; SKL-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
1013 ; SKL-NEXT: Cost Model: Found an estimated cost of 132 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
1014 ; SKL-NEXT: Cost Model: Found an estimated cost of 66 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
1015 ; SKL-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
1016 ; SKL-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
1017 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
1019 ; AVX512-LABEL: 'masked_compressstore'
1020 ; AVX512-NEXT: Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
1021 ; AVX512-NEXT: Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
1022 ; AVX512-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
1023 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
1024 ; AVX512-NEXT: Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
1025 ; AVX512-NEXT: Cost Model: Found an estimated cost of 23 for instruction: call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
1026 ; AVX512-NEXT: Cost Model: Found an estimated cost of 11 for instruction: call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
1027 ; AVX512-NEXT: Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
1028 ; AVX512-NEXT: Cost Model: Found an estimated cost of 27 for instruction: call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
1029 ; AVX512-NEXT: Cost Model: Found an estimated cost of 13 for instruction: call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
1030 ; AVX512-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
1031 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
1032 ; AVX512-NEXT: Cost Model: Found an estimated cost of 51 for instruction: call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
1033 ; AVX512-NEXT: Cost Model: Found an estimated cost of 25 for instruction: call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
1034 ; AVX512-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
1035 ; AVX512-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
1036 ; AVX512-NEXT: Cost Model: Found an estimated cost of 99 for instruction: call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
1037 ; AVX512-NEXT: Cost Model: Found an estimated cost of 49 for instruction: call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
1038 ; AVX512-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
1039 ; AVX512-NEXT: Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
1040 ; AVX512-NEXT: Cost Model: Found an estimated cost of 195 for instruction: call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
1041 ; AVX512-NEXT: Cost Model: Found an estimated cost of 97 for instruction: call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
1042 ; AVX512-NEXT: Cost Model: Found an estimated cost of 48 for instruction: call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
1043 ; AVX512-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
1044 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
1046 call void @llvm.masked.compressstore.v8f64(<8 x double> undef, double* undef, <8 x i1> undef)
1047 call void @llvm.masked.compressstore.v4f64(<4 x double> undef, double* undef, <4 x i1> undef)
1048 call void @llvm.masked.compressstore.v2f64(<2 x double> undef, double* undef, <2 x i1> undef)
1049 call void @llvm.masked.compressstore.v1f64(<1 x double> undef, double* undef, <1 x i1> undef)
1051 call void @llvm.masked.compressstore.v16f32(<16 x float> undef, float* undef, <16 x i1> undef)
1052 call void @llvm.masked.compressstore.v8f32(<8 x float> undef, float* undef, <8 x i1> undef)
1053 call void @llvm.masked.compressstore.v4f32(<4 x float> undef, float* undef, <4 x i1> undef)
1054 call void @llvm.masked.compressstore.v2f32(<2 x float> undef, float* undef, <2 x i1> undef)
1056 call void @llvm.masked.compressstore.v8i64(<8 x i64> undef, i64* undef, <8 x i1> undef)
1057 call void @llvm.masked.compressstore.v4i64(<4 x i64> undef, i64* undef, <4 x i1> undef)
1058 call void @llvm.masked.compressstore.v2i64(<2 x i64> undef, i64* undef, <2 x i1> undef)
1059 call void @llvm.masked.compressstore.v1i64(<1 x i64> undef, i64* undef, <1 x i1> undef)
1061 call void @llvm.masked.compressstore.v16i32(<16 x i32> undef, i32* undef, <16 x i1> undef)
1062 call void @llvm.masked.compressstore.v8i32(<8 x i32> undef, i32* undef, <8 x i1> undef)
1063 call void @llvm.masked.compressstore.v4i32(<4 x i32> undef, i32* undef, <4 x i1> undef)
1064 call void @llvm.masked.compressstore.v2i32(<2 x i32> undef, i32* undef, <2 x i1> undef)
1066 call void @llvm.masked.compressstore.v32i16(<32 x i16> undef, i16* undef, <32 x i1> undef)
1067 call void @llvm.masked.compressstore.v16i16(<16 x i16> undef, i16* undef, <16 x i1> undef)
1068 call void @llvm.masked.compressstore.v8i16(<8 x i16> undef, i16* undef, <8 x i1> undef)
1069 call void @llvm.masked.compressstore.v4i16(<4 x i16> undef, i16* undef, <4 x i1> undef)
1071 call void @llvm.masked.compressstore.v64i8(<64 x i8> undef, i8* undef, <64 x i1> undef)
1072 call void @llvm.masked.compressstore.v32i8(<32 x i8> undef, i8* undef, <32 x i1> undef)
1073 call void @llvm.masked.compressstore.v16i8(<16 x i8> undef, i8* undef, <16 x i1> undef)
1074 call void @llvm.masked.compressstore.v8i8(<8 x i8> undef, i8* undef, <8 x i1> undef)
1079 define <2 x double> @test1(<2 x i64> %trigger, <2 x double>* %addr, <2 x double> %dst) {
1080 ; SSE2-LABEL: 'test1'
1081 ; SSE2-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %mask = icmp eq <2 x i64> %trigger, zeroinitializer
1082 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* %addr, i32 4, <2 x i1> %mask, <2 x double> %dst)
1083 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1085 ; SSE42-LABEL: 'test1'
1086 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i64> %trigger, zeroinitializer
1087 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %res = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* %addr, i32 4, <2 x i1> %mask, <2 x double> %dst)
1088 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1090 ; AVX-LABEL: 'test1'
1091 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i64> %trigger, zeroinitializer
1092 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* %addr, i32 4, <2 x i1> %mask, <2 x double> %dst)
1093 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1095 ; AVX512-LABEL: 'test1'
1096 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i64> %trigger, zeroinitializer
