1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve.fp -verify-machineinstrs -o - %s | FileCheck %s
4 define arm_aapcs_vfpcc <16 x i8> @test_vhsubq_u8(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr #0 {
5 ; CHECK-LABEL: test_vhsubq_u8:
6 ; CHECK: @ %bb.0: @ %entry
7 ; CHECK-NEXT: vhsub.u8 q0, q0, q1
10 %0 = tail call <16 x i8> @llvm.arm.mve.vhsub.v16i8(<16 x i8> %a, <16 x i8> %b, i32 1)
14 declare <16 x i8> @llvm.arm.mve.vhsub.v16i8(<16 x i8>, <16 x i8>, i32) #1
16 define arm_aapcs_vfpcc <8 x i16> @test_vhsubq_s16(<8 x i16> %a, <8 x i16> %b) local_unnamed_addr #0 {
17 ; CHECK-LABEL: test_vhsubq_s16:
18 ; CHECK: @ %bb.0: @ %entry
19 ; CHECK-NEXT: vhsub.s16 q0, q0, q1
22 %0 = tail call <8 x i16> @llvm.arm.mve.vhsub.v8i16(<8 x i16> %a, <8 x i16> %b, i32 0)
26 declare <8 x i16> @llvm.arm.mve.vhsub.v8i16(<8 x i16>, <8 x i16>, i32) #1
28 define arm_aapcs_vfpcc <4 x i32> @test_vhsubq_u32(<4 x i32> %a, <4 x i32> %b) local_unnamed_addr #0 {
29 ; CHECK-LABEL: test_vhsubq_u32:
30 ; CHECK: @ %bb.0: @ %entry
31 ; CHECK-NEXT: vhsub.u32 q0, q0, q1
34 %0 = tail call <4 x i32> @llvm.arm.mve.vhsub.v4i32(<4 x i32> %a, <4 x i32> %b, i32 1)
38 declare <4 x i32> @llvm.arm.mve.vhsub.v4i32(<4 x i32>, <4 x i32>, i32) #1
40 define arm_aapcs_vfpcc <16 x i8> @test_vhsubq_m_s8(<16 x i8> %inactive, <16 x i8> %a, <16 x i8> %b, i16 zeroext %p) local_unnamed_addr #0 {
41 ; CHECK-LABEL: test_vhsubq_m_s8:
42 ; CHECK: @ %bb.0: @ %entry
43 ; CHECK-NEXT: vmsr p0, r0
45 ; CHECK-NEXT: vhsubt.s8 q0, q1, q2
48 %0 = zext i16 %p to i32
49 %1 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %0)
50 %2 = tail call <16 x i8> @llvm.arm.mve.hsub.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 0, <16 x i1> %1, <16 x i8> %inactive)
54 declare <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32) #1
56 declare <16 x i8> @llvm.arm.mve.hsub.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>, <16 x i8>) #1
58 define arm_aapcs_vfpcc <8 x i16> @test_vhsubq_m_u16(<8 x i16> %inactive, <8 x i16> %a, <8 x i16> %b, i16 zeroext %p) local_unnamed_addr #0 {
59 ; CHECK-LABEL: test_vhsubq_m_u16:
60 ; CHECK: @ %bb.0: @ %entry
61 ; CHECK-NEXT: vmsr p0, r0
63 ; CHECK-NEXT: vhsubt.u16 q0, q1, q2
66 %0 = zext i16 %p to i32
67 %1 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
68 %2 = tail call <8 x i16> @llvm.arm.mve.hsub.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 1, <8 x i1> %1, <8 x i16> %inactive)
72 declare <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32) #1
74 declare <8 x i16> @llvm.arm.mve.hsub.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>, <8 x i16>) #1
76 define arm_aapcs_vfpcc <4 x i32> @test_vhsubq_m_s32(<4 x i32> %inactive, <4 x i32> %a, <4 x i32> %b, i16 zeroext %p) local_unnamed_addr #0 {
77 ; CHECK-LABEL: test_vhsubq_m_s32:
78 ; CHECK: @ %bb.0: @ %entry
79 ; CHECK-NEXT: vmsr p0, r0
81 ; CHECK-NEXT: vhsubt.s32 q0, q1, q2
84 %0 = zext i16 %p to i32
85 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
86 %2 = tail call <4 x i32> @llvm.arm.mve.hsub.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 0, <4 x i1> %1, <4 x i32> %inactive)
90 declare <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32) #1
92 declare <4 x i32> @llvm.arm.mve.hsub.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>, <4 x i32>) #1
94 define arm_aapcs_vfpcc <16 x i8> @test_vhsubq_n_u8(<16 x i8> %a, i8 zeroext %b) {
95 ; CHECK-LABEL: test_vhsubq_n_u8:
96 ; CHECK: @ %bb.0: @ %entry
97 ; CHECK-NEXT: vhsub.u8 q0, q0, r0
100 %.splatinsert = insertelement <16 x i8> undef, i8 %b, i32 0
