84 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 i32, 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 i32] 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 i32, ptr %in0, align 417; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 418; SSE2: LV: Found an estimated cost of 28 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 419; SSE2: LV: Found an estimated cost of 60 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 420; SSE2: LV: Found an estimated cost of 120 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 421;22; AVX1-LABEL: 'test'23; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 424; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 425; AVX1: LV: Found an estimated cost of 18 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 426; AVX1: LV: Found an estimated cost of 36 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 427; AVX1: LV: Found an estimated cost of 76 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 428; AVX1: LV: Found an estimated cost of 152 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 429;30; AVX2-LABEL: 'test'31; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 432; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 433; AVX2: LV: Found an estimated cost of 5 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 434; AVX2: LV: Found an estimated cost of 10 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 435; AVX2: LV: Found an estimated cost of 20 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 436; AVX2: LV: Found an estimated cost of 40 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 437;38; AVX512-LABEL: 'test'39; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 440; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 441; AVX512: LV: Found an estimated cost of 5 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 442; AVX512: LV: Found an estimated cost of 5 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 443; AVX512: LV: Found an estimated cost of 8 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 444; AVX512: LV: Found an estimated cost of 22 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 445; AVX512: LV: Found an estimated cost of 92 for VF 32 For instruction: %v0 = load i32, ptr %in0, 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 58 %in0 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.059 %in1 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.160 %in2 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.261 %in3 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.362 63 %v0 = load i32, ptr %in064 %v1 = load i32, ptr %in165 %v2 = load i32, ptr %in266 %v3 = load i32, ptr %in367 68 %reduce.add.0 = add i32 %v0, %v169 %reduce.add.1 = add i32 %reduce.add.0, %v270 %reduce.add.2 = add i32 %reduce.add.1, %v371 72 %reduce.add.2.narrow = trunc i32 %reduce.add.2 to i873 74 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.075 store i8 %reduce.add.2.narrow, ptr %out76 77 %iv.next = add nuw nsw i64 %iv.0, 478 %cmp = icmp ult i64 %iv.next, 102479 br i1 %cmp, label %for.body, label %for.cond.cleanup80 81for.cond.cleanup:82 ret void83}84