1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2 // REQUIRES: aarch64-registered-target
3 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +bf16 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
4 // RUN: %clang_cc1 -triple aarch64 -target-feature +sve -target-feature +bf16 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
5 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +bf16 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
6 // RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64 -target-feature +sve -target-feature +bf16 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
7 // RUN: %clang_cc1 -fclang-abi-compat=latest -triple aarch64 -target-feature +sve -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
8 // RUN: %clang_cc1 -fclang-abi-compat=latest -triple aarch64 -target-feature +sme -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
12 #if defined __ARM_FEATURE_SME
13 #define MODE_ATTR __arm_streaming
19 #ifdef SVE_OVERLOADED_FORMS
20 // A simple used,unused... macro, long enough to represent any SVE builtin.
21 #define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
23 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
25 // CHECK-LABEL: @test_svst2_bf16(
27 // CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
28 // CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
29 // CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 0
30 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 1
31 // CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
32 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2.nxv8bf16(<vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x i1> [[TMP4]], ptr [[BASE:%.*]])
33 // CHECK-NEXT: ret void
35 // CPP-CHECK-LABEL: @_Z15test_svst2_bf16u10__SVBool_tPu6__bf1614svbfloat16x2_t(
36 // CPP-CHECK-NEXT: entry:
37 // CPP-CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
38 // CPP-CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
39 // CPP-CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 0
40 // CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 1
41 // CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
42 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2.nxv8bf16(<vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x i1> [[TMP4]], ptr [[BASE:%.*]])
43 // CPP-CHECK-NEXT: ret void
45 void test_svst2_bf16(svbool_t pg
, bfloat16_t
*base
, svbfloat16x2_t data
) MODE_ATTR
47 return SVE_ACLE_FUNC(svst2
,_bf16
,,)(pg
, base
, data
);
50 // CHECK-LABEL: @test_svst2_vnum_bf16(
52 // CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
53 // CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
54 // CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 0
55 // CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 1
56 // CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
57 // CHECK-NEXT: [[TMP5:%.*]] = getelementptr <vscale x 8 x bfloat>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
58 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st2.nxv8bf16(<vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x i1> [[TMP4]], ptr [[TMP5]])
59 // CHECK-NEXT: ret void
61 // CPP-CHECK-LABEL: @_Z20test_svst2_vnum_bf16u10__SVBool_tPu6__bf16l14svbfloat16x2_t(
62 // CPP-CHECK-NEXT: entry:
63 // CPP-CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
64 // CPP-CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
65 // CPP-CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 0
66 // CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], 1
67 // CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
68 // CPP-CHECK-NEXT: [[TMP5:%.*]] = getelementptr <vscale x 8 x bfloat>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
69 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st2.nxv8bf16(<vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x i1> [[TMP4]], ptr [[TMP5]])
70 // CPP-CHECK-NEXT: ret void
72 void test_svst2_vnum_bf16(svbool_t pg
, bfloat16_t
*base
, int64_t vnum
, svbfloat16x2_t data
) MODE_ATTR
74 return SVE_ACLE_FUNC(svst2_vnum
,_bf16
,,)(pg
, base
, vnum
, data
);