1097 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %res = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* %addr, i32 4, <2 x i1> %mask, <2 x double> %dst)
1098 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1100 %mask = icmp eq <2 x i64> %trigger, zeroinitializer
1101 %res = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* %addr, i32 4, <2 x i1>%mask, <2 x double>%dst)
1102 ret <2 x double> %res
1105 define <4 x i32> @test2(<4 x i32> %trigger, <4 x i32>* %addr, <4 x i32> %dst) {
1106 ; SSE2-LABEL: 'test2'
1107 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1108 ; SSE2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %res = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr, i32 4, <4 x i1> %mask, <4 x i32> %dst)
1109 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1111 ; SSE42-LABEL: 'test2'
1112 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1113 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %res = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr, i32 4, <4 x i1> %mask, <4 x i32> %dst)
1114 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1116 ; AVX-LABEL: 'test2'
1117 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1118 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr, i32 4, <4 x i1> %mask, <4 x i32> %dst)
1119 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1121 ; AVX512-LABEL: 'test2'
1122 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1123 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %res = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr, i32 4, <4 x i1> %mask, <4 x i32> %dst)
1124 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1126 %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1127 %res = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %addr, i32 4, <4 x i1>%mask, <4 x i32>%dst)
1131 define void @test3(<4 x i32> %trigger, <4 x i32>* %addr, <4 x i32> %val) {
1132 ; SSE2-LABEL: 'test3'
1133 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1134 ; SSE2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %val, <4 x i32>* %addr, i32 4, <4 x i1> %mask)
1135 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1137 ; SSE42-LABEL: 'test3'
1138 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1139 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %val, <4 x i32>* %addr, i32 4, <4 x i1> %mask)
1140 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1142 ; AVX-LABEL: 'test3'
1143 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1144 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %val, <4 x i32>* %addr, i32 4, <4 x i1> %mask)
1145 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1147 ; AVX512-LABEL: 'test3'
1148 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1149 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> %val, <4 x i32>* %addr, i32 4, <4 x i1> %mask)
1150 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1152 %mask = icmp eq <4 x i32> %trigger, zeroinitializer
1153 call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32>%val, <4 x i32>* %addr, i32 4, <4 x i1>%mask)
1157 define <8 x float> @test4(<8 x i32> %trigger, <8 x float>* %addr, <8 x float> %dst) {
1158 ; SSE2-LABEL: 'test4'
1159 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1160 ; SSE2-NEXT: Cost Model: Found an estimated cost of 37 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1161 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1163 ; SSE42-LABEL: 'test4'
1164 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1165 ; SSE42-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1166 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1168 ; AVX1-LABEL: 'test4'
1169 ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1170 ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1171 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1173 ; AVX2-LABEL: 'test4'
1174 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1175 ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1176 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1178 ; SKL-LABEL: 'test4'
1179 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1180 ; SKL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1181 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1183 ; AVX512-LABEL: 'test4'
1184 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1185 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1> %mask, <8 x float> %dst)
1186 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %res
1188 %mask = icmp eq <8 x i32> %trigger, zeroinitializer
1189 %res = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %addr, i32 4, <8 x i1>%mask, <8 x float>%dst)
1190 ret <8 x float> %res
1193 define void @test5(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %val) {
1194 ; SSE2-LABEL: 'test5'
1195 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1196 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
1197 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1199 ; SSE42-LABEL: 'test5'
1200 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1201 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
1202 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1204 ; AVX-LABEL: 'test5'
1205 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1206 ; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
1207 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1209 ; AVX512-LABEL: 'test5'
1210 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1211 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
1212 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1214 %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1215 call void @llvm.masked.store.v2f32.p0v2f32(<2 x float>%val, <2 x float>* %addr, i32 4, <2 x i1>%mask)
1219 define void @test6(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %val) {
1220 ; SSE2-LABEL: 'test6'
1221 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1222 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
1223 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1225 ; SSE42-LABEL: 'test6'
1226 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1227 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
1228 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1230 ; AVX-LABEL: 'test6'
1231 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1232 ; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
1233 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1235 ; AVX512-LABEL: 'test6'
1236 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1237 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
