1; RUN: opt < %s -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -dce -instcombine -S | FileCheck %s 2; RUN: opt < %s -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -force-vector-interleave=0 -dce -instcombine -S | FileCheck %s -check-prefix=UNROLL 3 4target 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" 5target triple = "x86_64-apple-macosx10.8.0" 6 7@b = common global [2048 x i32] zeroinitializer, align 16 8@c = common global [2048 x i32] zeroinitializer, align 16 9@a = common global [2048 x i32] zeroinitializer, align 16 10 11; Select VF = 8; 12;CHECK-LABEL: @example1( 13;CHECK: load <4 x i32> 14;CHECK: add nsw <4 x i32> 15;CHECK: store <4 x i32> 16;CHECK: ret void 17 18;UNROLL-LABEL: @example1( 19;UNROLL: load <4 x i32> 20;UNROLL: load <4 x i32> 21;UNROLL: add nsw <4 x i32> 22;UNROLL: add nsw <4 x i32> 23;UNROLL: store <4 x i32> 24;UNROLL: store <4 x i32> 25;UNROLL: ret void 26define void @example1() nounwind uwtable ssp { 27 br label %1 28 29; <label>:1 ; preds = %1, %0 30 %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ] 31 %2 = getelementptr inbounds [2048 x i32], [2048 x i32]* @b, i64 0, i64 %indvars.iv 32 %3 = load i32, i32* %2, align 4 33 %4 = getelementptr inbounds [2048 x i32], [2048 x i32]* @c, i64 0, i64 %indvars.iv 34 %5 = load i32, i32* %4, align 4 35 %6 = add nsw i32 %5, %3 36 %7 = getelementptr inbounds [2048 x i32], [2048 x i32]* @a, i64 0, i64 %indvars.iv 37 store i32 %6, i32* %7, align 4 38 %indvars.iv.next = add i64 %indvars.iv, 1 39 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 40 %exitcond = icmp eq i32 %lftr.wideiv, 256 41 br i1 %exitcond, label %8, label %1 42 43; <label>:8 ; preds = %1 44 ret void 45} 46 47; Select VF=4 because sext <8 x i1> to <8 x i32> is expensive. 48;CHECK-LABEL: @example10b( 49;CHECK: load <4 x i16> 50;CHECK: sext <4 x i16> 51;CHECK: store <4 x i32> 52;CHECK: ret void 53;UNROLL-LABEL: @example10b( 54;UNROLL: load <4 x i16> 55;UNROLL: load <4 x i16> 56;UNROLL: store <4 x i32> 57;UNROLL: store <4 x i32> 58;UNROLL: ret void 59define 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 { 60 br label %1 61 62; <label>:1 ; preds = %1, %0 63 %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ] 64 %2 = getelementptr inbounds i16, i16* %sb, i64 %indvars.iv 65 %3 = load i16, i16* %2, align 2 66 %4 = sext i16 %3 to i32 67 %5 = getelementptr inbounds i32, i32* %ia, i64 %indvars.iv 68 store i32 %4, i32* %5, align 4 69 %indvars.iv.next = add i64 %indvars.iv, 1 70 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 71 %exitcond = icmp eq i32 %lftr.wideiv, 1024 72 br i1 %exitcond, label %6, label %1 73 74; <label>:6 ; preds = %1 75 ret void 76} 77 78