1 ; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,SSE2
2 ; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX1
3 ; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX2
4 ; RUN: opt -loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512bw,+avx512vl --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefixes=CHECK,AVX512
7 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
8 target triple = "x86_64-unknown-linux-gnu"
10 @A = global [1024 x i8] zeroinitializer, align 128
11 @B = global [1024 x i8] zeroinitializer, align 128
13 ; CHECK: LV: Checking a loop in "test"
15 ; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i8, i8* %in0, align 1
16 ; SSE2: LV: Found an estimated cost of 47 for VF 2 For instruction: %v0 = load i8, i8* %in0, align 1
17 ; SSE2: LV: Found an estimated cost of 90 for VF 4 For instruction: %v0 = load i8, i8* %in0, align 1
18 ; SSE2: LV: Found an estimated cost of 186 for VF 8 For instruction: %v0 = load i8, i8* %in0, align 1
19 ; SSE2: LV: Found an estimated cost of 378 for VF 16 For instruction: %v0 = load i8, i8* %in0, align 1
21 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i8, i8* %in0, align 1
22 ; AVX1: LV: Found an estimated cost of 27 for VF 2 For instruction: %v0 = load i8, i8* %in0, align 1
23 ; AVX1: LV: Found an estimated cost of 59 for VF 4 For instruction: %v0 = load i8, i8* %in0, align 1
24 ; AVX1: LV: Found an estimated cost of 114 for VF 8 For instruction: %v0 = load i8, i8* %in0, align 1
25 ; AVX1: LV: Found an estimated cost of 243 for VF 16 For instruction: %v0 = load i8, i8* %in0, align 1
26 ; AVX1: LV: Found an estimated cost of 498 for VF 32 For instruction: %v0 = load i8, i8* %in0, align 1
28 ; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i8, i8* %in0, align 1
29 ; AVX2: LV: Found an estimated cost of 9 for VF 2 For instruction: %v0 = load i8, i8* %in0, align 1
30 ; AVX2: LV: Found an estimated cost of 17 for VF 4 For instruction: %v0 = load i8, i8* %in0, align 1
31 ; AVX2: LV: Found an estimated cost of 20 for VF 8 For instruction: %v0 = load i8, i8* %in0, align 1
32 ; AVX2: LV: Found an estimated cost of 46 for VF 16 For instruction: %v0 = load i8, i8* %in0, align 1
33 ; AVX2: LV: Found an estimated cost of 88 for VF 32 For instruction: %v0 = load i8, i8* %in0, align 1
35 ; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i8, i8* %in0, align 1
36 ; AVX512: LV: Found an estimated cost of 7 for VF 2 For instruction: %v0 = load i8, i8* %in0, align 1
37 ; AVX512: LV: Found an estimated cost of 25 for VF 4 For instruction: %v0 = load i8, i8* %in0, align 1
38 ; AVX512: LV: Found an estimated cost of 49 for VF 8 For instruction: %v0 = load i8, i8* %in0, align 1
39 ; AVX512: LV: Found an estimated cost of 119 for VF 16 For instruction: %v0 = load i8, i8* %in0, align 1
40 ; AVX512: LV: Found an estimated cost of 237 for VF 32 For instruction: %v0 = load i8, i8* %in0, align 1
41 ; AVX512: LV: Found an estimated cost of 591 for VF 64 For instruction: %v0 = load i8, i8* %in0, align 1
43 ; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i8, i8* %in0, align 1
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
55 %iv.3 = add nuw nsw i64 %iv, 3
56 %iv.4 = add nuw nsw i64 %iv, 4
57 %iv.5 = add nuw nsw i64 %iv, 5
59 %in0 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.0
60 %in1 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.1
61 %in2 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.2
62 %in3 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.3
63 %in4 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.4
64 %in5 = getelementptr inbounds [1024 x i8], [1024 x i8]* @A, i64 0, i64 %iv.5
66 %v0 = load i8, i8* %in0
67 %v1 = load i8, i8* %in1
68 %v2 = load i8, i8* %in2
69 %v3 = load i8, i8* %in3
70 %v4 = load i8, i8* %in4
71 %v5 = load i8, i8* %in5
73 %reduce.add.0 = add i8 %v0, %v1
74 %reduce.add.1 = add i8 %reduce.add.0, %v2
75 %reduce.add.2 = add i8 %reduce.add.1, %v3
76 %reduce.add.3 = add i8 %reduce.add.2, %v4
77 %reduce.add.4 = add i8 %reduce.add.3, %v5
79 %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
80 store i8 %reduce.add.4, i8* %out
82 %iv.next = add nuw nsw i64 %iv.0, 6
83 %cmp = icmp ult i64 %iv.next, 1024
84 br i1 %cmp, label %for.body, label %for.cond.cleanup