189 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 i32, 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 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: %v1 = load i32, ptr %in1, align 418; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 419; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 420; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 421; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 422; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 423; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 424; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 425; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 426; SSE2: LV: Found an estimated cost of 35 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 427; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i32, ptr %in1, align 428; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i32, ptr %in2, align 429; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i32, ptr %in3, align 430; SSE2: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i32, ptr %in4, align 431; SSE2: LV: Found an estimated cost of 75 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 432; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i32, ptr %in1, align 433; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i32, ptr %in2, align 434; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i32, ptr %in3, align 435; SSE2: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i32, ptr %in4, align 436; SSE2: LV: Found an estimated cost of 150 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 437; SSE2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i32, ptr %in1, align 438; SSE2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i32, ptr %in2, align 439; SSE2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i32, ptr %in3, align 440; SSE2: LV: Found an estimated cost of 0 for VF 8 For instruction: %v4 = load i32, ptr %in4, align 441;42; AVX1-LABEL: 'test'43; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 444; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 445; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 446; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 447; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 448; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 449; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 450; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 451; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 452; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 453; AVX1: LV: Found an estimated cost of 23 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 454; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i32, ptr %in1, align 455; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i32, ptr %in2, align 456; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i32, ptr %in3, align 457; AVX1: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i32, ptr %in4, align 458; AVX1: LV: Found an estimated cost of 45 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 459; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i32, ptr %in1, align 460; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i32, ptr %in2, align 461; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i32, ptr %in3, align 462; AVX1: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i32, ptr %in4, align 463; AVX1: LV: Found an estimated cost of 95 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 464; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i32, ptr %in1, align 465; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i32, ptr %in2, align 466; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i32, ptr %in3, align 467; AVX1: LV: Found an estimated cost of 0 for VF 8 For instruction: %v4 = load i32, ptr %in4, align 468; AVX1: LV: Found an estimated cost of 190 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 469; AVX1: LV: Found an estimated cost of 0 for VF 16 For instruction: %v1 = load i32, ptr %in1, align 470; AVX1: LV: Found an estimated cost of 0 for VF 16 For instruction: %v2 = load i32, ptr %in2, align 471; AVX1: LV: Found an estimated cost of 0 for VF 16 For instruction: %v3 = load i32, ptr %in3, align 472; AVX1: LV: Found an estimated cost of 0 for VF 16 For instruction: %v4 = load i32, ptr %in4, align 473;74; AVX2-LABEL: 'test'75; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 476; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 477; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 478; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 479; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 480; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 481; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 482; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 483; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 484; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 485; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 486; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v1 = load i32, ptr %in1, align 487; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v2 = load i32, ptr %in2, align 488; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v3 = load i32, ptr %in3, align 489; AVX2: LV: Found an estimated cost of 4 for VF 2 For instruction: %v4 = load i32, ptr %in4, align 490; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 491; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v1 = load i32, ptr %in1, align 492; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v2 = load i32, ptr %in2, align 493; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v3 = load i32, ptr %in3, align 494; AVX2: LV: Found an estimated cost of 8 for VF 4 For instruction: %v4 = load i32, ptr %in4, align 495; AVX2: LV: Found an estimated cost of 17 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 496; AVX2: LV: Found an estimated cost of 17 for VF 8 For