| blob | d8eaa0aad61d57aef18047e513924482705f40f5 |
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 i16, 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=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
7 ; REQUIRES: asserts
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 i16] zeroinitializer, align 128
13 @B = global [1024 x i8] zeroinitializer, align 128
15 define void @test() {
16 ; SSE2-LABEL: 'test'
17 ; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, ptr %in0, align 2
18 ; SSE2: LV: Found an estimated cost of 22 for VF 2 For instruction: %v0 = load i16, ptr %in0, align 2
19 ; SSE2: LV: Found an estimated cost of 43 for VF 4 For instruction: %v0 = load i16, ptr %in0, align 2
20 ; SSE2: LV: Found an estimated cost of 85 for VF 8 For instruction: %v0 = load i16, ptr %in0, align 2
21 ; SSE2: LV: Found an estimated cost of 170 for VF 16 For instruction: %v0 = load i16, ptr %in0, align 2
22 ;
23 ; AVX1-LABEL: 'test'
24 ; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, ptr %in0, align 2
25 ; AVX1: LV: Found an estimated cost of 25 for VF 2 For instruction: %v0 = load i16, ptr %in0, align 2
26 ; AVX1: LV: Found an estimated cost of 43 for VF 4 For instruction: %v0 = load i16, ptr %in0, align 2
27 ; AVX1: LV: Found an estimated cost of 85 for VF 8 For instruction: %v0 = load i16, ptr %in0, align 2
28 ; AVX1: LV: Found an estimated cost of 175 for VF 16 For instruction: %v0 = load i16, ptr %in0, align 2
29 ; AVX1: LV: Found an estimated cost of 350 for VF 32 For instruction: %v0 = load i16, ptr %in0, align 2
30 ;
31 ; AVX2-LABEL: 'test'
32 ; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, ptr %in0, align 2
33 ; AVX2: LV: Found an estimated cost of 20 for VF 2 For instruction: %v0 = load i16, ptr %in0, align 2
34 ; AVX2: LV: Found an estimated cost of 40 for VF 4 For instruction: %v0 = load i16, ptr %in0, align 2
35 ; AVX2: LV: Found an estimated cost of 80 for VF 8 For instruction: %v0 = load i16, ptr %in0, align 2
36 ; AVX2: LV: Found an estimated cost of 165 for VF 16 For instruction: %v0 = load i16, ptr %in0, align 2
37 ; AVX2: LV: Found an estimated cost of 330 for VF 32 For instruction: %v0 = load i16, ptr %in0, align 2
38 ;
39 ; AVX512DQ-LABEL: 'test'
40 ; AVX512DQ: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, ptr %in0, align 2
41 ; AVX512DQ: LV: Found an estimated cost of 25 for VF 2 For instruction: %v0 = load i16, ptr %in0, align 2
42 ; AVX512DQ: LV: Found an estimated cost of 45 for VF 4 For instruction: %v0 = load i16, ptr %in0, align 2
43 ; AVX512DQ: LV: Found an estimated cost of 85 for VF 8 For instruction: %v0 = load i16, ptr %in0, align 2
44 ; AVX512DQ: LV: Found an estimated cost of 175 for VF 16 For instruction: %v0 = load i16, ptr %in0, align 2
45 ; AVX512DQ: LV: Found an estimated cost of 355 for VF 32 For instruction: %v0 = load i16, ptr %in0, align 2
46 ; AVX512DQ: LV: Found an estimated cost of 710 for VF 64 For instruction: %v0 = load i16, ptr %in0, align 2
47 ;
48 ; AVX512BW-LABEL: 'test'
49 ; AVX512BW: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i16, ptr %in0, align 2
50 ; AVX512BW: LV: Found an estimated cost of 11 for VF 2 For instruction: %v0 = load i16, ptr %in0, align 2
51 ; AVX512BW: LV: Found an estimated cost of 11 for VF 4 For instruction: %v0 = load i16, ptr %in0, align 2
52 ; AVX512BW: LV: Found an estimated cost of 14 for VF 8 For instruction: %v0 = load i16, ptr %in0, align 2
53 ; AVX512BW: LV: Found an estimated cost of 28 for VF 16 For instruction: %v0 = load i16, ptr %in0, align 2
54 ; AVX512BW: LV: Found an estimated cost of 55 for VF 32 For instruction: %v0 = load i16, ptr %in0, align 2
55 ; AVX512BW: LV: Found an estimated cost of 235 for VF 64 For instruction: %v0 = load i16, ptr %in0, align 2
56 ;
57 entry:
58 br label %for.body
60 for.body:
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
66 %iv.3 = add nuw nsw i64 %iv, 3
67 %iv.4 = add nuw nsw i64 %iv, 4
69 %in0 = getelementptr inbounds [1024 x i16], ptr @A, i64 0, i64 %iv.0
70 %in1 = getelementptr inbounds [1024 x i16], ptr @A, i64 0, i64 %iv.1
71 %in2 = getelementptr inbounds [1024 x i16], ptr @A, i64 0, i64 %iv.2
72 %in3 = getelementptr inbounds [1024 x i16], ptr @A, i64 0, i64 %iv.3
73 %in4 = getelementptr inbounds [1024 x i16], ptr @A, i64 0, i64 %iv.4
75 %v0 = load i16, ptr %in0
76 %v1 = load i16, ptr %in1
77 %v2 = load i16, ptr %in2
78 %v3 = load i16, ptr %in3
79 %v4 = load i16, ptr %in4
81 %reduce.add.0 = add i16 %v0, %v1
82 %reduce.add.1 = add i16 %reduce.add.0, %v2
83 %reduce.add.2 = add i16 %reduce.add.1, %v3
84 %reduce.add.3 = add i16 %reduce.add.2, %v4
86 %reduce.add.3.narrow = trunc i16 %reduce.add.3 to i8
88 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0
89 store i8 %reduce.add.3.narrow, ptr %out
91 %iv.next = add nuw nsw i64 %iv.0, 5
92 %cmp = icmp ult i64 %iv.next, 1024
93 br i1 %cmp, label %for.body, label %for.cond.cleanup
95 for.cond.cleanup:
96 ret void
97 }