1238 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1240 %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1241 call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32>%val, <2 x i32>* %addr, i32 4, <2 x i1>%mask)
1245 define <2 x float> @test7(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %dst) {
1246 ; SSE2-LABEL: 'test7'
1247 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1248 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
1249 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
1251 ; SSE42-LABEL: 'test7'
1252 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1253 ; SSE42-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
1254 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
1256 ; AVX-LABEL: 'test7'
1257 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1258 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
1259 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
1261 ; AVX512-LABEL: 'test7'
1262 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1263 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
1264 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
1266 %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1267 %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1>%mask, <2 x float>%dst)
1268 ret <2 x float> %res
1271 define <2 x i32> @test8(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %dst) {
1272 ; SSE2-LABEL: 'test8'
1273 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1274 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
1275 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
1277 ; SSE42-LABEL: 'test8'
1278 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1279 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
1280 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
1282 ; AVX-LABEL: 'test8'
1283 ; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1284 ; AVX-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
1285 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
1287 ; AVX512-LABEL: 'test8'
1288 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1289 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
1290 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
1292 %mask = icmp eq <2 x i32> %trigger, zeroinitializer
1293 %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1>%mask, <2 x i32>%dst)
1297 define <2 x double> @test_gather_2f64(<2 x double*> %ptrs, <2 x i1> %mask, <2 x double> %src0) {
1298 ; SSE2-LABEL: 'test_gather_2f64'
1299 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1300 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1302 ; SSE42-LABEL: 'test_gather_2f64'
1303 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1304 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1306 ; AVX1-LABEL: 'test_gather_2f64'
1307 ; AVX1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1308 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1310 ; AVX2-LABEL: 'test_gather_2f64'
1311 ; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1312 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1314 ; SKL-LABEL: 'test_gather_2f64'
1315 ; SKL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1316 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1318 ; AVX512-LABEL: 'test_gather_2f64'
1319 ; AVX512-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1320 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %res
1322 %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*> %ptrs, i32 4, <2 x i1> %mask, <2 x double> %src0)
1323 ret <2 x double> %res
1326 define <4 x i32> @test_gather_4i32(<4 x i32*> %ptrs, <4 x i1> %mask, <4 x i32> %src0) {
1327 ; SSE2-LABEL: 'test_gather_4i32'
1328 ; SSE2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1329 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1331 ; SSE42-LABEL: 'test_gather_4i32'
1332 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1333 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1335 ; AVX1-LABEL: 'test_gather_4i32'
1336 ; AVX1-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1337 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1339 ; AVX2-LABEL: 'test_gather_4i32'
1340 ; AVX2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1341 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1343 ; SKL-LABEL: 'test_gather_4i32'
1344 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1345 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1347 ; KNL-LABEL: 'test_gather_4i32'
1348 ; KNL-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1349 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1351 ; SKX-LABEL: 'test_gather_4i32'
1352 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1353 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1355 %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> %mask, <4 x i32> %src0)
1359 define <4 x i32> @test_gather_4i32_const_mask(<4 x i32*> %ptrs, <4 x i32> %src0) {
1360 ; SSE2-LABEL: 'test_gather_4i32_const_mask'
1361 ; SSE2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1362 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1364 ; SSE42-LABEL: 'test_gather_4i32_const_mask'
1365 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1366 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1368 ; AVX1-LABEL: 'test_gather_4i32_const_mask'
1369 ; AVX1-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1370 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1372 ; AVX2-LABEL: 'test_gather_4i32_const_mask'
1373 ; AVX2-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1374 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1376 ; SKL-LABEL: 'test_gather_4i32_const_mask'
1377 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1378 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1380 ; KNL-LABEL: 'test_gather_4i32_const_mask'
1381 ; KNL-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1382 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1384 ; SKX-LABEL: 'test_gather_4i32_const_mask'
1385 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1386 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %res
1388 %res = call <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> %src0)
1392 define <16 x float> @test_gather_16f32_const_mask(float* %base, <16 x i32> %ind) {
1393 ; SSE2-LABEL: 'test_gather_16f32_const_mask'
