clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c

   1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
   2 // REQUIRES: aarch64-registered-target
   3
   4 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
   5 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
   6 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
   7 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
   8
   9 #include <arm_sve.h>
  10
  11 #ifdef SVE_OVERLOADED_FORMS
  12 // A simple used,unused... macro, long enough to represent any SVE builtin.
  13 #define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
  14 #else
  15 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
  16 #endif
  17
  18 // CHECK-LABEL: @test_svpmullt_pair_u8(
  19 // CHECK-NEXT:  entry:
  20 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
  21 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
  22 //
  23 // CPP-CHECK-LABEL: @_Z21test_svpmullt_pair_u8u11__SVUint8_tS_(
  24 // CPP-CHECK-NEXT:  entry:
  25 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
  26 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
  27 //
  28 svuint8_t test_svpmullt_pair_u8(svuint8_t op1, svuint8_t op2)
  29 {
  30   return SVE_ACLE_FUNC(svpmullt_pair,_u8,,)(op1, op2);
  31 }
  32
  33 // CHECK-LABEL: @test_svpmullt_pair_u32(
  34 // CHECK-NEXT:  entry:
  35 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
  36 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
  37 //
  38 // CPP-CHECK-LABEL: @_Z22test_svpmullt_pair_u32u12__SVUint32_tS_(
  39 // CPP-CHECK-NEXT:  entry:
  40 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
  41 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
  42 //
  43 svuint32_t test_svpmullt_pair_u32(svuint32_t op1, svuint32_t op2)
  44 {
  45   return SVE_ACLE_FUNC(svpmullt_pair,_u32,,)(op1, op2);
  46 }
  47
  48 // CHECK-LABEL: @test_svpmullt_pair_n_u8(
  49 // CHECK-NEXT:  entry:
  50 // CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
  51 // CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
  52 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
  53 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
  54 //
  55 // CPP-CHECK-LABEL: @_Z23test_svpmullt_pair_n_u8u11__SVUint8_th(
  56 // CPP-CHECK-NEXT:  entry:
  57 // CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
  58 // CPP-CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
  59 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
  60 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
  61 //
  62 svuint8_t test_svpmullt_pair_n_u8(svuint8_t op1, uint8_t op2)
  63 {
  64   return SVE_ACLE_FUNC(svpmullt_pair,_n_u8,,)(op1, op2);
  65 }
  66
  67 // CHECK-LABEL: @test_svpmullt_pair_n_u32(
  68 // CHECK-NEXT:  entry:
  69 // CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
  70 // CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
  71 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
  72 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
  73 //
  74 // CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u32u12__SVUint32_tj(
  75 // CPP-CHECK-NEXT:  entry:
  76 // CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
  77 // CPP-CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
  78 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
  79 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
  80 //
  81 svuint32_t test_svpmullt_pair_n_u32(svuint32_t op1, uint32_t op2)
  82 {
  83   return SVE_ACLE_FUNC(svpmullt_pair,_n_u32,,)(op1, op2);
  84 }
  85
  86 // CHECK-LABEL: @test_svpmullt_u16(
  87 // CHECK-NEXT:  entry:
  88 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
  89 // CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 16 x i8> [[TMP0]] to <vscale x 8 x i16>
  90 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
  91 //
  92 // CPP-CHECK-LABEL: @_Z17test_svpmullt_u16u11__SVUint8_tS_(
  93 // CPP-CHECK-NEXT:  entry:
  94 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
  95 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 16 x i8> [[TMP0]] to <vscale x 8 x i16>
  96 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
  97 //
  98 svuint16_t test_svpmullt_u16(svuint8_t op1, svuint8_t op2)
  99 {
 100   return SVE_ACLE_FUNC(svpmullt,_u16,,)(op1, op2);
 101 }
 102
 103 // CHECK-LABEL: @test_svpmullt_u64(
 104 // CHECK-NEXT:  entry:
 105 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
 106 // CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 4 x i32> [[TMP0]] to <vscale x 2 x i64>
 107 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 108 //
 109 // CPP-CHECK-LABEL: @_Z17test_svpmullt_u64u12__SVUint32_tS_(
 110 // CPP-CHECK-NEXT:  entry:
 111 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
 112 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 4 x i32> [[TMP0]] to <vscale x 2 x i64>
 113 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 114 //
 115 svuint64_t test_svpmullt_u64(svuint32_t op1, svuint32_t op2)
 116 {
 117   return SVE_ACLE_FUNC(svpmullt,_u64,,)(op1, op2);
 118 }
 119
 120 // CHECK-LABEL: @test_svpmullt_n_u16(
 121 // CHECK-NEXT:  entry:
 122 // CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
 123 // CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 124 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
 125 // CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 16 x i8> [[TMP0]] to <vscale x 8 x i16>
 126 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 127 //
 128 // CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u16u11__SVUint8_th(
 129 // CPP-CHECK-NEXT:  entry:
 130 // CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
 131 // CPP-CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 132 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
 133 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 16 x i8> [[TMP0]] to <vscale x 8 x i16>
 134 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 135 //
 136 svuint16_t test_svpmullt_n_u16(svuint8_t op1, uint8_t op2)
 137 {
 138   return SVE_ACLE_FUNC(svpmullt,_n_u16,,)(op1, op2);
 139 }
 140
 141 // CHECK-LABEL: @test_svpmullt_n_u64(
 142 // CHECK-NEXT:  entry:
 143 // CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
 144 // CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 145 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
 146 // CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 4 x i32> [[TMP0]] to <vscale x 2 x i64>
 147 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 148 //
 149 // CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u64u12__SVUint32_tj(
 150 // CPP-CHECK-NEXT:  entry:
 151 // CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
 152 // CPP-CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 153 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
 154 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = bitcast <vscale x 4 x i32> [[TMP0]] to <vscale x 2 x i64>
 155 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 156 //
 157 svuint64_t test_svpmullt_n_u64(svuint32_t op1, uint32_t op2)
 158 {
 159   return SVE_ACLE_FUNC(svpmullt,_n_u64,,)(op1, op2);
 160 }