1 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --filter "LV: Found an estimated cost of [0-9]+ for VF [0-9]+ For instruction:\s*%valB.loaded = load i32, ptr %inB, align 4"
2 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=SSE
3 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse4.2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=SSE
4 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=AVX1
5 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2,-fast-gather --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=AVX2-SLOWGATHER
6 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2,+fast-gather --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=AVX2-FASTGATHER
7 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=AVX512
11 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
12 target triple = "x86_64-unknown-linux-gnu"
14 @A = global [1024 x i8] zeroinitializer, align 128
15 @B = global [1024 x i32] zeroinitializer, align 128
16 @C = global [1024 x i32] zeroinitializer, align 128
20 ; SSE: LV: Found an estimated cost of 1 for VF 1 For instruction: %valB.loaded = load i32, ptr %inB, align 4
21 ; SSE: LV: Found an estimated cost of 3000000 for VF 2 For instruction: %valB.loaded = load i32, ptr %inB, align 4
22 ; SSE: LV: Found an estimated cost of 3000000 for VF 4 For instruction: %valB.loaded = load i32, ptr %inB, align 4
23 ; SSE: LV: Found an estimated cost of 3000000 for VF 8 For instruction: %valB.loaded = load i32, ptr %inB, align 4
24 ; SSE: LV: Found an estimated cost of 3000000 for VF 16 For instruction: %valB.loaded = load i32, ptr %inB, align 4
27 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %valB.loaded = load i32, ptr %inB, align 4
28 ; AVX1: LV: Found an estimated cost of 3000000 for VF 2 For instruction: %valB.loaded = load i32, ptr %inB, align 4
29 ; AVX1: LV: Found an estimated cost of 3000000 for VF 4 For instruction: %valB.loaded = load i32, ptr %inB, align 4
30 ; AVX1: LV: Found an estimated cost of 3000000 for VF 8 For instruction: %valB.loaded = load i32, ptr %inB, align 4
31 ; AVX1: LV: Found an estimated cost of 3000000 for VF 16 For instruction: %valB.loaded = load i32, ptr %inB, align 4
32 ; AVX1: LV: Found an estimated cost of 3000000 for VF 32 For instruction: %valB.loaded = load i32, ptr %inB, align 4
34 ; AVX2-SLOWGATHER-LABEL: 'test'
35 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 1 for VF 1 For instruction: %valB.loaded = load i32, ptr %inB, align 4
36 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 3000000 for VF 2 For instruction: %valB.loaded = load i32, ptr %inB, align 4
37 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 3000000 for VF 4 For instruction: %valB.loaded = load i32, ptr %inB, align 4
38 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 3000000 for VF 8 For instruction: %valB.loaded = load i32, ptr %inB, align 4
39 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 3000000 for VF 16 For instruction: %valB.loaded = load i32, ptr %inB, align 4
40 ; AVX2-SLOWGATHER: LV: Found an estimated cost of 3000000 for VF 32 For instruction: %valB.loaded = load i32, ptr %inB, align 4
42 ; AVX2-FASTGATHER-LABEL: 'test'
43 ; AVX2-FASTGATHER: LV: Found an estimated cost of 1 for VF 1 For instruction: %valB.loaded = load i32, ptr %inB, align 4
44 ; AVX2-FASTGATHER: LV: Found an estimated cost of 4 for VF 2 For instruction: %valB.loaded = load i32, ptr %inB, align 4
45 ; AVX2-FASTGATHER: LV: Found an estimated cost of 6 for VF 4 For instruction: %valB.loaded = load i32, ptr %inB, align 4
46 ; AVX2-FASTGATHER: LV: Found an estimated cost of 12 for VF 8 For instruction: %valB.loaded = load i32, ptr %inB, align 4
47 ; AVX2-FASTGATHER: LV: Found an estimated cost of 24 for VF 16 For instruction: %valB.loaded = load i32, ptr %inB, align 4
48 ; AVX2-FASTGATHER: LV: Found an estimated cost of 48 for VF 32 For instruction: %valB.loaded = load i32, ptr %inB, align 4
50 ; AVX512-LABEL: 'test'
51 ; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %valB.loaded = load i32, ptr %inB, align 4
52 ; AVX512: LV: Found an estimated cost of 8 for VF 2 For instruction: %valB.loaded = load i32, ptr %inB, align 4
53 ; AVX512: LV: Found an estimated cost of 17 for VF 4 For instruction: %valB.loaded = load i32, ptr %inB, align 4
54 ; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %valB.loaded = load i32, ptr %inB, align 4
55 ; AVX512: LV: Found an estimated cost of 18 for VF 16 For instruction: %valB.loaded = load i32, ptr %inB, align 4
56 ; AVX512: LV: Found an estimated cost of 36 for VF 32 For instruction: %valB.loaded = load i32, ptr %inB, align 4
57 ; AVX512: LV: Found an estimated cost of 72 for VF 64 For instruction: %valB.loaded = load i32, ptr %inB, align 4
63 %iv = phi i64 [ 0, %entry ], [ %iv.next, %end ]
65 %inA = getelementptr inbounds [1024 x i8], ptr @A, i64 0, i64 %iv
66 %valA = load i8, ptr %inA
67 %canLoad = icmp ne i8 %valA, 0
68 br i1 %canLoad, label %load, label %mask
71 %valA.ext = sext i8 %valA to i64
72 %inB = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %valA.ext
73 %valB.loaded = load i32, ptr %inB
80 %valB = phi i32 [ %valB.loaded, %load ], [ 0, %mask ]
81 %out = getelementptr inbounds [1024 x i32], ptr @C, i64 0, i64 %iv
82 store i32 %valB, ptr %out
84 %iv.next = add nuw nsw i64 %iv, 1
85 %cmp = icmp ult i64 %iv.next, 1024
86 br i1 %cmp, label %for.body, label %for.cond.cleanup