1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2; RUN: opt -S -passes=slp-vectorizer,slp-vectorizer -mcpu=bdver1 < %s | FileCheck %s 3 4target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" 5target triple = "x86_64-unknown-linux-gnu" 6 7@a = common local_unnamed_addr global [1 x i32] zeroinitializer, align 4 8@b = common local_unnamed_addr global [1 x i32] zeroinitializer, align 4 9 10define i32 @slp_schedule_bundle() local_unnamed_addr #0 { 11; CHECK-LABEL: @slp_schedule_bundle( 12; CHECK-NEXT: entry: 13; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr @b, align 4 14; CHECK-NEXT: [[TMP1:%.*]] = lshr <4 x i32> [[TMP0]], splat (i32 31) 15; CHECK-NEXT: [[TMP2:%.*]] = xor <4 x i32> [[TMP1]], splat (i32 1) 16; CHECK-NEXT: store <4 x i32> [[TMP2]], ptr @a, align 4 17; CHECK-NEXT: [[TMP3:%.*]] = load <2 x i32>, ptr getelementptr ([1 x i32], ptr @b, i64 4, i64 0), align 4 18; CHECK-NEXT: [[TMP4:%.*]] = lshr <2 x i32> [[TMP3]], splat (i32 31) 19; CHECK-NEXT: [[TMP5:%.*]] = xor <2 x i32> [[TMP4]], splat (i32 1) 20; CHECK-NEXT: store <2 x i32> [[TMP5]], ptr getelementptr ([1 x i32], ptr @a, i64 4, i64 0), align 4 21; CHECK-NEXT: ret i32 undef 22; 23entry: 24 %0 = load i32, ptr @b, align 4 25 %.lobit = lshr i32 %0, 31 26 %.lobit.not = xor i32 %.lobit, 1 27 store i32 %.lobit.not, ptr @a, align 4 28 %1 = load i32, ptr getelementptr inbounds ([1 x i32], ptr @b, i64 1, i64 0), align 4 29 %.lobit.1 = lshr i32 %1, 31 30 %.lobit.not.1 = xor i32 %.lobit.1, 1 31 store i32 %.lobit.not.1, ptr getelementptr inbounds ([1 x i32], ptr @a, i64 1, i64 0), align 4 32 %2 = load i32, ptr getelementptr ([1 x i32], ptr @b, i64 2, i64 0), align 4 33 %.lobit.2 = lshr i32 %2, 31 34 %.lobit.not.2 = xor i32 %.lobit.2, 1 35 store i32 %.lobit.not.2, ptr getelementptr ([1 x i32], ptr @a, i64 2, i64 0), align 4 36 %3 = load i32, ptr getelementptr ([1 x i32], ptr @b, i64 3, i64 0), align 4 37 %.lobit.3 = lshr i32 %3, 31 38 %.lobit.not.3 = xor i32 %.lobit.3, 1 39 store i32 %.lobit.not.3, ptr getelementptr ([1 x i32], ptr @a, i64 3, i64 0), align 4 40 %4 = load i32, ptr getelementptr ([1 x i32], ptr @b, i64 4, i64 0), align 4 41 %.lobit.4 = lshr i32 %4, 31 42 %.lobit.not.4 = xor i32 %.lobit.4, 1 43 store i32 %.lobit.not.4, ptr getelementptr ([1 x i32], ptr @a, i64 4, i64 0), align 4 44 %5 = load i32, ptr getelementptr ([1 x i32], ptr @b, i64 5, i64 0), align 4 45 %.lobit.5 = lshr i32 %5, 31 46 %.lobit.not.5 = xor i32 %.lobit.5, 1 47 store i32 %.lobit.not.5, ptr getelementptr ([1 x i32], ptr @a, i64 5, i64 0), align 4 48 ret i32 undef 49} 50