1 target 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" 2 target triple = "x86_64-unknown-linux-gnu" 3 ; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s 4 ; RUN: opt < %s -basicaa -loop-unroll -unroll-threshold=45 -unroll-allow-partial -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-UNRL 5 ; The second check covers the use of alias analysis (with loop unrolling). 6 7 define void @test1(double* noalias %out, double* noalias %in1, double* noalias %in2) nounwind uwtable { 8 entry: 9 br label %for.body 10 ; CHECK: @test1 11 ; CHECK-UNRL: @test1 12 13 for.body: ; preds = %for.body, %entry 14 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] 15 %arrayidx = getelementptr inbounds double* %in1, i64 %indvars.iv 16 %0 = load double* %arrayidx, align 8 17 %arrayidx2 = getelementptr inbounds double* %in2, i64 %indvars.iv 18 %1 = load double* %arrayidx2, align 8 19 %mul = fmul double %0, %0 20 %mul3 = fmul double %0, %1 21 %add = fadd double %mul, %mul3 22 %add4 = fadd double %1, %1 23 %add5 = fadd double %add4, %0 24 %mul6 = fmul double %0, %add5 25 %add7 = fadd double %add, %mul6 26 %mul8 = fmul double %1, %1 27 %add9 = fadd double %0, %0 28 %add10 = fadd double %add9, %0 29 %mul11 = fmul double %mul8, %add10 30 %add12 = fadd double %add7, %mul11 31 %arrayidx14 = getelementptr inbounds double* %out, i64 %indvars.iv 32 store double %add12, double* %arrayidx14, align 8 33 %indvars.iv.next = add i64 %indvars.iv, 1 34 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 35 %exitcond = icmp eq i32 %lftr.wideiv, 10 36 br i1 %exitcond, label %for.end, label %for.body 37 ; CHECK: %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] 38 ; CHECK: %arrayidx = getelementptr inbounds double* %in1, i64 %indvars.iv 39 ; CHECK: %0 = load double* %arrayidx, align 8 40 ; CHECK: %arrayidx2 = getelementptr inbounds double* %in2, i64 %indvars.iv 41 ; CHECK: %1 = load double* %arrayidx2, align 8 42 ; CHECK: %mul = fmul double %0, %0 43 ; CHECK: %mul3 = fmul double %0, %1 44 ; CHECK: %add = fadd double %mul, %mul3 45 ; CHECK: %mul8 = fmul double %1, %1 46 ; CHECK: %add4.v.i1.1 = insertelement <2 x double> undef, double %1, i32 0 47 ; CHECK: %add4.v.i1.2 = insertelement <2 x double> %add4.v.i1.1, double %0, i32 1 48 ; CHECK: %add4 = fadd <2 x double> %add4.v.i1.2, %add4.v.i1.2 49 ; CHECK: %add5.v.i1.1 = insertelement <2 x double> undef, double %0, i32 0 50 ; CHECK: %add5.v.i1.2 = insertelement <2 x double> %add5.v.i1.1, double %0, i32 1 51 ; CHECK: %add5 = fadd <2 x double> %add4, %add5.v.i1.2 52 ; CHECK: %mul6.v.i0.2 = insertelement <2 x double> %add5.v.i1.1, double %mul8, i32 1 53 ; CHECK: %mul6 = fmul <2 x double> %mul6.v.i0.2, %add5 54 ; CHECK: %mul6.v.r1 = extractelement <2 x double> %mul6, i32 0 55 ; CHECK: %mul6.v.r2 = extractelement <2 x double> %mul6, i32 1 56 ; CHECK: %add7 = fadd double %add, %mul6.v.r1 57 ; CHECK: %add12 = fadd double %add7, %mul6.v.r2 58 ; CHECK: %arrayidx14 = getelementptr inbounds double* %out, i64 %indvars.iv 59 ; CHECK: store double %add12, double* %arrayidx14, align 8 60 ; CHECK: %indvars.iv.next = add i64 %indvars.iv, 1 61 ; CHECK: %lftr.wideiv = trunc i64 %indvars.iv.next to i32 62 ; CHECK: %exitcond = icmp eq i32 %lftr.wideiv, 10 63 ; CHECK: br i1 %exitcond, label %for.end, label %for.body 64 ; CHECK-UNRL: %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next.1, %for.body ] 65 ; CHECK-UNRL: %arrayidx = getelementptr inbounds double* %in1, i64 %indvars.iv 66 ; CHECK-UNRL: %0 = bitcast double* %arrayidx to <2 x double>* 67 ; CHECK-UNRL: %arrayidx2 = getelementptr inbounds double* %in2, i64 %indvars.iv 68 ; CHECK-UNRL: %1 = bitcast double* %arrayidx2 to <2 x double>* 69 ; CHECK-UNRL: %arrayidx14 = getelementptr inbounds double* %out, i64 %indvars.iv 70 ; CHECK-UNRL: %2 = load <2 x double>* %0, align 8 71 ; CHECK-UNRL: %3 = load <2 x double>* %1, align 8 72 ; CHECK-UNRL: %mul = fmul <2 x double> %2, %2 73 ; CHECK-UNRL: %mul3 = fmul <2 x double> %2, %3 74 ; CHECK-UNRL: %add = fadd <2 x double> %mul, %mul3 75 ; CHECK-UNRL: %add4 = fadd <2 x double> %3, %3 76 ; CHECK-UNRL: %add5 = fadd <2 x double> %add4, %2 77 ; CHECK-UNRL: %mul6 = fmul <2 x double> %2, %add5 78 ; CHECK-UNRL: %add7 = fadd <2 x double> %add, %mul6 79 ; CHECK-UNRL: %mul8 = fmul <2 x double> %3, %3 80 ; CHECK-UNRL: %add9 = fadd <2 x double> %2, %2 81 ; CHECK-UNRL: %add10 = fadd <2 x double> %add9, %2 82 ; CHECK-UNRL: %mul11 = fmul <2 x double> %mul8, %add10 83 ; CHECK-UNRL: %add12 = fadd <2 x double> %add7, %mul11 84 ; CHECK-UNRL: %4 = bitcast double* %arrayidx14 to <2 x double>* 85 ; CHECK-UNRL: store <2 x double> %add12, <2 x double>* %4, align 8 86 ; CHECK-UNRL: %indvars.iv.next.1 = add i64 %indvars.iv, 2 87 ; CHECK-UNRL: %lftr.wideiv.1 = trunc i64 %indvars.iv.next.1 to i32 88 ; CHECK-UNRL: %exitcond.1 = icmp eq i32 %lftr.wideiv.1, 10 89 ; CHECK-UNRL: br i1 %exitcond.1, label %for.end, label %for.body 90 91 for.end: ; preds = %for.body 92 ret void 93 } 94
