1 ; RUN: opt -loop-vectorize -dce -instcombine -mtriple aarch64-linux-gnu -mattr=+sve -scalable-vectorization=on < %s -S | FileCheck %s
4 target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
5 target triple = "aarch64-unknown-linux-gnu"
7 define void @cmpsel_i32(i32* noalias nocapture %a, i32* noalias nocapture readonly %b, i64 %n) {
8 ; CHECK-LABEL: @cmpsel_i32(
11 ; CHECK: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, <vscale x 4 x i32>* {{.*}}, align 4
12 ; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <vscale x 4 x i32> [[WIDE_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 0, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
13 ; CHECK-NEXT: [[TMP2:%.*]] = select <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 10, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
14 ; CHECK: store <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32>* {{.*}}, align 4
19 for.body: ; preds = %entry, %for.body
20 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
21 %arrayidx = getelementptr inbounds i32, i32* %b, i64 %indvars.iv
22 %0 = load i32, i32* %arrayidx, align 4
23 %tobool.not = icmp eq i32 %0, 0
24 %cond = select i1 %tobool.not, i32 2, i32 10
25 %arrayidx2 = getelementptr inbounds i32, i32* %a, i64 %indvars.iv
26 store i32 %cond, i32* %arrayidx2, align 4
27 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
28 %exitcond.not = icmp eq i64 %indvars.iv.next, %n
29 br i1 %exitcond.not, label %for.end.loopexit, label %for.body, !llvm.loop !0
31 for.end.loopexit: ; preds = %for.body
34 for.end: ; preds = %for.end.loopexit, %entry
38 define void @cmpsel_f32(float* noalias nocapture %a, float* noalias nocapture readonly %b, i64 %n) {
39 ; CHECK-LABEL: @cmpsel_f32(
42 ; CHECK: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, <vscale x 4 x float>* {{.*}}, align 4
43 ; CHECK-NEXT: [[TMP1:%.*]] = fcmp ogt <vscale x 4 x float> [[WIDE_LOAD]], shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i32 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
44 ; CHECK-NEXT: [[TMP2:%.*]] = select <vscale x 4 x i1> [[TMP1]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 1.000000e+01, i32 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 2.000000e+00, i32 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
45 ; CHECK: store <vscale x 4 x float> [[TMP2]], <vscale x 4 x float>* {{.*}}, align 4
50 for.body: ; preds = %entry, %for.body
51 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
52 %arrayidx = getelementptr inbounds float, float* %b, i64 %indvars.iv
53 %0 = load float, float* %arrayidx, align 4
54 %cmp1 = fcmp ogt float %0, 3.000000e+00
55 %conv = select i1 %cmp1, float 1.000000e+01, float 2.000000e+00
56 %arrayidx3 = getelementptr inbounds float, float* %a, i64 %indvars.iv
57 store float %conv, float* %arrayidx3, align 4
58 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
59 %exitcond.not = icmp eq i64 %indvars.iv.next, %n
60 br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
62 for.end: ; preds = %for.body, %entry
66 define void @fneg_f32(float* noalias nocapture %a, float* noalias nocapture readonly %b, i64 %n) {
67 ; CHECK-LABEL: @fneg_f32(
70 ; CHECK: [[WIDE_LOAD:%.*]] = load <vscale x 4 x float>, <vscale x 4 x float>* {{.*}}, align 4
71 ; CHECK-NEXT: [[TMP1:%.*]] = fneg <vscale x 4 x float> [[WIDE_LOAD]]
72 ; CHECK: store <vscale x 4 x float> [[TMP1]], <vscale x 4 x float>* {{.*}}, align 4
77 for.body: ; preds = %entry, %for.body
78 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
79 %arrayidx = getelementptr inbounds float, float* %b, i64 %indvars.iv
80 %0 = load float, float* %arrayidx, align 4
82 %arrayidx3 = getelementptr inbounds float, float* %a, i64 %indvars.iv
83 store float %fneg, float* %arrayidx3, align 4
84 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
85 %exitcond.not = icmp eq i64 %indvars.iv.next, %n
86 br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
88 for.end: ; preds = %for.body, %entry
92 !0 = distinct !{!0, !1, !2, !3, !4, !5}
93 !1 = !{!"llvm.loop.mustprogress"}
94 !2 = !{!"llvm.loop.vectorize.width", i32 4}
95 !3 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
96 !4 = !{!"llvm.loop.interleave.count", i32 1}
97 !5 = !{!"llvm.loop.vectorize.enable", i1 true}