instruction: %v1 = load i32, ptr %in1, align 497; AVX2: LV: Found an estimated cost of 17 for VF 8 For instruction: %v2 = load i32, ptr %in2, align 498; AVX2: LV: Found an estimated cost of 17 for VF 8 For instruction: %v3 = load i32, ptr %in3, align 499; AVX2: LV: Found an estimated cost of 17 for VF 8 For instruction: %v4 = load i32, ptr %in4, align 4100; AVX2: LV: Found an estimated cost of 34 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4101; AVX2: LV: Found an estimated cost of 34 for VF 16 For instruction: %v1 = load i32, ptr %in1, align 4102; AVX2: LV: Found an estimated cost of 34 for VF 16 For instruction: %v2 = load i32, ptr %in2, align 4103; AVX2: LV: Found an estimated cost of 34 for VF 16 For instruction: %v3 = load i32, ptr %in3, align 4104; AVX2: LV: Found an estimated cost of 34 for VF 16 For instruction: %v4 = load i32, ptr %in4, align 4105; AVX2: LV: Found an estimated cost of 68 for VF 32 For instruction: %v0 = load i32, ptr %in0, align 4106; AVX2: LV: Found an estimated cost of 68 for VF 32 For instruction: %v1 = load i32, ptr %in1, align 4107; AVX2: LV: Found an estimated cost of 68 for VF 32 For instruction: %v2 = load i32, ptr %in2, align 4108; AVX2: LV: Found an estimated cost of 68 for VF 32 For instruction: %v3 = load i32, ptr %in3, align 4109; AVX2: LV: Found an estimated cost of 68 for VF 32 For instruction: %v4 = load i32, ptr %in4, align 4110;111; AVX512-LABEL: 'test'112; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4113; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 4114; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 4115; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 4116; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 4117; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v0 = load i32, ptr %in0, align 4118; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v1 = load i32, ptr %in1, align 4119; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v2 = load i32, ptr %in2, align 4120; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v3 = load i32, ptr %in3, align 4121; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: %v4 = load i32, ptr %in4, align 4122; AVX512: LV: Found an estimated cost of 6 for VF 2 For instruction: %v0 = load i32, ptr %in0, align 4123; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v1 = load i32, ptr %in1, align 4124; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v2 = load i32, ptr %in2, align 4125; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v3 = load i32, ptr %in3, align 4126; AVX512: LV: Found an estimated cost of 0 for VF 2 For instruction: %v4 = load i32, ptr %in4, align 4127; AVX512: LV: Found an estimated cost of 9 for VF 4 For instruction: %v0 = load i32, ptr %in0, align 4128; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v1 = load i32, ptr %in1, align 4129; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v2 = load i32, ptr %in2, align 4130; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v3 = load i32, ptr %in3, align 4131; AVX512: LV: Found an estimated cost of 0 for VF 4 For instruction: %v4 = load i32, ptr %in4, align 4132; AVX512: LV: Found an estimated cost of 18 for VF 8 For instruction: %v0 = load i32, ptr %in0, align 4133; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v1 = load i32, ptr %in1, align 4134; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v2 = load i32, ptr %in2, align 4135; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v3 = load i32, ptr %in3, align 4136; AVX512: LV: Found an estimated cost of 0 for VF 8 For instruction: %v4 = load i32, ptr %in4, align 4137; AVX512: LV: Found an estimated cost of 35 for VF 16 For instruction: %v0 = load i32, ptr %in0, align 4138; AVX512: LV: Found an estimated cost of 0 for VF 16 For instruction: %v1 = load i32, ptr %in1, align 4139; AVX512: LV: Found an estimated cost of 0 for VF 16 For instruction: %v2 = load i32, ptr %in2, align 4140; AVX512: LV: Found an estimated cost of 0 for VF 16 For instruction: %v3 = load i32, ptr %in3, align 4141; AVX512: LV: Found an estimated cost of 0 for VF 16 For instruction: %v4 = load i32, ptr %in4, align 4142; AVX512: LV: Found an estimated cost of 145 for VF 32 For instruction: %v0 = load i32, ptr %in0, align 4143; AVX512: LV: Found an estimated cost of 0 for VF 32 For instruction: %v1 = load i32, ptr %in1, align 4144; AVX512: LV: Found an estimated cost of 0 for VF 32 For instruction: %v2 = load i32, ptr %in2, align 4145; AVX512: LV: Found an estimated cost of 0 for VF 32 For instruction: %v3 = load i32, ptr %in3, align 4146; AVX512: LV: Found an estimated cost of 0 for VF 32 For instruction: %v4 = load i32, ptr %in4, align 4147;148entry:149 br label %for.body150 151for.body:152 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]153 154 %iv.0 = add nuw nsw i64 %iv, 0155 %iv.1 = add nuw nsw i64 %iv, 1156 %iv.2 = add nuw nsw i64 %iv, 2157 %iv.3 = add nuw nsw i64 %iv, 3158 %iv.4 = add nuw nsw i64 %iv, 4159 160 %in0 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.0161 %in1 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.1162 %in2 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.2163 %in3 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.3164 %in4 = getelementptr inbounds [1024 x i32], ptr @A, i64 0, i64 %iv.4165 166 %v0 = load i32, ptr %in0167 %v1 = load i32, ptr %in1168 %v2 = load i32, ptr %in2169 %v3 = load i32, ptr %in3170 %v4 = load i32, ptr %in4171 172 %reduce.add.0 = add i32 %v0, %v1173 %reduce.add.1 = add i32 %reduce.add.0, %v2174 %reduce.add.2 = add i32 %reduce.add.1, %v3175 %reduce.add.3 = add i32 %reduce.add.2, %v4176 177 %reduce.add.3.narrow = trunc i32 %reduce.add.3 to i8178 179 %out = getelementptr inbounds [1024 x i8], ptr @B, i64 0, i64 %iv.0180 store i8 %reduce.add.3.narrow, ptr %out181 182 %iv.next = add nuw nsw i64 %iv.0, 5183 %cmp = icmp ult i64 %iv.next, 1024184 br i1 %cmp, label %for.body, label %for.cond.cleanup185 186for.cond.cleanup:187 ret void188}189