xref: /llvm-project/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll (revision acdc419c897f8a9414c7a00c8908ac32312afee2)
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt -S -passes=instcombine < %s | FileCheck %s
3
4target triple = "aarch64-unknown-linux-gnu"
5
6define <vscale x 16 x i8> @dup_insertelement_0(<vscale x 16 x i8> %v, i8 %s) #0 {
7; CHECK-LABEL: @dup_insertelement_0(
8; CHECK-NEXT:    [[INSERT:%.*]] = insertelement <vscale x 16 x i8> [[V:%.*]], i8 [[S:%.*]], i64 0
9; CHECK-NEXT:    ret <vscale x 16 x i8> [[INSERT]]
10;
11  %pg = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 1)
12  %insert = tail call <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8> %v, <vscale x 16 x i1> %pg, i8 %s)
13  ret <vscale x 16 x i8> %insert
14}
15
16define <vscale x 16 x i8> @dup_insertelement_1(<vscale x 16 x i8> %v, i8 %s) #0 {
17; CHECK-LABEL: @dup_insertelement_1(
18; CHECK-NEXT:    [[PG:%.*]] = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 2)
19; CHECK-NEXT:    [[INSERT:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8> [[V:%.*]], <vscale x 16 x i1> [[PG]], i8 [[S:%.*]])
20; CHECK-NEXT:    ret <vscale x 16 x i8> [[INSERT]]
21;
22  %pg = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 2)
23  %insert = tail call <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8> %v, <vscale x 16 x i1> %pg, i8 %s)
24  ret <vscale x 16 x i8> %insert
25}
26
27define <vscale x 16 x i8> @dup_insertelement_x(<vscale x 16 x i8> %v, i8 %s, <vscale x 16 x i1> %pg) #0 {
28; CHECK-LABEL: @dup_insertelement_x(
29; CHECK-NEXT:    [[INSERT:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8> [[V:%.*]], <vscale x 16 x i1> [[PG:%.*]], i8 [[S:%.*]])
30; CHECK-NEXT:    ret <vscale x 16 x i8> [[INSERT]]
31;
32  %insert = tail call <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8> %v, <vscale x 16 x i1> %pg, i8 %s)
33  ret <vscale x 16 x i8> %insert
34}
35
36define <vscale x 8 x i16> @dup_insertelement_0_convert(<vscale x 8 x i16> %v, i16 %s) #0 {
37; CHECK-LABEL: @dup_insertelement_0_convert(
38; CHECK-NEXT:    [[INSERT:%.*]] = insertelement <vscale x 8 x i16> [[V:%.*]], i16 [[S:%.*]], i64 0
39; CHECK-NEXT:    ret <vscale x 8 x i16> [[INSERT]]
40;
41  %pg = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 1)
42  %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg)
43  %2 = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> %1)
44  %insert = tail call <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16> %v, <vscale x 8 x i1> %2, i16 %s)
45  ret <vscale x 8 x i16> %insert
46}
47
48define <vscale x 8 x i16> @dupx_splat_convert(i16 %s) #0 {
49; CHECK-LABEL: @dupx_splat_convert(
50; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[S:%.*]], i64 0
51; CHECK-NEXT:    [[SPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
52; CHECK-NEXT:    ret <vscale x 8 x i16> [[SPLAT]]
53;
54  %splat = tail call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %s)
55  ret <vscale x 8 x i16> %splat
56}
57
58declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
59
60declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i1>, i8)
61declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, i16)
62
63declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32)
64declare <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32)
65
66declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1>)
67declare <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1>)
68
69attributes #0 = { "target-features"="+sve" }
70