97 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*store float %v6, ptr %out6"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 i8] zeroinitializer, align 12812@B = global [1024 x float] zeroinitializer, align 12813 14define void @test() {15; SSE2-LABEL: 'test'16; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 417; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 418; SSE2: LV: Found an estimated cost of 23 for VF 2 For instruction: store float %v6, ptr %out6, align 419; SSE2: LV: Found an estimated cost of 52 for VF 4 For instruction: store float %v6, ptr %out6, align 420; SSE2: LV: Found an estimated cost of 104 for VF 8 For instruction: store float %v6, ptr %out6, align 421;22; AVX1-LABEL: 'test'23; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 424; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 425; AVX1: LV: Found an estimated cost of 24 for VF 2 For instruction: store float %v6, ptr %out6, align 426; AVX1: LV: Found an estimated cost of 50 for VF 4 For instruction: store float %v6, ptr %out6, align 427; AVX1: LV: Found an estimated cost of 105 for VF 8 For instruction: store float %v6, ptr %out6, align 428; AVX1: LV: Found an estimated cost of 210 for VF 16 For instruction: store float %v6, ptr %out6, align 429;30; AVX2-LABEL: 'test'31; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 432; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 433; AVX2: LV: Found an estimated cost of 24 for VF 2 For instruction: store float %v6, ptr %out6, align 434; AVX2: LV: Found an estimated cost of 50 for VF 4 For instruction: store float %v6, ptr %out6, align 435; AVX2: LV: Found an estimated cost of 105 for VF 8 For instruction: store float %v6, ptr %out6, align 436; AVX2: LV: Found an estimated cost of 210 for VF 16 For instruction: store float %v6, ptr %out6, align 437;38; AVX512-LABEL: 'test'39; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 440; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store float %v6, ptr %out6, align 441; AVX512: LV: Found an estimated cost of 10 for VF 2 For instruction: store float %v6, ptr %out6, align 442; AVX512: LV: Found an estimated cost of 20 for VF 4 For instruction: store float %v6, ptr %out6, align 443; AVX512: LV: Found an estimated cost of 40 for VF 8 For instruction: store float %v6, ptr %out6, align 444; AVX512: LV: Found an estimated cost of 70 for VF 16 For instruction: store float %v6, ptr %out6, align 445; AVX512: LV: Found an estimated cost of 140 for VF 32 For instruction: store float %v6, ptr %out6, align 446;47entry:48 br label %for.body49 50for.body:51 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]52 53 %iv.0 = add nuw nsw i64 %iv, 054 %iv.1 = add nuw nsw i64 %iv, 155 %iv.2 = add nuw nsw i64 %iv, 256 %iv.3 = add nuw nsw i64 %iv, 357 %iv.4 = add nuw nsw i64 %iv, 458 %iv.5 = add nuw nsw i64 %iv, 559 %iv.6 = add nuw nsw i64 %iv, 660 61 %in = getelementptr inbounds [1024 x i8], ptr @A, i64 0, i64 %iv.062 %v.narrow = load i8, ptr %in63 64 %v = uitofp i8 %v.narrow to float65 66 %v0 = fadd float %v, 0.067 %v1 = fadd float %v, 1.068 %v2 = fadd float %v, 2.069 %v3 = fadd float %v, 3.070 %v4 = fadd float %v, 4.071 %v5 = fadd float %v, 5.072 %v6 = fadd float %v, 6.073 74 %out0 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.075 %out1 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.176 %out2 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.277 %out3 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.378 %out4 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.479 %out5 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.580 %out6 = getelementptr inbounds [1024 x float], ptr @B, i64 0, i64 %iv.681 82 store float %v0, ptr %out083 store float %v1, ptr %out184 store float %v2, ptr %out285 store float %v3, ptr %out386 store float %v4, ptr %out487 store float %v5, ptr %out588 store float %v6, ptr %out689 90 %iv.next = add nuw nsw i64 %iv.0, 791 %cmp = icmp ult i64 %iv.next, 102492 br i1 %cmp, label %for.body, label %for.cond.cleanup93 94for.cond.cleanup:95 ret void96}97