1394 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1395 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1396 ; SSE2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1397 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1399 ; SSE42-LABEL: 'test_gather_16f32_const_mask'
1400 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1401 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1402 ; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1403 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1405 ; AVX1-LABEL: 'test_gather_16f32_const_mask'
1406 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1407 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1408 ; AVX1-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1409 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1411 ; AVX2-LABEL: 'test_gather_16f32_const_mask'
1412 ; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1413 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1414 ; AVX2-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1415 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1417 ; SKL-LABEL: 'test_gather_16f32_const_mask'
1418 ; SKL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1419 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1420 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1421 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1423 ; AVX512-LABEL: 'test_gather_16f32_const_mask'
1424 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1425 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1426 ; AVX512-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1427 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1429 %sext_ind = sext <16 x i32> %ind to <16 x i64>
1430 %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1432 %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1433 ret <16 x float>%res
1436 define <16 x float> @test_gather_16f32_var_mask(float* %base, <16 x i32> %ind, <16 x i1>%mask) {
1437 ; SSE2-LABEL: 'test_gather_16f32_var_mask'
1438 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1439 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1440 ; SSE2-NEXT: Cost Model: Found an estimated cost of 77 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1441 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1443 ; SSE42-LABEL: 'test_gather_16f32_var_mask'
1444 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1445 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1446 ; SSE42-NEXT: Cost Model: Found an estimated cost of 61 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1447 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1449 ; AVX1-LABEL: 'test_gather_16f32_var_mask'
1450 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1451 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1452 ; AVX1-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1453 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1455 ; AVX2-LABEL: 'test_gather_16f32_var_mask'
1456 ; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1457 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1458 ; AVX2-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1459 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1461 ; SKL-LABEL: 'test_gather_16f32_var_mask'
1462 ; SKL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1463 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1464 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1465 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1467 ; AVX512-LABEL: 'test_gather_16f32_var_mask'
1468 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1469 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1470 ; AVX512-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1471 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1473 %sext_ind = sext <16 x i32> %ind to <16 x i64>
1474 %gep.v = getelementptr float, float* %base, <16 x i64> %sext_ind
1476 %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1477 ret <16 x float>%res
1480 define <16 x float> @test_gather_16f32_ra_var_mask(<16 x float*> %ptrs, <16 x i32> %ind, <16 x i1>%mask) {
1481 ; SSE2-LABEL: 'test_gather_16f32_ra_var_mask'
1482 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1483 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1484 ; SSE2-NEXT: Cost Model: Found an estimated cost of 77 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1485 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1487 ; SSE42-LABEL: 'test_gather_16f32_ra_var_mask'
1488 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1489 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1490 ; SSE42-NEXT: Cost Model: Found an estimated cost of 61 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1491 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1493 ; AVX1-LABEL: 'test_gather_16f32_ra_var_mask'
1494 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1495 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1496 ; AVX1-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1497 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1499 ; AVX2-LABEL: 'test_gather_16f32_ra_var_mask'
1500 ; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1501 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1502 ; AVX2-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1503 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1505 ; SKL-LABEL: 'test_gather_16f32_ra_var_mask'
1506 ; SKL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1507 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1508 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1509 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1511 ; AVX512-LABEL: 'test_gather_16f32_ra_var_mask'
1512 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1513 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1514 ; AVX512-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1515 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1517 %sext_ind = sext <16 x i32> %ind to <16 x i64>
1518 %gep.v = getelementptr float, <16 x float*> %ptrs, <16 x i64> %sext_ind
1520 %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.v, i32 4, <16 x i1> %mask, <16 x float> undef)
1521 ret <16 x float>%res
1524 define <16 x float> @test_gather_16f32_const_mask2(float* %base, <16 x i32> %ind) {
1525 ; SSE2-LABEL: 'test_gather_16f32_const_mask2'
1526 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1527 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1528 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1529 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1530 ; SSE2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1531 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1533 ; SSE42-LABEL: 'test_gather_16f32_const_mask2'
