1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2 // REQUIRES: aarch64-registered-target
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
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
15 #define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
18 // CHECK-LABEL: @test_svqrdmlsh_s8(
20 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[OP3:%.*]])
21 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
23 // CPP-CHECK-LABEL: @_Z17test_svqrdmlsh_s8u10__SVInt8_tS_S_(
24 // CPP-CHECK-NEXT: entry:
25 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[OP3:%.*]])
26 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
28 svint8_t
test_svqrdmlsh_s8(svint8_t op1
, svint8_t op2
, svint8_t op3
)
30 return SVE_ACLE_FUNC(svqrdmlsh
,_s8
,,)(op1
, op2
, op3
);
33 // CHECK-LABEL: @test_svqrdmlsh_s16(
35 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
36 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
38 // CPP-CHECK-LABEL: @_Z18test_svqrdmlsh_s16u11__SVInt16_tS_S_(
39 // CPP-CHECK-NEXT: entry:
40 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
41 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
43 svint16_t
test_svqrdmlsh_s16(svint16_t op1
, svint16_t op2
, svint16_t op3
)
45 return SVE_ACLE_FUNC(svqrdmlsh
,_s16
,,)(op1
, op2
, op3
);
48 // CHECK-LABEL: @test_svqrdmlsh_s32(
50 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
51 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
53 // CPP-CHECK-LABEL: @_Z18test_svqrdmlsh_s32u11__SVInt32_tS_S_(
54 // CPP-CHECK-NEXT: entry:
55 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
56 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
58 svint32_t
test_svqrdmlsh_s32(svint32_t op1
, svint32_t op2
, svint32_t op3
)
60 return SVE_ACLE_FUNC(svqrdmlsh
,_s32
,,)(op1
, op2
, op3
);
63 // CHECK-LABEL: @test_svqrdmlsh_s64(
65 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
66 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
68 // CPP-CHECK-LABEL: @_Z18test_svqrdmlsh_s64u11__SVInt64_tS_S_(
69 // CPP-CHECK-NEXT: entry:
70 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
71 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
73 svint64_t
test_svqrdmlsh_s64(svint64_t op1
, svint64_t op2
, svint64_t op3
)
75 return SVE_ACLE_FUNC(svqrdmlsh
,_s64
,,)(op1
, op2
, op3
);
78 // CHECK-LABEL: @test_svqrdmlsh_n_s8(
80 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i64 0
81 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
82 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
83 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
85 // CPP-CHECK-LABEL: @_Z19test_svqrdmlsh_n_s8u10__SVInt8_tS_a(
86 // CPP-CHECK-NEXT: entry:
87 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i64 0
88 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
89 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[DOTSPLAT]])
90 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
92 svint8_t
test_svqrdmlsh_n_s8(svint8_t op1
, svint8_t op2
, int8_t op3
)
94 return SVE_ACLE_FUNC(svqrdmlsh
,_n_s8
,,)(op1
, op2
, op3
);
97 // CHECK-LABEL: @test_svqrdmlsh_n_s16(
99 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
100 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
101 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
102 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
104 // CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s16u11__SVInt16_tS_s(
105 // CPP-CHECK-NEXT: entry:
106 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
107 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
108 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
109 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
111 svint16_t
test_svqrdmlsh_n_s16(svint16_t op1
, svint16_t op2
, int16_t op3
)
113 return SVE_ACLE_FUNC(svqrdmlsh
,_n_s16
,,)(op1
, op2
, op3
);
116 // CHECK-LABEL: @test_svqrdmlsh_n_s32(
117 // CHECK-NEXT: entry:
118 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
119 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
120 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
121 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
123 // CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s32u11__SVInt32_tS_i(
124 // CPP-CHECK-NEXT: entry:
125 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
126 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
127 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
128 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
130 svint32_t
test_svqrdmlsh_n_s32(svint32_t op1
, svint32_t op2
, int32_t op3
)
132 return SVE_ACLE_FUNC(svqrdmlsh
,_n_s32
,,)(op1
, op2
, op3
);
135 // CHECK-LABEL: @test_svqrdmlsh_n_s64(
136 // CHECK-NEXT: entry:
137 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
138 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
139 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
