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 i16] 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 i16, i16* %in0, align 2
16 ; SSE2: LV: Found an estimated cost of 22 for VF 2 For instruction: %v0 = load i16, i16* %in0, align 2
17 ; SSE2: LV: Found an estimated cost of 43 for VF 4 For instruction: %v0 = load i16, i16* %in0, align 2
18 ; SSE2: LV: Found an estimated cost of 85 for VF 8 For instruction: %v0 = load i16, i16* %in0, align 2
19 ; SSE2: LV: Found an estimated cost of 170 for VF 16 For instruction: %v0 = load i16, i16* %in0, align 2
21 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, i16* %in0, align 2
22 ; AVX1: LV: Found an estimated cost of 26 for VF 2 For instruction: %v0 = load i16, i16* %in0, align 2
23 ; AVX1: LV: Found an estimated cost of 50 for VF 4 For instruction: %v0 = load i16, i16* %in0, align 2
24 ; AVX1: LV: Found an estimated cost of 99 for VF 8 For instruction: %v0 = load i16, i16* %in0, align 2
25 ; AVX1: LV: Found an estimated cost of 215 for VF 16 For instruction: %v0 = load i16, i16* %in0, align 2
26 ; AVX1: LV: Found an estimated cost of 430 for VF 32 For instruction: %v0 = load i16, i16* %in0, align 2
28 ; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, i16* %in0, align 2
29 ; AVX2: LV: Found an estimated cost of 20 for VF 2 For instruction: %v0 = load i16, i16* %in0, align 2
30 ; AVX2: LV: Found an estimated cost of 40 for VF 4 For instruction: %v0 = load i16, i16* %in0, align 2
31 ; AVX2: LV: Found an estimated cost of 80 for VF 8 For instruction: %v0 = load i16, i16* %in0, align 2
32 ; AVX2: LV: Found an estimated cost of 170 for VF 16 For instruction: %v0 = load i16, i16* %in0, align 2
33 ; AVX2: LV: Found an estimated cost of 340 for VF 32 For instruction: %v0 = load i16, i16* %in0, align 2
35 ; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, i16* %in0, align 2
36 ; AVX512: LV: Found an estimated cost of 11 for VF 2 For instruction: %v0 = load i16, i16* %in0, align 2
37 ; AVX512: LV: Found an estimated cost of 11 for VF 4 For instruction: %v0 = load i16, i16* %in0, align 2
38 ; AVX512: LV: Found an estimated cost of 14 for VF 8 For instruction: %v0 = load i16, i16* %in0, align 2
39 ; AVX512: LV: Found an estimated cost of 28 for VF 16 For instruction: %v0 = load i16, i16* %in0, align 2
40 ; AVX512: LV: Found an estimated cost of 55 for VF 32 For instruction: %v0 = load i16, i16* %in0, align 2
41 ; AVX512: LV: Found an estimated cost of 235 for VF 64 For instruction: %v0 = load i16, i16* %in0, align 2
43 ; CHECK-NOT: LV: Found an estimated cost of {{[0-9]+}} for VF {{[0-9]+}} For instruction: %v0 = load i16, i16* %in0, align 2
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
58 %in0 = getelementptr inbounds [1024 x i16], [1024 x i16]* @A, i64 0, i64 %iv.0
59 %in1 = getelementptr inbounds [1024 x i16], [1024 x i16]* @A, i64 0, i64 %iv.1
60 %in2 = getelementptr inbounds [1024 x i16], [1024 x i16]* @A, i64 0, i64 %iv.2
61 %in3 = getelementptr inbounds [1024 x i16], [1024 x i16]* @A, i64 0, i64 %iv.3
62 %in4 = getelementptr inbounds [1024 x i16], [1024 x i16]* @A, i64 0, i64 %iv.4
64 %v0 = load i16, i16* %in0
65 %v1 = load i16, i16* %in1
66 %v2 = load i16, i16* %in2
67 %v3 = load i16, i16* %in3
68 %v4 = load i16, i16* %in4
70 %reduce.add.0 = add i16 %v0, %v1
71 %reduce.add.1 = add i16 %reduce.add.0, %v2
72 %reduce.add.2 = add i16 %reduce.add.1, %v3
73 %reduce.add.3 = add i16 %reduce.add.2, %v4
75 %reduce.add.3.narrow = trunc i16 %reduce.add.3 to i8
77 %out = getelementptr inbounds [1024 x i8], [1024 x i8]* @B, i64 0, i64 %iv.0
78 store i8 %reduce.add.3.narrow, i8* %out
80 %iv.next = add nuw nsw i64 %iv.0, 5
81 %cmp = icmp ult i64 %iv.next, 1024
82 br i1 %cmp, label %for.body, label %for.cond.cleanup