1; RUN: opt -mtriple aarch64-linux-gnu -mattr=+sve -loop-vectorize -dce -instcombine -S <%s | FileCheck %s
2
3define void @stride7_i32(i32* noalias nocapture %dst, i64 %n) {
4; CHECK-LABEL: @stride7_i32(
5; CHECK:      vector.body
6; CHECK:        %[[VEC_IND:.*]] = phi <vscale x 4 x i64> [ %{{.*}}, %vector.ph ], [ %{{.*}}, %vector.body ]
7; CHECK-NEXT:   %[[PTR_INDICES:.*]] = mul nuw nsw <vscale x 4 x i64> %[[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 7, i32 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
8; CHECK-NEXT:   %[[PTRS:.*]] = getelementptr inbounds i32, i32* %dst, <vscale x 4 x i64> %[[PTR_INDICES]]
9; CHECK-NEXT:   %[[GLOAD:.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0i32(<vscale x 4 x i32*> %[[PTRS]]
10; CHECK-NEXT:   %[[VALS:.*]] = add nsw <vscale x 4 x i32> %[[GLOAD]],
11; CHECK-NEXT:   call void @llvm.masked.scatter.nxv4i32.nxv4p0i32(<vscale x 4 x i32> %[[VALS]], <vscale x 4 x i32*> %[[PTRS]]
12entry:
13  br label %for.body
14
15for.body:                                         ; preds = %entry, %for.body
16  %i.05 = phi i64 [ %inc, %for.body ], [ 0, %entry ]
17  %mul = mul nuw nsw i64 %i.05, 7
18  %arrayidx = getelementptr inbounds i32, i32* %dst, i64 %mul
19  %0 = load i32, i32* %arrayidx, align 4
20  %add = add nsw i32 %0, 3
21  store i32 %add, i32* %arrayidx, align 4
22  %inc = add nuw nsw i64 %i.05, 1
23  %exitcond.not = icmp eq i64 %inc, %n
24  br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !0
25
26for.end:                                          ; preds = %for.end.loopexit, %entry
27  ret void
28}
29
30define void @stride7_f64(double* noalias nocapture %dst, i64 %n) {
31; CHECK-LABEL: @stride7_f64(
32; CHECK:      vector.body
33; CHECK:        %[[VEC_IND:.*]] = phi <vscale x 2 x i64> [ %{{.*}}, %vector.ph ], [ %{{.*}}, %vector.body ]
34; CHECK-NEXT:   %[[PTR_INDICES:.*]] = mul nuw nsw <vscale x 2 x i64> %[[VEC_IND]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 7, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
35; CHECK-NEXT:   %[[PTRS:.*]] = getelementptr inbounds double, double* %dst, <vscale x 2 x i64> %[[PTR_INDICES]]
36; CHECK-NEXT:   %[[GLOAD:.*]] = call <vscale x 2 x double> @llvm.masked.gather.nxv2f64.nxv2p0f64(<vscale x 2 x double*> %[[PTRS]],
37; CHECK-NEXT:   %[[VALS:.*]] = fadd <vscale x 2 x double> %[[GLOAD]],
38; CHECK-NEXT:  call void @llvm.masked.scatter.nxv2f64.nxv2p0f64(<vscale x 2 x double> %[[VALS]], <vscale x 2 x double*> %[[PTRS]],
39entry:
40  br label %for.body
41
42for.body:                                         ; preds = %entry, %for.body
43  %i.05 = phi i64 [ %inc, %for.body ], [ 0, %entry ]
44  %mul = mul nuw nsw i64 %i.05, 7
45  %arrayidx = getelementptr inbounds double, double* %dst, i64 %mul
46  %0 = load double, double* %arrayidx, align 8
47  %add = fadd double %0, 1.000000e+00
48  store double %add, double* %arrayidx, align 8
49  %inc = add nuw nsw i64 %i.05, 1
50  %exitcond.not = icmp eq i64 %inc, %n
51  br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !6
52
53for.end:                                          ; preds = %for.end.loopexit, %entry
54  ret void
55}
56
57
58define void @cond_stride7_f64(double* noalias nocapture %dst, i64* noalias nocapture readonly %cond, i64 %n) {
59; CHECK-LABEL: @cond_stride7_f64(
60; CHECK:      vector.body
61; CHECK:        %[[MASK:.*]] = icmp ne <vscale x 2 x i64>
62; CHECK:        %[[PTRS:.*]] = getelementptr inbounds double, double* %dst, <vscale x 2 x i64> %{{.*}}
63; CHECK-NEXT:   %[[GLOAD:.*]] = call <vscale x 2 x double> @llvm.masked.gather.nxv2f64.nxv2p0f64(<vscale x 2 x double*> %[[PTRS]], i32 8, <vscale x 2 x i1> %[[MASK]]
64; CHECK-NEXT:   %[[VALS:.*]] = fadd <vscale x 2 x double> %[[GLOAD]],
65; CHECK-NEXT:  call void @llvm.masked.scatter.nxv2f64.nxv2p0f64(<vscale x 2 x double> %[[VALS]], <vscale x 2 x double*> %[[PTRS]], i32 8, <vscale x 2 x i1> %[[MASK]])
66entry:
67  br label %for.body
68
69for.body:                                         ; preds = %entry, %for.inc
70  %i.07 = phi i64 [ %inc, %for.inc ], [ 0, %entry ]
71  %arrayidx = getelementptr inbounds i64, i64* %cond, i64 %i.07
72  %0 = load i64, i64* %arrayidx, align 8
73  %tobool.not = icmp eq i64 %0, 0
74  br i1 %tobool.not, label %for.inc, label %if.then
75
76if.then:                                          ; preds = %for.body
77  %mul = mul nsw i64 %i.07, 7
78  %arrayidx1 = getelementptr inbounds double, double* %dst, i64 %mul
79  %1 = load double, double* %arrayidx1, align 8
80  %add = fadd double %1, 1.000000e+00
81  store double %add, double* %arrayidx1, align 8
82  br label %for.inc
83
84for.inc:                                          ; preds = %for.body, %if.then
85  %inc = add nuw nsw i64 %i.07, 1
86  %exitcond.not = icmp eq i64 %inc, %n
87  br i1 %exitcond.not, label %for.end, label %for.body, !llvm.loop !6
88
89for.end:                                          ; preds = %for.end.loopexit, %entry
90  ret void
91}
92
93
94!0 = distinct !{!0, !1, !2, !3, !4, !5}
95!1 = !{!"llvm.loop.mustprogress"}
96!2 = !{!"llvm.loop.vectorize.width", i32 4}
97!3 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
98!4 = !{!"llvm.loop.interleave.count", i32 1}
99!5 = !{!"llvm.loop.vectorize.enable", i1 true}
100!6 = distinct !{!6, !1, !7, !3, !4, !5}
101!7 = !{!"llvm.loop.vectorize.width", i32 2}
102