140 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
142 // CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s64u11__SVInt64_tS_l(
143 // CPP-CHECK-NEXT: entry:
144 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
145 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
146 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
147 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
149 svint64_t
test_svqrdmlsh_n_s64(svint64_t op1
, svint64_t op2
, int64_t op3
)
151 return SVE_ACLE_FUNC(svqrdmlsh
,_n_s64
,,)(op1
, op2
, op3
);
154 // CHECK-LABEL: @test_svqrdmlsh_lane_s16(
155 // CHECK-NEXT: entry:
156 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.lane.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 0)
157 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
159 // CPP-CHECK-LABEL: @_Z23test_svqrdmlsh_lane_s16u11__SVInt16_tS_S_(
160 // CPP-CHECK-NEXT: entry:
161 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.lane.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 0)
162 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
164 svint16_t
test_svqrdmlsh_lane_s16(svint16_t op1
, svint16_t op2
, svint16_t op3
)
166 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s16
,,)(op1
, op2
, op3
, 0);
169 // CHECK-LABEL: @test_svqrdmlsh_lane_s16_1(
170 // CHECK-NEXT: entry:
171 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.lane.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 7)
172 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
174 // CPP-CHECK-LABEL: @_Z25test_svqrdmlsh_lane_s16_1u11__SVInt16_tS_S_(
175 // CPP-CHECK-NEXT: entry:
176 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.lane.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 7)
177 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
179 svint16_t
test_svqrdmlsh_lane_s16_1(svint16_t op1
, svint16_t op2
, svint16_t op3
)
181 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s16
,,)(op1
, op2
, op3
, 7);
184 // CHECK-LABEL: @test_svqrdmlsh_lane_s32(
185 // CHECK-NEXT: entry:
186 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.lane.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]], i32 0)
187 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
189 // CPP-CHECK-LABEL: @_Z23test_svqrdmlsh_lane_s32u11__SVInt32_tS_S_(
190 // CPP-CHECK-NEXT: entry:
191 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.lane.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]], i32 0)
192 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
194 svint32_t
test_svqrdmlsh_lane_s32(svint32_t op1
, svint32_t op2
, svint32_t op3
)
196 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s32
,,)(op1
, op2
, op3
, 0);
199 // CHECK-LABEL: @test_svqrdmlsh_lane_s32_1(
200 // CHECK-NEXT: entry:
201 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.lane.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]], i32 3)
202 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
204 // CPP-CHECK-LABEL: @_Z25test_svqrdmlsh_lane_s32_1u11__SVInt32_tS_S_(
205 // CPP-CHECK-NEXT: entry:
206 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.lane.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]], i32 3)
207 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
209 svint32_t
test_svqrdmlsh_lane_s32_1(svint32_t op1
, svint32_t op2
, svint32_t op3
)
211 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s32
,,)(op1
, op2
, op3
, 3);
214 // CHECK-LABEL: @test_svqrdmlsh_lane_s64(
215 // CHECK-NEXT: entry:
216 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.lane.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]], i32 0)
217 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
219 // CPP-CHECK-LABEL: @_Z23test_svqrdmlsh_lane_s64u11__SVInt64_tS_S_(
220 // CPP-CHECK-NEXT: entry:
221 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.lane.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]], i32 0)
222 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
224 svint64_t
test_svqrdmlsh_lane_s64(svint64_t op1
, svint64_t op2
, svint64_t op3
)
226 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s64
,,)(op1
, op2
, op3
, 0);
229 // CHECK-LABEL: @test_svqrdmlsh_lane_s64_1(
230 // CHECK-NEXT: entry:
231 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.lane.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]], i32 1)
232 // CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
234 // CPP-CHECK-LABEL: @_Z25test_svqrdmlsh_lane_s64_1u11__SVInt64_tS_S_(
235 // CPP-CHECK-NEXT: entry:
236 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.lane.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]], i32 1)
237 // CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
239 svint64_t
test_svqrdmlsh_lane_s64_1(svint64_t op1
, svint64_t op2
, svint64_t op3
)
241 return SVE_ACLE_FUNC(svqrdmlsh_lane
,_s64
,,)(op1
, op2
, op3
, 1);