Home | History | Annotate | Download | only in X86
      1 ; RUN: llc -mtriple=x86_64-pc-linux-gnu -mattr=+sse4.2 < %s | FileCheck %s
      2 
      3 ; CHECK-LABEL: testss:
      4 ; CHECK: movss {{.*}}, %[[XMM0:xmm[0-9]+]]
      5 ; CHECK: xorps %[[XMM1:xmm[0-9]+]], %[[XMM1]]
      6 ; CHECK: roundss $9, %[[XMM0]], %[[XMM1]]
      7 
      8 define void @testss(float* nocapture %a, <4 x float>* nocapture %b, i32 %k) {
      9 entry:
     10   br label %for.body
     11 
     12 for.body:
     13   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
     14   %arrayidx = getelementptr inbounds float, float* %a, i64 %indvars.iv
     15   %v = load float, float* %arrayidx, align 4
     16   %floor = call float @floorf(float %v)
     17   %sub = fsub float %floor, %v
     18   %v1 = insertelement <4 x float> undef, float %sub, i32 0
     19   %br = shufflevector <4 x float> %v1, <4 x float> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
     20   store volatile <4 x float> %br, <4 x float>* %b, align 4
     21   %indvars.iv.next = add i64 %indvars.iv, 1
     22   %lftr.wideiv = trunc i64 %indvars.iv.next to i32
     23   %exitcond = icmp eq i32 %lftr.wideiv, %k
     24   br i1 %exitcond, label %for.end, label %for.body
     25 
     26 for.end:
     27   ret void
     28 }
     29 
     30 ; CHECK-LABEL: testsd:
     31 ; CHECK: movsd {{.*}}, %[[XMM0:xmm[0-9]+]]
     32 ; CHECK: xorps %[[XMM1:xmm[0-9]+]], %[[XMM1]]
     33 ; CHECK: roundsd $9, %[[XMM0]], %[[XMM1]]
     34 
     35 define void @testsd(double* nocapture %a, <2 x double>* nocapture %b, i32 %k) {
     36 entry:
     37   br label %for.body
     38 
     39 for.body:
     40   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
     41   %arrayidx = getelementptr inbounds double, double* %a, i64 %indvars.iv
     42   %v = load double, double* %arrayidx, align 4
     43   %floor = call double @floor(double %v)
     44   %sub = fsub double %floor, %v
     45   %v1 = insertelement <2 x double> undef, double %sub, i32 0
     46   %br = shufflevector <2 x double> %v1, <2 x double> undef, <2 x i32> <i32 0, i32 0>
     47   store volatile <2 x double> %br, <2 x double>* %b, align 4
     48   %indvars.iv.next = add i64 %indvars.iv, 1
     49   %lftr.wideiv = trunc i64 %indvars.iv.next to i32
     50   %exitcond = icmp eq i32 %lftr.wideiv, %k
     51   br i1 %exitcond, label %for.end, label %for.body
     52 
     53 for.end:
     54   ret void
     55 }
     56 
     57 declare float @floorf(float) nounwind readnone
     58 
     59 declare double @floor(double) nounwind readnone
     60 
     61