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*store i16 %v2, ptr %out2"
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=AVX512DQ
6 ; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512vl,+avx512bw --debug-only=loop-vectorize < %s 2>&1 | FileCheck %s --check-prefix=AVX512BW
9 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
10 target triple = "x86_64-unknown-linux-gnu"
12 @A = global [1024 x i8] zeroinitializer, align 128
13 @B = global [1024 x i16] zeroinitializer, align 128
17 ; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i16 %v2, ptr %out2, align 2
18 ; SSE2: LV: Found an estimated cost of 16 for VF 2 For instruction: store i16 %v2, ptr %out2, align 2
19 ; SSE2: LV: Found an estimated cost of 26 for VF 4 For instruction: store i16 %v2, ptr %out2, align 2
20 ; SSE2: LV: Found an estimated cost of 51 for VF 8 For instruction: store i16 %v2, ptr %out2, align 2
21 ; SSE2: LV: Found an estimated cost of 102 for VF 16 For instruction: store i16 %v2, ptr %out2, align 2
24 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store i16 %v2, ptr %out2, align 2
25 ; AVX1: LV: Found an estimated cost of 15 for VF 2 For instruction: store i16 %v2, ptr %out2, align 2
26 ; AVX1: LV: Found an estimated cost of 29 for VF 4 For instruction: store i16 %v2, ptr %out2, align 2
27 ; AVX1: LV: Found an estimated cost of 52 for VF 8 For instruction: store i16 %v2, ptr %out2, align 2
28 ; AVX1: LV: Found an estimated cost of 105 for VF 16 For instruction: store i16 %v2, ptr %out2, align 2
29 ; AVX1: LV: Found an estimated cost of 210 for VF 32 For instruction: store i16 %v2, ptr %out2, align 2
32 ; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i16 %v2, ptr %out2, align 2
33 ; AVX2: LV: Found an estimated cost of 7 for VF 2 For instruction: store i16 %v2, ptr %out2, align 2
34 ; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: store i16 %v2, ptr %out2, align 2
35 ; AVX2: LV: Found an estimated cost of 14 for VF 8 For instruction: store i16 %v2, ptr %out2, align 2
36 ; AVX2: LV: Found an estimated cost of 30 for VF 16 For instruction: store i16 %v2, ptr %out2, align 2
37 ; AVX2: LV: Found an estimated cost of 60 for VF 32 For instruction: store i16 %v2, ptr %out2, align 2
39 ; AVX512DQ-LABEL: 'test'
40 ; AVX512DQ: LV: Found an estimated cost of 1 for VF 1 For instruction: store i16 %v2, ptr %out2, align 2
41 ; AVX512DQ: LV: Found an estimated cost of 7 for VF 2 For instruction: store i16 %v2, ptr %out2, align 2
42 ; AVX512DQ: LV: Found an estimated cost of 9 for VF 4 For instruction: store i16 %v2, ptr %out2, align 2
43 ; AVX512DQ: LV: Found an estimated cost of 15 for VF 8 For instruction: store i16 %v2, ptr %out2, align 2
44 ; AVX512DQ: LV: Found an estimated cost of 29 for VF 16 For instruction: store i16 %v2, ptr %out2, align 2
45 ; AVX512DQ: LV: Found an estimated cost of 57 for VF 32 For instruction: store i16 %v2, ptr %out2, align 2
46 ; AVX512DQ: LV: Found an estimated cost of 426 for VF 64 For instruction: store i16 %v2, ptr %out2, align 2
48 ; AVX512BW-LABEL: 'test'
49 ; AVX512BW: LV: Found an estimated cost of 1 for VF 1 For instruction: store i16 %v2, ptr %out2, align 2
50 ; AVX512BW: LV: Found an estimated cost of 6 for VF 2 For instruction: store i16 %v2, ptr %out2, align 2
51 ; AVX512BW: LV: Found an estimated cost of 6 for VF 4 For instruction: store i16 %v2, ptr %out2, align 2
52 ; AVX512BW: LV: Found an estimated cost of 6 for VF 8 For instruction: store i16 %v2, ptr %out2, align 2
53 ; AVX512BW: LV: Found an estimated cost of 12 for VF 16 For instruction: store i16 %v2, ptr %out2, align 2
54 ; AVX512BW: LV: Found an estimated cost of 18 for VF 32 For instruction: store i16 %v2, ptr %out2, align 2
55 ; AVX512BW: LV: Found an estimated cost of 36 for VF 64 For instruction: store i16 %v2, ptr %out2, align 2
61 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
63 %iv.0 = add nuw nsw i64 %iv, 0
64 %iv.1 = add nuw nsw i64 %iv, 1
65 %iv.2 = add nuw nsw i64 %iv, 2
67 %in = getelementptr inbounds [1024 x i8], ptr @A, i64 0, i64 %iv.0
68 %v.narrow = load i8, ptr %in
70 %v = zext i8 %v.narrow to i16
76 %out0 = getelementptr inbounds [1024 x i16], ptr @B, i64 0, i64 %iv.0
77 %out1 = getelementptr inbounds [1024 x i16], ptr @B, i64 0, i64 %iv.1
78 %out2 = getelementptr inbounds [1024 x i16], ptr @B, i64 0, i64 %iv.2
80 store i16 %v0, ptr %out0
81 store i16 %v1, ptr %out1
82 store i16 %v2, ptr %out2
84 %iv.next = add nuw nsw i64 %iv.0, 3
85 %cmp = icmp ult i64 %iv.next, 1024
86 br i1 %cmp, label %for.body, label %for.cond.cleanup