93 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*%v0 = load float, ptr %in0"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 float] 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 float, ptr %in0, align 417; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 418; SSE2: LV: Found an estimated cost of 18 for VF 2 For instruction: %v0 = load float, ptr %in0, align 419; SSE2: LV: Found an estimated cost of 42 for VF 4 For instruction: %v0 = load float, ptr %in0, align 420; SSE2: LV: Found an estimated cost of 84 for VF 8 For instruction: %v0 = load float, ptr %in0, align 421;22; AVX1-LABEL: 'test'23; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 424; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 425; AVX1: LV: Found an estimated cost of 18 for VF 2 For instruction: %v0 = load float, ptr %in0, align 426; AVX1: LV: Found an estimated cost of 42 for VF 4 For instruction: %v0 = load float, ptr %in0, align 427; AVX1: LV: Found an estimated cost of 90 for VF 8 For instruction: %v0 = load float, ptr %in0, align 428; AVX1: LV: Found an estimated cost of 180 for VF 16 For instruction: %v0 = load float, ptr %in0, align 429;30; AVX2-LABEL: 'test'31; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 432; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 433; AVX2: LV: Found an estimated cost of 8 for VF 2 For instruction: %v0 = load float, ptr %in0, align 434; AVX2: LV: Found an estimated cost of 18 for VF 4 For instruction: %v0 = load float, ptr %in0, align 435; AVX2: LV: Found an estimated cost of 37 for VF 8 For instruction: %v0 = load float, ptr %in0, align 436; AVX2: LV: Found an estimated cost of 76 for VF 16 For instruction: %v0 = load float, ptr %in0, align 437; AVX2: LV: Found an estimated cost of 60 for VF 32 For instruction: %v0 = load float, ptr %in0, align 438;39; AVX512-LABEL: 'test'40; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 441; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load float, ptr %in0, align 442; AVX512: LV: Found an estimated cost of 7 for VF 2 For instruction: %v0 = load float, ptr %in0, align 443; AVX512: LV: Found an estimated cost of 11 for VF 4 For instruction: %v0 = load float, ptr %in0, align 444; AVX512: LV: Found an estimated cost of 21 for VF 8 For instruction: %v0 = load float, ptr %in0, align 445; AVX512: LV: Found an estimated cost of 51 for VF 16 For instruction: %v0 = load float, ptr %in0, align 446; AVX512: LV: Found an estimated cost of 210 for VF 32 For instruction: %v0 = load float, ptr %in0, align 447;48entry:49 br label %for.body50 51for.body:52 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]53 54 %iv.0 = add nuw nsw i64 %iv, 055 %iv.1 = add nuw nsw i64 %iv, 156 %iv.2 = add nuw nsw i64 %iv, 257 %iv.3 = add nuw nsw i64 %iv, 358 %iv.4 = add nuw nsw i64 %iv, 459 %iv.5 = add nuw nsw i64 %iv, 560 61 %in0 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.062 %in1 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.163 %in2 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.264 %in3 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.365 %in4 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.466 %in5 = getelementptr inbounds [1024 x float], ptr @A, i64 0, i64 %iv.567 68 %v0 = load float, ptr %in069 %v1 = load float, ptr %in170 %v2 = load float, ptr %in271 %v3 = load float, ptr %in372 %v4 = load float, ptr %in473 %v5 = load float, ptr %in574 75 %reduce.add.0 = fadd float %v0, %v176 %reduce.add.1 = fadd float %reduce.add.0, %v277 %reduce.add.2 = fadd float %reduce.add.1, %v378 %reduce.add.3 = fadd float %reduce.add.2, %v479 %reduce.add.4 = fadd float %reduce.add.3, %v580 81 %reduce.add.4.narrow = fptoui float %reduce.add.4 to i882 83 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.084 store i8 %reduce.add.4.narrow, ptr %out85 86 %iv.next = add nuw nsw i64 %iv.0, 687 %cmp = icmp ult i64 %iv.next, 102488 br i1 %cmp, label %for.body, label %for.cond.cleanup89 90for.cond.cleanup:91 ret void92}93