Run DCE after a LoopFlatten test to reduce spurious output [nfc]
[llvm-project.git] / clang / test / CodeGen / aarch64-sve2-intrinsics / acle_sve2_rsubhnt.c
blob115b16c53e09170ce7844031d5337f14caa006bd
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
9 #include <arm_sve.h>
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
18 // CHECK-LABEL: @test_svrsubhnt_s16(
19 // CHECK-NEXT: entry:
20 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
21 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
23 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_s16u10__SVInt8_tu11__SVInt16_tS0_(
24 // CPP-CHECK-NEXT: entry:
25 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
26 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
28 svint8_t test_svrsubhnt_s16(svint8_t op1, svint16_t op2, svint16_t op3)
30 return SVE_ACLE_FUNC(svrsubhnt,_s16,,)(op1, op2, op3);
33 // CHECK-LABEL: @test_svrsubhnt_s32(
34 // CHECK-NEXT: entry:
35 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
36 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
38 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_s32u11__SVInt16_tu11__SVInt32_tS0_(
39 // CPP-CHECK-NEXT: entry:
40 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
41 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
43 svint16_t test_svrsubhnt_s32(svint16_t op1, svint32_t op2, svint32_t op3)
45 return SVE_ACLE_FUNC(svrsubhnt,_s32,,)(op1, op2, op3);
48 // CHECK-LABEL: @test_svrsubhnt_s64(
49 // CHECK-NEXT: entry:
50 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
51 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
53 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_s64u11__SVInt32_tu11__SVInt64_tS0_(
54 // CPP-CHECK-NEXT: entry:
55 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
56 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
58 svint32_t test_svrsubhnt_s64(svint32_t op1, svint64_t op2, svint64_t op3)
60 return SVE_ACLE_FUNC(svrsubhnt,_s64,,)(op1, op2, op3);
63 // CHECK-LABEL: @test_svrsubhnt_u16(
64 // CHECK-NEXT: entry:
65 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
66 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
68 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_u16u11__SVUint8_tu12__SVUint16_tS0_(
69 // CPP-CHECK-NEXT: entry:
70 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
71 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
73 svuint8_t test_svrsubhnt_u16(svuint8_t op1, svuint16_t op2, svuint16_t op3)
75 return SVE_ACLE_FUNC(svrsubhnt,_u16,,)(op1, op2, op3);
78 // CHECK-LABEL: @test_svrsubhnt_u32(
79 // CHECK-NEXT: entry:
80 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
81 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
83 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_u32u12__SVUint16_tu12__SVUint32_tS0_(
84 // CPP-CHECK-NEXT: entry:
85 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[OP3:%.*]])
86 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
88 svuint16_t test_svrsubhnt_u32(svuint16_t op1, svuint32_t op2, svuint32_t op3)
90 return SVE_ACLE_FUNC(svrsubhnt,_u32,,)(op1, op2, op3);
93 // CHECK-LABEL: @test_svrsubhnt_u64(
94 // CHECK-NEXT: entry:
95 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
96 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
98 // CPP-CHECK-LABEL: @_Z18test_svrsubhnt_u64u12__SVUint32_tu12__SVUint64_tS0_(
99 // CPP-CHECK-NEXT: entry:
100 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[OP3:%.*]])
101 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
103 svuint32_t test_svrsubhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
105 return SVE_ACLE_FUNC(svrsubhnt,_u64,,)(op1, op2, op3);
108 // CHECK-LABEL: @test_svrsubhnt_n_s16(
109 // CHECK-NEXT: entry:
110 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
111 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
112 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
113 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
115 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
116 // CPP-CHECK-NEXT: entry:
117 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
118 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
119 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
120 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
122 svint8_t test_svrsubhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
124 return SVE_ACLE_FUNC(svrsubhnt,_n_s16,,)(op1, op2, op3);
127 // CHECK-LABEL: @test_svrsubhnt_n_s32(
128 // CHECK-NEXT: entry:
129 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
130 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
131 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
132 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
134 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
135 // CPP-CHECK-NEXT: entry:
136 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
137 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
138 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
139 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
141 svint16_t test_svrsubhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
143 return SVE_ACLE_FUNC(svrsubhnt,_n_s32,,)(op1, op2, op3);
146 // CHECK-LABEL: @test_svrsubhnt_n_s64(
147 // CHECK-NEXT: entry:
148 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
149 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
150 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
151 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
153 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
154 // CPP-CHECK-NEXT: entry:
155 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
156 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
157 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
158 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
160 svint32_t test_svrsubhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
162 return SVE_ACLE_FUNC(svrsubhnt,_n_s64,,)(op1, op2, op3);
165 // CHECK-LABEL: @test_svrsubhnt_n_u16(
166 // CHECK-NEXT: entry:
167 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
168 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
169 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
170 // CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
172 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
173 // CPP-CHECK-NEXT: entry:
174 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i64 0
175 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
176 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[DOTSPLAT]])
177 // CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
179 svuint8_t test_svrsubhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
181 return SVE_ACLE_FUNC(svrsubhnt,_n_u16,,)(op1, op2, op3);
184 // CHECK-LABEL: @test_svrsubhnt_n_u32(
185 // CHECK-NEXT: entry:
186 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
187 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
188 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
189 // CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
191 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
192 // CPP-CHECK-NEXT: entry:
193 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i64 0
194 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
195 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[DOTSPLAT]])
196 // CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
198 svuint16_t test_svrsubhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
200 return SVE_ACLE_FUNC(svrsubhnt,_n_u32,,)(op1, op2, op3);
203 // CHECK-LABEL: @test_svrsubhnt_n_u64(
204 // CHECK-NEXT: entry:
205 // CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
206 // CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
207 // CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
208 // CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
210 // CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
211 // CPP-CHECK-NEXT: entry:
212 // CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i64 0
213 // CPP-CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
214 // CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[DOTSPLAT]])
215 // CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
217 svuint32_t test_svrsubhnt_n_u64(svuint32_t op1, svuint64_t op2, uint64_t op3)
219 return SVE_ACLE_FUNC(svrsubhnt,_n_u64,,)(op1, op2, op3);