1534 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1535 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1536 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1537 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1538 ; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1539 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1541 ; AVX1-LABEL: 'test_gather_16f32_const_mask2'
1542 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1543 ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1544 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1545 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1546 ; AVX1-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1547 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1549 ; AVX2-LABEL: 'test_gather_16f32_const_mask2'
1550 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1551 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1552 ; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1553 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1554 ; AVX2-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1555 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1557 ; SKL-LABEL: 'test_gather_16f32_const_mask2'
1558 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1559 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1560 ; SKL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1561 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1562 ; SKL-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1563 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1565 ; AVX512-LABEL: 'test_gather_16f32_const_mask2'
1566 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1567 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1568 ; AVX512-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <16 x i32> %ind to <16 x i64>
1569 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1570 ; AVX512-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1571 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x float> %res
1573 %broadcast.splatinsert = insertelement <16 x float*> poison, float* %base, i32 0
1574 %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> poison, <16 x i32> zeroinitializer
1576 %sext_ind = sext <16 x i32> %ind to <16 x i64>
1577 %gep.random = getelementptr float, <16 x float*> %broadcast.splat, <16 x i64> %sext_ind
1579 %res = call <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*> %gep.random, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x float> undef)
1580 ret <16 x float>%res
1583 define void @test_scatter_16i32(i32* %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) {
1584 ; SSE2-LABEL: 'test_scatter_16i32'
1585 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1586 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1587 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1588 ; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1589 ; SSE2-NEXT: Cost Model: Found an estimated cost of 93 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1590 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1592 ; SSE42-LABEL: 'test_scatter_16i32'
1593 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1594 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1595 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1596 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1597 ; SSE42-NEXT: Cost Model: Found an estimated cost of 65 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1598 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1600 ; AVX1-LABEL: 'test_scatter_16i32'
1601 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1602 ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1603 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1604 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1605 ; AVX1-NEXT: Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1606 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1608 ; AVX2-LABEL: 'test_scatter_16i32'
1609 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1610 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1611 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1612 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1613 ; AVX2-NEXT: Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1614 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1616 ; SKL-LABEL: 'test_scatter_16i32'
1617 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1618 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1619 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1620 ; SKL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1621 ; SKL-NEXT: Cost Model: Found an estimated cost of 71 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1622 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1624 ; AVX512-LABEL: 'test_scatter_16i32'
1625 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1626 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1627 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1628 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %imask = bitcast i16 %mask to <16 x i1>
1629 ; AVX512-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1630 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1632 %broadcast.splatinsert = insertelement <16 x i32*> poison, i32* %base, i32 0
1633 %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> poison, <16 x i32> zeroinitializer
1635 %gep.random = getelementptr i32, <16 x i32*> %broadcast.splat, <16 x i32> %ind
1636 %imask = bitcast i16 %mask to <16 x i1>
1637 call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32>%val, <16 x i32*> %gep.random, i32 4, <16 x i1> %imask)
1641 define void @test_scatter_8i32(<8 x i32>%a1, <8 x i32*> %ptr, <8 x i1>%mask) {
1642 ; SSE2-LABEL: 'test_scatter_8i32'
1643 ; SSE2-NEXT: Cost Model: Found an estimated cost of 47 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> %a1, <8 x i32*> %ptr, i32 4, <8 x i1> %mask)
1644 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1646 ; SSE42-LABEL: 'test_scatter_8i32'
1647 ; SSE42-NEXT: Cost Model: Found an estimated cost of 33 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> %a1, <8 x i32*> %ptr, i32 4, <8 x i1> %mask)
1648 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1650 ; AVX-LABEL: 'test_scatter_8i32'
