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
26 // CHECK-LABEL: @test_svst4_bf16(
28 // CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
29 // CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
30 // CHECK-NEXT: [[TMP2:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], <vscale x 8 x bfloat> [[DATA_COERCE2:%.*]], 2
31 // CHECK-NEXT: [[TMP3:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], <vscale x 8 x bfloat> [[DATA_COERCE3:%.*]], 3
32 // CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 0
33 // CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 1
34 // CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 2
35 // CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 3
36 // CHECK-NEXT: [[TMP8:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
37 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4.nxv8bf16(<vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], <vscale x 8 x i1> [[TMP8]], ptr [[BASE:%.*]])
38 // CHECK-NEXT: ret void
40 // CPP-CHECK-LABEL: @_Z15test_svst4_bf16u10__SVBool_tPu6__bf1614svbfloat16x4_t(
41 // CPP-CHECK-NEXT: entry:
42 // CPP-CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
43 // CPP-CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
44 // CPP-CHECK-NEXT: [[TMP2:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], <vscale x 8 x bfloat> [[DATA_COERCE2:%.*]], 2
45 // CPP-CHECK-NEXT: [[TMP3:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], <vscale x 8 x bfloat> [[DATA_COERCE3:%.*]], 3
46 // CPP-CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 0
47 // CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 1
48 // CPP-CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 2
49 // CPP-CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 3
50 // CPP-CHECK-NEXT: [[TMP8:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
51 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4.nxv8bf16(<vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], <vscale x 8 x i1> [[TMP8]], ptr [[BASE:%.*]])
52 // CPP-CHECK-NEXT: ret void
54 void test_svst4_bf16(svbool_t pg
, bfloat16_t
*base
, svbfloat16x4_t data
) MODE_ATTR
56 return SVE_ACLE_FUNC(svst4
,_bf16
,,)(pg
, base
, data
);
59 // CHECK-LABEL: @test_svst4_vnum_bf16(
61 // CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
62 // CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
63 // CHECK-NEXT: [[TMP2:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], <vscale x 8 x bfloat> [[DATA_COERCE2:%.*]], 2
64 // CHECK-NEXT: [[TMP3:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], <vscale x 8 x bfloat> [[DATA_COERCE3:%.*]], 3
65 // CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 0
66 // CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 1
67 // CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 2
68 // CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 3
69 // CHECK-NEXT: [[TMP8:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
70 // CHECK-NEXT: [[TMP9:%.*]] = getelementptr <vscale x 8 x bfloat>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
71 // CHECK-NEXT: tail call void @llvm.aarch64.sve.st4.nxv8bf16(<vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], <vscale x 8 x i1> [[TMP8]], ptr [[TMP9]])
72 // CHECK-NEXT: ret void
74 // CPP-CHECK-LABEL: @_Z20test_svst4_vnum_bf16u10__SVBool_tPu6__bf16l14svbfloat16x4_t(
75 // CPP-CHECK-NEXT: entry:
76 // CPP-CHECK-NEXT: [[TMP0:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } poison, <vscale x 8 x bfloat> [[DATA_COERCE0:%.*]], 0
77 // CPP-CHECK-NEXT: [[TMP1:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], <vscale x 8 x bfloat> [[DATA_COERCE1:%.*]], 1
78 // CPP-CHECK-NEXT: [[TMP2:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP1]], <vscale x 8 x bfloat> [[DATA_COERCE2:%.*]], 2
79 // CPP-CHECK-NEXT: [[TMP3:%.*]] = insertvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], <vscale x 8 x bfloat> [[DATA_COERCE3:%.*]], 3
80 // CPP-CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 0
81 // CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 1
82 // CPP-CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 2
83 // CPP-CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP3]], 3
84 // CPP-CHECK-NEXT: [[TMP8:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
85 // CPP-CHECK-NEXT: [[TMP9:%.*]] = getelementptr <vscale x 8 x bfloat>, ptr [[BASE:%.*]], i64 [[VNUM:%.*]]
86 // CPP-CHECK-NEXT: tail call void @llvm.aarch64.sve.st4.nxv8bf16(<vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], <vscale x 8 x i1> [[TMP8]], ptr [[TMP9]])
87 // CPP-CHECK-NEXT: ret void
89 void test_svst4_vnum_bf16(svbool_t pg
, bfloat16_t
*base
, int64_t vnum
, svbfloat16x4_t data
) MODE_ATTR
91 return SVE_ACLE_FUNC(svst4_vnum
,_bf16
,,)(pg
, base
, vnum
, data
);