148 lines · plain
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*%v. = load i64, ptr %in."2; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=SSE23; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX14; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX25; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512vl --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX5126; REQUIRES: asserts7 8target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"9target triple = "x86_64-unknown-linux-gnu"10 11@A = global [1024 x i64] zeroinitializer, align 12812@B = global [1024 x i8] zeroinitializer, align 12813 14define void @test() {15; SSE2-LABEL: 'test'16; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 817; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 818; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 819; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 820; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 821; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 822; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 823; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 824; SSE2: LV: Found an estimated cost of 28 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 825; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 826; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 827; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 828; SSE2: LV: Found an estimated cost of 56 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 829; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 830; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 831; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 832;33; AVX1-LABEL: 'test'34; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 835; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 836; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 837; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 838; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 839; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 840; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 841; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 842; AVX1: LV: Found an estimated cost of 20 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 843; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 844; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 845; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 846; AVX1: LV: Found an estimated cost of 44 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 847; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 848; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 849; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 850; AVX1: LV: Found an estimated cost of 88 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 851; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 852; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 853; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 854;55; AVX2-LABEL: 'test'56; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 857; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 858; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 859; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 860; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 861; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 862; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 863; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 864; AVX2: LV: Found an estimated cost of 8 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 865; AVX2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 866; AVX2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 867; AVX2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 868; AVX2: LV: Found an estimated cost of 12 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 869; AVX2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 870; AVX2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 871; AVX2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 872; AVX2: LV: Found an estimated cost of 28 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 873; AVX2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 874; AVX2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 875; AVX2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 876; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v0 = load i64, ptr %in0, align 877; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v1 = load i64, ptr %in1, align 878; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v2 = load i64, ptr %in2, align 879; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v3 = load i64, ptr %in3, align 880;81; AVX512-LABEL: 'test'82; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 883; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 884; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 885; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 886; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 887; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 888; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 889; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 890; AVX512: LV: Found an estimated cost of 5 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 891; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 892; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 893; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 894; AVX512: LV: Found an estimated cost of 8 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 895; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 896; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 897; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 898; AVX512: LV: Found an estimated cost of 22 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 899; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 8100; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 8101; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 8102; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i64, ptr %in0, align 8103; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v1 = load i64, ptr %in1, align 8104; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v2 = load i64, ptr %in2, align 8105; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v3 = load i64, ptr %in3, align 8106; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i64, ptr %in0, align 8107; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v1 = load i64, ptr %in1, align 8108; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v2 = load i64, ptr %in2, align 8109; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v3 = load i64, ptr %in3, align 8110;111entry:112 br label %for.body113 114for.body:115 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]116 117 %iv.0 = add nuw nsw i64 %iv, 0118 %iv.1 = add nuw nsw i64 %iv, 1119 %iv.2 = add nuw nsw i64 %iv, 2120 %iv.3 = add nuw nsw i64 %iv, 3121 122 %in0 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.0123 %in1 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.1124 %in2 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.2125 %in3 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.3126 127 %v0 = load i64, ptr %in0128 %v1 = load i64, ptr %in1129 %v2 = load i64, ptr %in2130 %v3 = load i64, ptr %in3131 132 %reduce.add.0 = add i64 %v0, %v1133 %reduce.add.1 = add i64 %reduce.add.0, %v2134 %reduce.add.2 = add i64 %reduce.add.1, %v3135 136 %reduce.add.2.narrow = trunc i64 %reduce.add.2 to i8137 138 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0139 store i8 %reduce.add.2.narrow, ptr %out140 141 %iv.next = add nuw nsw i64 %iv.0, 4142 %cmp = icmp ult i64 %iv.next, 1024143 br i1 %cmp, label %for.body, label %for.cond.cleanup144 145for.cond.cleanup:146 ret void147}148