1651 ; AVX-NEXT: Cost Model: Found an estimated cost of 36 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> %a1, <8 x i32*> %ptr, i32 4, <8 x i1> %mask)
1652 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1654 ; AVX512-LABEL: 'test_scatter_8i32'
1655 ; AVX512-NEXT: Cost Model: Found an estimated cost of 10 for instruction: call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> %a1, <8 x i32*> %ptr, i32 4, <8 x i1> %mask)
1656 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1658 call void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32> %a1, <8 x i32*> %ptr, i32 4, <8 x i1> %mask)
1662 define void @test_scatter_4i32(<4 x i32>%a1, <4 x i32*> %ptr, <4 x i1>%mask) {
1663 ; SSE2-LABEL: 'test_scatter_4i32'
1664 ; SSE2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1665 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1667 ; SSE42-LABEL: 'test_scatter_4i32'
1668 ; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1669 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1671 ; AVX-LABEL: 'test_scatter_4i32'
1672 ; AVX-NEXT: Cost Model: Found an estimated cost of 18 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1673 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1675 ; KNL-LABEL: 'test_scatter_4i32'
1676 ; KNL-NEXT: Cost Model: Found an estimated cost of 21 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1677 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1679 ; SKX-LABEL: 'test_scatter_4i32'
1680 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1681 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
1683 call void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32> %a1, <4 x i32*> %ptr, i32 4, <4 x i1> %mask)
1687 define <4 x float> @test_gather_4f32(float* %ptr, <4 x i32> %ind, <4 x i1>%mask) {
1688 ; SSE2-LABEL: 'test_gather_4f32'
1689 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1690 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1691 ; SSE2-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1692 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1694 ; SSE42-LABEL: 'test_gather_4f32'
1695 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1696 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1697 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1698 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1700 ; AVX1-LABEL: 'test_gather_4f32'
1701 ; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1702 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1703 ; AVX1-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1704 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1706 ; AVX2-LABEL: 'test_gather_4f32'
1707 ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1708 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1709 ; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1710 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1712 ; SKL-LABEL: 'test_gather_4f32'
1713 ; SKL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1714 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1715 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1716 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1718 ; KNL-LABEL: 'test_gather_4f32'
1719 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1720 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1721 ; KNL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1722 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1724 ; SKX-LABEL: 'test_gather_4f32'
1725 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1726 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1727 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1728 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1730 %sext_ind = sext <4 x i32> %ind to <4 x i64>
1731 %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1733 %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> %mask, <4 x float> undef)
1737 define <4 x float> @test_gather_4f32_const_mask(float* %ptr, <4 x i32> %ind) {
1738 ; SSE2-LABEL: 'test_gather_4f32_const_mask'
1739 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1740 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1741 ; SSE2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1742 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1744 ; SSE42-LABEL: 'test_gather_4f32_const_mask'
1745 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1746 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1747 ; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1748 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1750 ; AVX1-LABEL: 'test_gather_4f32_const_mask'
1751 ; AVX1-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1752 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1753 ; AVX1-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1754 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1756 ; AVX2-LABEL: 'test_gather_4f32_const_mask'
1757 ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1758 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1759 ; AVX2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1760 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1762 ; SKL-LABEL: 'test_gather_4f32_const_mask'
1763 ; SKL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1764 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1765 ; SKL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1766 ; SKL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1768 ; KNL-LABEL: 'test_gather_4f32_const_mask'
1769 ; KNL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1770 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1771 ; KNL-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1772 ; KNL-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1774 ; SKX-LABEL: 'test_gather_4f32_const_mask'
1775 ; SKX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %sext_ind = sext <4 x i32> %ind to <4 x i64>
1776 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1777 ; SKX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1778 ; SKX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %res
1780 %sext_ind = sext <4 x i32> %ind to <4 x i64>
1781 %gep.v = getelementptr float, float* %ptr, <4 x i64> %sext_ind
1783 %res = call <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*> %gep.v, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x float> undef)
1787 declare <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>*, i32, <8 x i1>, <8 x double>)
