clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.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 +sve -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
   5 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
   6 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
   7 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | 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_bfdot_f32(
  19 // CHECK-NEXT:  entry:
  20 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]])
  21 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  22 //
  23 // CPP-CHECK-LABEL: @_Z14test_bfdot_f32u13__SVFloat32_tu14__SVBfloat16_tS0_(
  24 // CPP-CHECK-NEXT:  entry:
  25 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]])
  26 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  27 //
  28 svfloat32_t test_bfdot_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t z) {
  29   return SVE_ACLE_FUNC(svbfdot, _f32, , )(x, y, z);
  30 }
  31
  32 // CHECK-LABEL: @test_bfdot_lane_0_f32(
  33 // CHECK-NEXT:  entry:
  34 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]], i32 0)
  35 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  36 //
  37 // CPP-CHECK-LABEL: @_Z21test_bfdot_lane_0_f32u13__SVFloat32_tu14__SVBfloat16_tS0_(
  38 // CPP-CHECK-NEXT:  entry:
  39 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]], i32 0)
  40 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  41 //
  42 svfloat32_t test_bfdot_lane_0_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t z) {
  43   return SVE_ACLE_FUNC(svbfdot_lane, _f32, , )(x, y, z, 0);
  44 }
  45
  46 // CHECK-LABEL: @test_bfdot_lane_3_f32(
  47 // CHECK-NEXT:  entry:
  48 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]], i32 3)
  49 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  50 //
  51 // CPP-CHECK-LABEL: @_Z21test_bfdot_lane_3_f32u13__SVFloat32_tu14__SVBfloat16_tS0_(
  52 // CPP-CHECK-NEXT:  entry:
  53 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[Z:%.*]], i32 3)
  54 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  55 //
  56 svfloat32_t test_bfdot_lane_3_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t z) {
  57   return SVE_ACLE_FUNC(svbfdot_lane, _f32, , )(x, y, z, 3);
  58 }
  59
  60 // CHECK-LABEL: @test_bfdot_n_f32(
  61 // CHECK-NEXT:  entry:
  62 // CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i64 0
  63 // CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
  64 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[DOTSPLAT]])
  65 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  66 //
  67 // CPP-CHECK-LABEL: @_Z16test_bfdot_n_f32u13__SVFloat32_tu14__SVBfloat16_tu6__bf16(
  68 // CPP-CHECK-NEXT:  entry:
  69 // CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i64 0
  70 // CPP-CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
  71 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[DOTSPLAT]])
  72 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
  73 //
  74 svfloat32_t test_bfdot_n_f32(svfloat32_t x, svbfloat16_t y, bfloat16_t z) {
  75   return SVE_ACLE_FUNC(svbfdot, _n_f32, , )(x, y, z);
  76 }