1; RUN: opt < %s -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -dce -instcombine -licm -S | FileCheck %s 2 3target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" 4target triple = "x86_64-apple-macosx10.8.0" 5 6@b = common global [2048 x i32] zeroinitializer, align 16 7@c = common global [2048 x i32] zeroinitializer, align 16 8@a = common global [2048 x i32] zeroinitializer, align 16 9 10; Select VF = 8; 11;CHECK: @example1 12;CHECK: load <4 x i32> 13;CHECK: add nsw <4 x i32> 14;CHECK: store <4 x i32> 15;CHECK: ret void 16define void @example1() nounwind uwtable ssp { 17 br label %1 18 19; <label>:1 ; preds = %1, %0 20 %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ] 21 %2 = getelementptr inbounds [2048 x i32]* @b, i64 0, i64 %indvars.iv 22 %3 = load i32* %2, align 4 23 %4 = getelementptr inbounds [2048 x i32]* @c, i64 0, i64 %indvars.iv 24 %5 = load i32* %4, align 4 25 %6 = add nsw i32 %5, %3 26 %7 = getelementptr inbounds [2048 x i32]* @a, i64 0, i64 %indvars.iv 27 store i32 %6, i32* %7, align 4 28 %indvars.iv.next = add i64 %indvars.iv, 1 29 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 30 %exitcond = icmp eq i32 %lftr.wideiv, 256 31 br i1 %exitcond, label %8, label %1 32 33; <label>:8 ; preds = %1 34 ret void 35} 36 37 38; Select VF=4 because sext <8 x i1> to <8 x i32> is expensive. 39;CHECK: @example10b 40;CHECK: load <4 x i16> 41;CHECK: sext <4 x i16> 42;CHECK: store <4 x i32> 43;CHECK: ret void 44define void @example10b(i16* noalias nocapture %sa, i16* noalias nocapture %sb, i16* noalias nocapture %sc, i32* noalias nocapture %ia, i32* noalias nocapture %ib, i32* noalias nocapture %ic) nounwind uwtable ssp { 45 br label %1 46 47; <label>:1 ; preds = %1, %0 48 %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ] 49 %2 = getelementptr inbounds i16* %sb, i64 %indvars.iv 50 %3 = load i16* %2, align 2 51 %4 = sext i16 %3 to i32 52 %5 = getelementptr inbounds i32* %ia, i64 %indvars.iv 53 store i32 %4, i32* %5, align 4 54 %indvars.iv.next = add i64 %indvars.iv, 1 55 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 56 %exitcond = icmp eq i32 %lftr.wideiv, 1024 57 br i1 %exitcond, label %6, label %1 58 59; <label>:6 ; preds = %1 60 ret void 61} 62 63