1788 declare <4 x double> @llvm.masked.load.v4f64.p0v4f64(<4 x double>*, i32, <4 x i1>, <4 x double>)
1789 declare <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>*, i32, <2 x i1>, <2 x double>)
1790 declare <1 x double> @llvm.masked.load.v1f64.p0v1f64(<1 x double>*, i32, <1 x i1>, <1 x double>)
1792 declare <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>*, i32, <16 x i1>, <16 x float>)
1793 declare <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>*, i32, <8 x i1>, <8 x float>)
1794 declare <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>*, i32, <4 x i1>, <4 x float>)
1795 declare <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>*, i32, <2 x i1>, <2 x float>)
1797 declare <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>*, i32, <8 x i1>, <8 x i64>)
1798 declare <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>*, i32, <4 x i1>, <4 x i64>)
1799 declare <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>*, i32, <2 x i1>, <2 x i64>)
1800 declare <1 x i64> @llvm.masked.load.v1i64.p0v1i64(<1 x i64>*, i32, <1 x i1>, <1 x i64>)
1802 declare <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>*, i32, <16 x i1>, <16 x i32>)
1803 declare <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>*, i32, <8 x i1>, <8 x i32>)
1804 declare <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>*, i32, <4 x i1>, <4 x i32>)
1805 declare <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>*, i32, <2 x i1>, <2 x i32>)
1807 declare <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>*, i32, <32 x i1>, <32 x i16>)
1808 declare <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>*, i32, <16 x i1>, <16 x i16>)
1809 declare <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>*, i32, <8 x i1>, <8 x i16>)
1810 declare <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>*, i32, <4 x i1>, <4 x i16>)
1812 declare <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>*, i32, <64 x i1>, <64 x i8>)
1813 declare <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>*, i32, <32 x i1>, <32 x i8>)
1814 declare <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>*, i32, <16 x i1>, <16 x i8>)
1815 declare <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>*, i32, <8 x i1>, <8 x i8>)
1817 declare void @llvm.masked.store.v8f64.p0v8f64(<8 x double>, <8 x double>*, i32, <8 x i1>)
1818 declare void @llvm.masked.store.v4f64.p0v4f64(<4 x double>, <4 x double>*, i32, <4 x i1>)
1819 declare void @llvm.masked.store.v2f64.p0v2f64(<2 x double>, <2 x double>*, i32, <2 x i1>)
1820 declare void @llvm.masked.store.v1f64.p0v1f64(<1 x double>, <1 x double>*, i32, <1 x i1>)
1822 declare void @llvm.masked.store.v16f32.p0v16f32(<16 x float>, <16 x float>*, i32, <16 x i1>)
1823 declare void @llvm.masked.store.v8f32.p0v8f32(<8 x float>, <8 x float>*, i32, <8 x i1>)
1824 declare void @llvm.masked.store.v4f32.p0v4f32(<4 x float>, <4 x float>*, i32, <4 x i1>)
1825 declare void @llvm.masked.store.v2f32.p0v2f32(<2 x float>, <2 x float>*, i32, <2 x i1>)
1827 declare void @llvm.masked.store.v8i64.p0v8i64(<8 x i64>, <8 x i64>*, i32, <8 x i1>)
1828 declare void @llvm.masked.store.v4i64.p0v4i64(<4 x i64>, <4 x i64>*, i32, <4 x i1>)
1829 declare void @llvm.masked.store.v2i64.p0v2i64(<2 x i64>, <2 x i64>*, i32, <2 x i1>)
1830 declare void @llvm.masked.store.v1i64.p0v1i64(<1 x i64>, <1 x i64>*, i32, <1 x i1>)
1832 declare void @llvm.masked.store.v16i32.p0v16i32(<16 x i32>, <16 x i32>*, i32, <16 x i1>)
1833 declare void @llvm.masked.store.v8i32.p0v8i32(<8 x i32>, <8 x i32>*, i32, <8 x i1>)
1834 declare void @llvm.masked.store.v4i32.p0v4i32(<4 x i32>, <4 x i32>*, i32, <4 x i1>)
1835 declare void @llvm.masked.store.v2i32.p0v2i32(<2 x i32>, <2 x i32>*, i32, <2 x i1>)
1837 declare void @llvm.masked.store.v32i16.p0v32i16(<32 x i16>, <32 x i16>*, i32, <32 x i1>)
1838 declare void @llvm.masked.store.v16i16.p0v16i16(<16 x i16>, <16 x i16>*, i32, <16 x i1>)
1839 declare void @llvm.masked.store.v8i16.p0v8i16(<8 x i16>, <8 x i16>*, i32, <8 x i1>)
1840 declare void @llvm.masked.store.v4i16.p0v4i16(<4 x i16>, <4 x i16>*, i32, <4 x i1>)
1842 declare void @llvm.masked.store.v64i8.p0v64i8(<64 x i8>, <64 x i8>*, i32, <64 x i1>)
1843 declare void @llvm.masked.store.v32i8.p0v32i8(<32 x i8>, <32 x i8>*, i32, <32 x i1>)
1844 declare void @llvm.masked.store.v16i8.p0v16i8(<16 x i8>, <16 x i8>*, i32, <16 x i1>)
1845 declare void @llvm.masked.store.v8i8.p0v8i8(<8 x i8>, <8 x i8>*, i32, <8 x i1>)
1847 declare <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*>, i32, <8 x i1>, <8 x double>)
1848 declare <4 x double> @llvm.masked.gather.v4f64.v4p0f64(<4 x double*>, i32, <4 x i1>, <4 x double>)
1849 declare <2 x double> @llvm.masked.gather.v2f64.v2p0f64(<2 x double*>, i32, <2 x i1>, <2 x double>)
1850 declare <1 x double> @llvm.masked.gather.v1f64.v1p0f64(<1 x double*>, i32, <1 x i1>, <1 x double>)
1852 declare <16 x float> @llvm.masked.gather.v16f32.v16p0f32(<16 x float*>, i32, <16 x i1>, <16 x float>)
1853 declare <8 x float> @llvm.masked.gather.v8f32.v8p0f32(<8 x float*>, i32, <8 x i1>, <8 x float>)
1854 declare <4 x float> @llvm.masked.gather.v4f32.v4p0f32(<4 x float*>, i32, <4 x i1>, <4 x float>)
1855 declare <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*>, i32, <2 x i1>, <2 x float>)
1857 declare <8 x i64> @llvm.masked.gather.v8i64.v8p0i64(<8 x i64*>, i32, <8 x i1>, <8 x i64>)
1858 declare <4 x i64> @llvm.masked.gather.v4i64.v4p0i64(<4 x i64*>, i32, <4 x i1>, <4 x i64>)
1859 declare <2 x i64> @llvm.masked.gather.v2i64.v2p0i64(<2 x i64*>, i32, <2 x i1>, <2 x i64>)
1860 declare <1 x i64> @llvm.masked.gather.v1i64.v1p0i64(<1 x i64*>, i32, <1 x i1>, <1 x i64>)
1862 declare <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*>, i32, <16 x i1>, <16 x i32>)
1863 declare <8 x i32> @llvm.masked.gather.v8i32.v8p0i32(<8 x i32*>, i32, <8 x i1>, <8 x i32>)
1864 declare <4 x i32> @llvm.masked.gather.v4i32.v4p0i32(<4 x i32*>, i32, <4 x i1>, <4 x i32>)
1865 declare <2 x i32> @llvm.masked.gather.v2i32.v2p0i32(<2 x i32*>, i32, <2 x i1>, <2 x i32>)
1867 declare <32 x i16> @llvm.masked.gather.v32i16.v32p0i16(<32 x i16*>, i32, <32 x i1>, <32 x i16>)
1868 declare <16 x i16> @llvm.masked.gather.v16i16.v16p0i16(<16 x i16*>, i32, <16 x i1>, <16 x i16>)
1869 declare <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*>, i32, <8 x i1>, <8 x i16>)
1870 declare <4 x i16> @llvm.masked.gather.v4i16.v4p0i16(<4 x i16*>, i32, <4 x i1>, <4 x i16>)
1872 declare <64 x i8> @llvm.masked.gather.v64i8.v64p0i8(<64 x i8*>, i32, <64 x i1>, <64 x i8>)
