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*%v0 = load i32, ptr %in0"
2 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefix=SSE2
3 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefix=AVX1
4 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefix=AVX2
5 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefix=AVX512
8 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
9 target triple = "x86_64-unknown-linux-gnu"
11 @A = global [1024 x i32] zeroinitializer, align 128
12 @B = global [1024 x i8] zeroinitializer, align 128
16 ; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4
17 ; SSE2: LV: Found an estimated cost of 21 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 4
18 ; SSE2: LV: Found an estimated cost of 45 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 4
19 ; SSE2: LV: Found an estimated cost of 90 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 4
20 ; SSE2: LV: Found an estimated cost of 180 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4
23 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4
24 ; AVX1: LV: Found an estimated cost of 14 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 4
25 ; AVX1: LV: Found an estimated cost of 28 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 4
26 ; AVX1: LV: Found an estimated cost of 59 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 4
27 ; AVX1: LV: Found an estimated cost of 118 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4
28 ; AVX1: LV: Found an estimated cost of 236 for VF 32 For instruction: %v0 = load i32, ptr %in0, align 4
31 ; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4
32 ; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 4
33 ; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 4
34 ; AVX2: LV: Found an estimated cost of 16 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 4
35 ; AVX2: LV: Found an estimated cost of 32 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4
36 ; AVX2: LV: Found an estimated cost of 67 for VF 32 For instruction: %v0 = load i32, ptr %in0, align 4
38 ; AVX512-LABEL: 'test'
39 ; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4
40 ; AVX512: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 4
41 ; AVX512: LV: Found an estimated cost of 4 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 4
42 ; AVX512: LV: Found an estimated cost of 6 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 4
43 ; AVX512: LV: Found an estimated cost of 17 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4
44 ; AVX512: LV: Found an estimated cost of 71 for VF 32 For instruction: %v0 = load i32, ptr %in0, align 4
50 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
52 %iv.0 = add nuw nsw i64 %iv, 0
53 %iv.1 = add nuw nsw i64 %iv, 1
54 %iv.2 = add nuw nsw i64 %iv, 2
56 %in0 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.0
57 %in1 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.1
58 %in2 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.2
60 %v0 = load i32, ptr %in0
61 %v1 = load i32, ptr %in1
62 %v2 = load i32, ptr %in2
64 %reduce.add.0 = add i32 %v0, %v1
65 %reduce.add.1 = add i32 %reduce.add.0, %v2
66 %reduce.add.2 = add i32 %reduce.add.1, 0
68 %reduce.add.2.narrow = trunc i32 %reduce.add.2 to i8
70 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0
71 store i8 %reduce.add.2.narrow, ptr %out
73 %iv.next = add nuw nsw i64 %iv, 4
74 %cmp = icmp ult i64 %iv.next, 1024
75 br i1 %cmp, label %for.body, label %for.cond.cleanup