169 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: %v4 = load i64, ptr %in4, align 821; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 822; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 823; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 824; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 825; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 826; SSE2: LV: Found an estimated cost of 35 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 827; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 828; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 829; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 830; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 831; SSE2: LV: Found an estimated cost of 70 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 832; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 833; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 834; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 835; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 836;37; AVX1-LABEL: 'test'38; 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 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 843; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 844; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 845; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 846; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 847; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 848; AVX1: LV: Found an estimated cost of 25 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 849; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 850; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 851; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 852; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 853; AVX1: LV: Found an estimated cost of 55 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 854; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 855; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 856; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 857; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 858; AVX1: LV: Found an estimated cost of 110 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 859; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 860; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 861; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 862; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v4 = load i64, ptr %in4, align 863;64; AVX2-LABEL: 'test'65; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 866; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 867; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 868; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 869; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 870; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 871; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 872; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 873; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 874; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 875; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 876; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 877; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 878; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 879; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 880; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 881; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 882; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 883; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 884; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 885; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 886; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 887; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 888; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 889; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v4 = load i64, ptr %in4, align 890;91; AVX512-LABEL: 'test'92; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 893; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 894; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 895; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 896; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 897; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 898; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 899; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 8100; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 8101; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 8102; AVX512: LV: Found an estimated cost of 9 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 8103; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 8104; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 8105; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 8106; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 8107; AVX512: LV: Found an estimated cost of 18 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 8108; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 8109; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 8110; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 8111; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 8112; AVX512: LV: Found an estimated cost of 35 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 8113; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 8114; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 8115; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 8116; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v4 = load i64, ptr %in4, align 8117; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i64, ptr %in0, align 8118; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v1 = load i64, ptr %in1, align 8119; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v2 = load i64, ptr %in2, align 8120; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v3 = load i64, ptr %in3, align 8121; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v4 = load i64, ptr %in4, align 8122; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v0 = load i64, ptr %in0, align 8123; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v1 = load i64, ptr %in1, align 8124; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v2 = load i64, ptr %in2, align 8125; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v3 = load i64, ptr %in3, align 8126; AVX512: LV: Found an estimated cost of 40 for VF 32 For instruction: %v4 = load i64, ptr %in4, align 8127;128entry:129 br label %for.body130 131for.body:132 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]133 134 %iv.0 = add nuw nsw i64 %iv, 0135 %iv.1 = add nuw nsw i64 %iv, 1136 %iv.2 = add nuw nsw i64 %iv, 2137 %iv.3 = add nuw nsw i64 %iv, 3138 %iv.4 = add nuw nsw i64 %iv, 4139 140 %in0 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.0141 %in1 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.1142 %in2 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.2143 %in3 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.3144 %in4 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.4145 146 %v0 = load i64, ptr %in0147 %v1 = load i64, ptr %in1148 %v2 = load i64, ptr %in2149 %v3 = load i64, ptr %in3150 %v4 = load i64, ptr %in4151 152 %reduce.add.0 = add i64 %v0, %v1153 %reduce.add.1 = add i64 %reduce.add.0, %v2154 %reduce.add.2 = add i64 %reduce.add.1, %v3155 %reduce.add.3 = add i64 %reduce.add.2, %v4156 157 %reduce.add.3.narrow = trunc i64 %reduce.add.3 to i8158 159 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0160 store i8 %reduce.add.3.narrow, ptr %out161 162 %iv.next = add nuw nsw i64 %iv.0, 5163 %cmp = icmp ult i64 %iv.next, 1024164 br i1 %cmp, label %for.body, label %for.cond.cleanup165 166for.cond.cleanup:167 ret void168}169