1873 declare <32 x i8> @llvm.masked.gather.v32i8.v32p0i8(<32 x i8*>, i32, <32 x i1>, <32 x i8>)
1874 declare <16 x i8> @llvm.masked.gather.v16i8.v16p0i8(<16 x i8*>, i32, <16 x i1>, <16 x i8>)
1875 declare <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*>, i32, <8 x i1>, <8 x i8>)
1877 declare void @llvm.masked.scatter.v8f64.v8p0f64(<8 x double>, <8 x double*>, i32, <8 x i1>)
1878 declare void @llvm.masked.scatter.v4f64.v4p0f64(<4 x double>, <4 x double*>, i32, <4 x i1>)
1879 declare void @llvm.masked.scatter.v2f64.v2p0f64(<2 x double>, <2 x double*>, i32, <2 x i1>)
1880 declare void @llvm.masked.scatter.v1f64.v1p0f64(<1 x double>, <1 x double*>, i32, <1 x i1>)
1882 declare void @llvm.masked.scatter.v16f32.v16p0f32(<16 x float>, <16 x float*>, i32, <16 x i1>)
1883 declare void @llvm.masked.scatter.v8f32.v8p0f32(<8 x float>, <8 x float*>, i32, <8 x i1>)
1884 declare void @llvm.masked.scatter.v4f32.v4p0f32(<4 x float>, <4 x float*>, i32, <4 x i1>)
1885 declare void @llvm.masked.scatter.v2f32.v2p0f32(<2 x float>, <2 x float*>, i32, <2 x i1>)
1887 declare void @llvm.masked.scatter.v8i64.v8p0i64(<8 x i64>, <8 x i64*>, i32, <8 x i1>)
1888 declare void @llvm.masked.scatter.v4i64.v4p0i64(<4 x i64>, <4 x i64*>, i32, <4 x i1>)
1889 declare void @llvm.masked.scatter.v2i64.v2p0i64(<2 x i64>, <2 x i64*>, i32, <2 x i1>)
1890 declare void @llvm.masked.scatter.v1i64.v1p0i64(<1 x i64>, <1 x i64*>, i32, <1 x i1>)
1892 declare void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32>, <16 x i32*>, i32, <16 x i1>)
1893 declare void @llvm.masked.scatter.v8i32.v8p0i32(<8 x i32>, <8 x i32*>, i32, <8 x i1>)
1894 declare void @llvm.masked.scatter.v4i32.v4p0i32(<4 x i32>, <4 x i32*>, i32, <4 x i1>)
1895 declare void @llvm.masked.scatter.v2i32.v2p0i32(<2 x i32>, <2 x i32*>, i32, <2 x i1>)
1897 declare void @llvm.masked.scatter.v32i16.v32p0i16(<32 x i16>, <32 x i16*>, i32, <32 x i1>)
1898 declare void @llvm.masked.scatter.v16i16.v16p0i16(<16 x i16>, <16 x i16*>, i32, <16 x i1>)
1899 declare void @llvm.masked.scatter.v8i16.v8p0i16(<8 x i16>, <8 x i16*>, i32, <8 x i1>)
1900 declare void @llvm.masked.scatter.v4i16.v4p0i16(<4 x i16>, <4 x i16*>, i32, <4 x i1>)
1902 declare void @llvm.masked.scatter.v64i8.v64p0i8(<64 x i8>, <64 x i8*>, i32, <64 x i1>)
1903 declare void @llvm.masked.scatter.v32i8.v32p0i8(<32 x i8>, <32 x i8*>, i32, <32 x i1>)
1904 declare void @llvm.masked.scatter.v16i8.v16p0i8(<16 x i8>, <16 x i8*>, i32, <16 x i1>)
1905 declare void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8>, <8 x i8*>, i32, <8 x i1>)
1907 declare <8 x double> @llvm.masked.expandload.v8f64(double*, <8 x i1>, <8 x double>)
1908 declare <4 x double> @llvm.masked.expandload.v4f64(double*, <4 x i1>, <4 x double>)
1909 declare <2 x double> @llvm.masked.expandload.v2f64(double*, <2 x i1>, <2 x double>)
1910 declare <1 x double> @llvm.masked.expandload.v1f64(double*, <1 x i1>, <1 x double>)
1912 declare <16 x float> @llvm.masked.expandload.v16f32(float*, <16 x i1>, <16 x float>)
1913 declare <8 x float> @llvm.masked.expandload.v8f32(float*, <8 x i1>, <8 x float>)
1914 declare <4 x float> @llvm.masked.expandload.v4f32(float*, <4 x i1>, <4 x float>)
1915 declare <2 x float> @llvm.masked.expandload.v2f32(float*, <2 x i1>, <2 x float>)
1917 declare <8 x i64> @llvm.masked.expandload.v8i64(i64*, <8 x i1>, <8 x i64>)
1918 declare <4 x i64> @llvm.masked.expandload.v4i64(i64*, <4 x i1>, <4 x i64>)
1919 declare <2 x i64> @llvm.masked.expandload.v2i64(i64*, <2 x i1>, <2 x i64>)
1920 declare <1 x i64> @llvm.masked.expandload.v1i64(i64*, <1 x i1>, <1 x i64>)
1922 declare <16 x i32> @llvm.masked.expandload.v16i32(i32*, <16 x i1>, <16 x i32>)
1923 declare <8 x i32> @llvm.masked.expandload.v8i32(i32*, <8 x i1>, <8 x i32>)
1924 declare <4 x i32> @llvm.masked.expandload.v4i32(i32*, <4 x i1>, <4 x i32>)
1925 declare <2 x i32> @llvm.masked.expandload.v2i32(i32*, <2 x i1>, <2 x i32>)
1927 declare <32 x i16> @llvm.masked.expandload.v32i16(i16*, <32 x i1>, <32 x i16>)
1928 declare <16 x i16> @llvm.masked.expandload.v16i16(i16*, <16 x i1>, <16 x i16>)
1929 declare <8 x i16> @llvm.masked.expandload.v8i16(i16*, <8 x i1>, <8 x i16>)
1930 declare <4 x i16> @llvm.masked.expandload.v4i16(i16*, <4 x i1>, <4 x i16>)
1932 declare <64 x i8> @llvm.masked.expandload.v64i8(i8*, <64 x i1>, <64 x i8>)
1933 declare <32 x i8> @llvm.masked.expandload.v32i8(i8*, <32 x i1>, <32 x i8>)
1934 declare <16 x i8> @llvm.masked.expandload.v16i8(i8*, <16 x i1>, <16 x i8>)
1935 declare <8 x i8> @llvm.masked.expandload.v8i8(i8*, <8 x i1>, <8 x i8>)
1937 declare void @llvm.masked.compressstore.v8f64(<8 x double>, double*, <8 x i1>)
1938 declare void @llvm.masked.compressstore.v4f64(<4 x double>, double*, <4 x i1>)
1939 declare void @llvm.masked.compressstore.v2f64(<2 x double>, double*, <2 x i1>)
1940 declare void @llvm.masked.compressstore.v1f64(<1 x double>, double*, <1 x i1>)
1942 declare void @llvm.masked.compressstore.v16f32(<16 x float>, float*, <16 x i1>)
1943 declare void @llvm.masked.compressstore.v8f32(<8 x float>, float*, <8 x i1>)
1944 declare void @llvm.masked.compressstore.v4f32(<4 x float>, float*, <4 x i1>)
1945 declare void @llvm.masked.compressstore.v2f32(<2 x float>, float*, <2 x i1>)
1947 declare void @llvm.masked.compressstore.v8i64(<8 x i64>, i64*, <8 x i1>)
1948 declare void @llvm.masked.compressstore.v4i64(<4 x i64>, i64*, <4 x i1>)
1949 declare void @llvm.masked.compressstore.v2i64(<2 x i64>, i64*, <2 x i1>)
1950 declare void @llvm.masked.compressstore.v1i64(<1 x i64>, i64*, <1 x i1>)
1952 declare void @llvm.masked.compressstore.v16i32(<16 x i32>, i32*, <16 x i1>)
1953 declare void @llvm.masked.compressstore.v8i32(<8 x i32>, i32*, <8 x i1>)
1954 declare void @llvm.masked.compressstore.v4i32(<4 x i32>, i32*, <4 x i1>)
1955 declare void @llvm.masked.compressstore.v2i32(<2 x i32>, i32*, <2 x i1>)
1957 declare void @llvm.masked.compressstore.v32i16(<32 x i16>, i16*, <32 x i1>)
1958 declare void @llvm.masked.compressstore.v16i16(<16 x i16>, i16*, <16 x i1>)
1959 declare void @llvm.masked.compressstore.v8i16(<8 x i16>, i16*, <8 x i1>)
1960 declare void @llvm.masked.compressstore.v4i16(<4 x i16>, i16*, <4 x i1>)
1962 declare void @llvm.masked.compressstore.v64i8(<64 x i8>, i8*, <64 x i1>)
1963 declare void @llvm.masked.compressstore.v32i8(<32 x i8>, i8*, <32 x i1>)
1964 declare void @llvm.masked.compressstore.v16i8(<16 x i8>, i8*, <16 x i1>)
1965 declare void @llvm.masked.compressstore.v8i8(<8 x i8>, i8*, <8 x i1>)