236 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: %v5 = load i64, ptr %in5, align 822; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 823; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 824; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 825; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 826; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 827; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 828; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 829; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 830; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 831; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 832; SSE2: LV: Found an estimated cost of 56 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 833; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 834; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 835; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 836; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 837; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v5 = load i64, ptr %in5, align 838; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v6 = load i64, ptr %in6, align 839; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v7 = load i64, ptr %in7, align 840;41; AVX1-LABEL: 'test'42; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 843; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 844; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 845; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 846; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 847; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 848; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 849; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 850; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 851; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 852; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 853; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 854; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 855; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 856; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 857; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 858; AVX1: LV: Found an estimated cost of 40 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 859; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 860; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 861; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 862; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 863; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v5 = load i64, ptr %in5, align 864; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v6 = load i64, ptr %in6, align 865; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v7 = load i64, ptr %in7, align 866; AVX1: LV: Found an estimated cost of 88 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 867; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 868; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 869; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 870; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 871; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v5 = load i64, ptr %in5, align 872; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v6 = load i64, ptr %in6, align 873; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v7 = load i64, ptr %in7, align 874;75; AVX2-LABEL: 'test'76; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 877; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 878; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 879; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 880; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 881; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 882; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 883; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 884; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 885; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 886; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 887; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 888; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 889; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 890; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 891; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 892; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 893; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 894; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 895; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 896; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 897; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v5 = load i64, ptr %in5, align 898; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v6 = load i64, ptr %in6, align 899; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v7 = load i64, ptr %in7, align 8100; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 8101; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 8102; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 8103; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 8104; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 8105; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v5 = load i64, ptr %in5, align 8106; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v6 = load i64, ptr %in6, align 8107; AVX2: LV: Found an estimated cost of 9 for VF 4 For instruction: %v7 = load i64, ptr %in7, align 8108; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 8109; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 8110; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 8111; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 8112; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v4 = load i64, ptr %in4, align 8113; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v5 = load i64, ptr %in5, align 8114; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v6 = load i64, ptr %in6, align 8115; AVX2: LV: Found an estimated cost of 18 for VF 8 For instruction: %v7 = load i64, ptr %in7, align 8116; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v0 = load i64, ptr %in0, align 8117; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v1 = load i64, ptr %in1, align 8118; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v2 = load i64, ptr %in2, align 8119; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v3 = load i64, ptr %in3, align 8120; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v4 = load i64, ptr %in4, align 8121; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v5 = load i64, ptr %in5, align 8122; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v6 = load i64, ptr %in6, align 8123; AVX2: LV: Found an estimated cost of 36 for VF 16 For instruction: %v7 = load i64, ptr %in7, align 8124; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v0 = load i64, ptr %in0, align 8125; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v1 = load i64, ptr %in1, align 8126; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v2 = load i64, ptr %in2, align 8127; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v3 = load i64, ptr %in3, align 8128; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v4 = load i64, ptr %in4, align 8129; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v5 = load i64, ptr %in5, align 8130; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v6 = load i64, ptr %in6, align 8131; AVX2: LV: Found an estimated cost of 72 for VF 32 For instruction: %v7 = load i64, ptr %in7, align 8132;133; AVX512-LABEL: 'test'134; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 8135; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 8136; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 8137; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 8138; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 8139; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 8140; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 8141; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 8142; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i64, ptr %in0, align 8143; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i64, ptr %in1, align 8144; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i64, ptr %in2, align 8145; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i64, ptr %in3, align 8146; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i64, ptr %in4, align 8147; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v5 = load i64, ptr %in5, align 8148; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v6 = load i64, ptr %in6, align 8149; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v7 = load i64, ptr %in7, align 8150; AVX512: LV: Found an estimated cost of 14 for VF 2 For instruction: %v0 = load i64, ptr %in0, align 8151; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i64, ptr %in1, align 8152; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i64, ptr %in2, align 8153; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i64, ptr %in3, align 8154; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i64, ptr %in4, align 8155; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v5 = load i64, ptr %in5, align 8156; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v6 = load i64, ptr %in6, align 8157; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v7 = load i64, ptr %in7, align 8158; AVX512: LV: Found an estimated cost of 40 for VF 4 For instruction: %v0 = load i64, ptr %in0, align 8159; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i64, ptr %in1, align 8160; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i64, ptr %in2, align 8161; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i64, ptr %in3, align 8162; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i64, ptr %in4, align 8163; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v5 = load i64, ptr %in5, align 8164; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v6 = load i64, ptr %in6, align 8165; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v7 = load i64, ptr %in7, align 8166; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v0 = load i64, ptr %in0, align 8167; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v1 = load i64, ptr %in1, align 8168; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v2 = load i64, ptr %in2, align 8169; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v3 = load i64, ptr %in3, align 8170; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v4 = load i64, ptr %in4, align 8171; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v5 = load i64, ptr %in5, align 8172; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v6 = load i64, ptr %in6, align 8173; AVX512: LV: Found an estimated cost of 10 for VF 8 For instruction: %v7 = load i64, ptr %in7, align 8174; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v0 = load i64, ptr %in0, align 8175; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v1 = load i64, ptr %in1, align 8176; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v2 = load i64, ptr %in2, align 8177; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v3 = load i64, ptr %in3, align 8178; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v4 = load i64, ptr %in4, align 8179; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v5 = load i64, ptr %in5, align 8180; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v6 = load i64, ptr %in6, align 8181; AVX512: LV: Found an estimated cost of 20 for VF 16 For instruction: %v7 = load i64, ptr %in7, align 8182;183entry:184 br label %for.body185 186for.body:187 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]188 189 %iv.0 = add nuw nsw i64 %iv, 0190 %iv.1 = add nuw nsw i64 %iv, 1191 %iv.2 = add nuw nsw i64 %iv, 2192 %iv.3 = add nuw nsw i64 %iv, 3193 %iv.4 = add nuw nsw i64 %iv, 4194 %iv.5 = add nuw nsw i64 %iv, 5195 %iv.6 = add nuw nsw i64 %iv, 6196 %iv.7 = add nuw nsw i64 %iv, 7197 198 %in0 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.0199 %in1 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.1200 %in2 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.2201 %in3 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.3202 %in4 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.4203 %in5 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.5204 %in6 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.6205 %in7 = getelementptr inbounds [1024 x i64], ptr @A, i64 0, i64 %iv.7206 207 %v0 = load i64, ptr %in0208 %v1 = load i64, ptr %in1209 %v2 = load i64, ptr %in2210 %v3 = load i64, ptr %in3211 %v4 = load i64, ptr %in4212 %v5 = load i64, ptr %in5213 %v6 = load i64, ptr %in6214 %v7 = load i64, ptr %in7215 216 %reduce.add.0 = add i64 %v0, %v1217 %reduce.add.1 = add i64 %reduce.add.0, %v2218 %reduce.add.2 = add i64 %reduce.add.1, %v3219 %reduce.add.3 = add i64 %reduce.add.2, %v4220 %reduce.add.4 = add i64 %reduce.add.3, %v5221 %reduce.add.5 = add i64 %reduce.add.4, %v6222 %reduce.add.6 = add i64 %reduce.add.5, %v7223 224 %reduce.add.6.narrow = trunc i64 %reduce.add.6 to i8225 226 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0227 store i8 %reduce.add.6.narrow, ptr %out228 229 %iv.next = add nuw nsw i64 %iv.0, 8230 %cmp = icmp ult i64 %iv.next, 1024231 br i1 %cmp, label %for.body, label %for.cond.cleanup232 233for.cond.cleanup:234 ret void235}236