1; RUN: opt < %s -passes=loop-vectorize -S | FileCheck %s 2 3; CHECK: fadd 4; CHECK-NEXT: fadd 5; CHECK-NEXT: fadd 6; CHECK-NEXT: fadd 7; CHECK-NEXT: fadd 8; CHECK-NEXT: fadd 9; CHECK-NEXT: fadd 10; CHECK-NEXT: fadd 11; CHECK-NEXT: = 12; CHECK-NOT: fadd 13; CHECK-SAME: > 14 15target datalayout = "e-m:e-i64:64-n32:64" 16target triple = "powerpc64le-ibm-linux-gnu" 17 18define void @QLA_F3_r_veq_norm2_V(ptr noalias nocapture %r, ptr noalias nocapture readonly %a, i32 signext %n) #0 { 19entry: 20 %cmp24 = icmp sgt i32 %n, 0 21 br i1 %cmp24, label %for.cond1.preheader.preheader, label %for.end13 22 23for.cond1.preheader.preheader: ; preds = %entry 24 br label %for.cond1.preheader 25 26for.cond1.preheader: ; preds = %for.cond1.preheader.preheader, %for.cond1.preheader 27 %indvars.iv = phi i64 [ %indvars.iv.next, %for.cond1.preheader ], [ 0, %for.cond1.preheader.preheader ] 28 %sum.026 = phi double [ %add10.2, %for.cond1.preheader ], [ 0.000000e+00, %for.cond1.preheader.preheader ] 29 %arrayidx5.realp = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 0, i32 0 30 %arrayidx5.real = load float, ptr %arrayidx5.realp, align 8 31 %arrayidx5.imagp = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 0, i32 1 32 %arrayidx5.imag = load float, ptr %arrayidx5.imagp, align 8 33 %mul = fmul fast float %arrayidx5.real, %arrayidx5.real 34 %mul9 = fmul fast float %arrayidx5.imag, %arrayidx5.imag 35 %add = fadd fast float %mul9, %mul 36 %conv = fpext float %add to double 37 %add10 = fadd fast double %conv, %sum.026 38 %arrayidx5.realp.1 = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 1, i32 0 39 %arrayidx5.real.1 = load float, ptr %arrayidx5.realp.1, align 8 40 %arrayidx5.imagp.1 = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 1, i32 1 41 %arrayidx5.imag.1 = load float, ptr %arrayidx5.imagp.1, align 8 42 %mul.1 = fmul fast float %arrayidx5.real.1, %arrayidx5.real.1 43 %mul9.1 = fmul fast float %arrayidx5.imag.1, %arrayidx5.imag.1 44 %add.1 = fadd fast float %mul9.1, %mul.1 45 %conv.1 = fpext float %add.1 to double 46 %add10.1 = fadd fast double %conv.1, %add10 47 %arrayidx5.realp.2 = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 2, i32 0 48 %arrayidx5.real.2 = load float, ptr %arrayidx5.realp.2, align 8 49 %arrayidx5.imagp.2 = getelementptr inbounds [3 x { float, float }], ptr %a, i64 %indvars.iv, i64 2, i32 1 50 %arrayidx5.imag.2 = load float, ptr %arrayidx5.imagp.2, align 8 51 %mul.2 = fmul fast float %arrayidx5.real.2, %arrayidx5.real.2 52 %mul9.2 = fmul fast float %arrayidx5.imag.2, %arrayidx5.imag.2 53 %add.2 = fadd fast float %mul9.2, %mul.2 54 %conv.2 = fpext float %add.2 to double 55 %add10.2 = fadd fast double %conv.2, %add10.1 56 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1 57 %lftr.wideiv = trunc i64 %indvars.iv.next to i32 58 %exitcond = icmp eq i32 %lftr.wideiv, %n 59 br i1 %exitcond, label %for.cond.for.end13_crit_edge, label %for.cond1.preheader 60 61for.cond.for.end13_crit_edge: ; preds = %for.cond1.preheader 62 %add10.2.lcssa = phi double [ %add10.2, %for.cond1.preheader ] 63 %phitmp = fptrunc double %add10.2.lcssa to float 64 br label %for.end13 65 66for.end13: ; preds = %for.cond.for.end13_crit_edge, %entry 67 %sum.0.lcssa = phi float [ %phitmp, %for.cond.for.end13_crit_edge ], [ 0.000000e+00, %entry ] 68 store float %sum.0.lcssa, ptr %r, align 4 69 ret void 70} 71 72