101 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
102 %0 = call <16 x i8> @llvm.arm.mve.vhsub.v16i8(<16 x i8> %a, <16 x i8> %.splat, i32 1)
106 define arm_aapcs_vfpcc <8 x i16> @test_vhsubq_n_s16(<8 x i16> %a, i16 signext %b) {
107 ; CHECK-LABEL: test_vhsubq_n_s16:
108 ; CHECK: @ %bb.0: @ %entry
109 ; CHECK-NEXT: vhsub.s16 q0, q0, r0
112 %.splatinsert = insertelement <8 x i16> undef, i16 %b, i32 0
113 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
114 %0 = call <8 x i16> @llvm.arm.mve.vhsub.v8i16(<8 x i16> %a, <8 x i16> %.splat, i32 0)
118 define arm_aapcs_vfpcc <4 x i32> @test_vhsubq_n_u32(<4 x i32> %a, i32 %b) {
119 ; CHECK-LABEL: test_vhsubq_n_u32:
120 ; CHECK: @ %bb.0: @ %entry
121 ; CHECK-NEXT: vhsub.u32 q0, q0, r0
124 %.splatinsert = insertelement <4 x i32> undef, i32 %b, i32 0
125 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
126 %0 = call <4 x i32> @llvm.arm.mve.vhsub.v4i32(<4 x i32> %a, <4 x i32> %.splat, i32 1)
130 define arm_aapcs_vfpcc <16 x i8> @test_vhsubq_m_n_s8(<16 x i8> %inactive, <16 x i8> %a, i8 signext %b, i16 zeroext %p) {
131 ; CHECK-LABEL: test_vhsubq_m_n_s8:
132 ; CHECK: @ %bb.0: @ %entry
133 ; CHECK-NEXT: vmsr p0, r1
135 ; CHECK-NEXT: vhsubt.s8 q0, q1, r0
138 %.splatinsert = insertelement <16 x i8> undef, i8 %b, i32 0
139 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
140 %0 = zext i16 %p to i32
141 %1 = call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %0)
142 %2 = call <16 x i8> @llvm.arm.mve.hsub.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %.splat, i32 0, <16 x i1> %1, <16 x i8> %inactive)
146 define arm_aapcs_vfpcc <8 x i16> @test_vhsubq_m_n_u16(<8 x i16> %inactive, <8 x i16> %a, i16 zeroext %b, i16 zeroext %p) {
147 ; CHECK-LABEL: test_vhsubq_m_n_u16:
148 ; CHECK: @ %bb.0: @ %entry
149 ; CHECK-NEXT: vmsr p0, r1
151 ; CHECK-NEXT: vhsubt.u16 q0, q1, r0
154 %.splatinsert = insertelement <8 x i16> undef, i16 %b, i32 0
155 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
156 %0 = zext i16 %p to i32
157 %1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
158 %2 = call <8 x i16> @llvm.arm.mve.hsub.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %.splat, i32 1, <8 x i1> %1, <8 x i16> %inactive)
162 define arm_aapcs_vfpcc <4 x i32> @test_vhsubq_m_n_s32(<4 x i32> %inactive, <4 x i32> %a, i32 %b, i16 zeroext %p) {
163 ; CHECK-LABEL: test_vhsubq_m_n_s32:
164 ; CHECK: @ %bb.0: @ %entry
165 ; CHECK-NEXT: vmsr p0, r1
167 ; CHECK-NEXT: vhsubt.s32 q0, q1, r0
170 %.splatinsert = insertelement <4 x i32> undef, i32 %b, i32 0
171 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
172 %0 = zext i16 %p to i32
173 %1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
174 %2 = call <4 x i32> @llvm.arm.mve.hsub.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %.splat, i32 0, <4 x i1> %1, <4 x i32> %inactive)
178 define arm_aapcs_vfpcc <16 x i8> @test_vhsubq_x_n_u8(<16 x i8> %a, i8 zeroext %b, i16 zeroext %p) {
179 ; CHECK-LABEL: test_vhsubq_x_n_u8:
180 ; CHECK: @ %bb.0: @ %entry
181 ; CHECK-NEXT: vmsr p0, r1
183 ; CHECK-NEXT: vhsubt.u8 q0, q0, r0
186 %.splatinsert = insertelement <16 x i8> undef, i8 %b, i32 0
187 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
188 %0 = zext i16 %p to i32
189 %1 = call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %0)
190 %2 = call <16 x i8> @llvm.arm.mve.hsub.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %.splat, i32 1, <16 x i1> %1, <16 x i8> undef)
194 define arm_aapcs_vfpcc <8 x i16> @test_vhsubq_x_n_s16(<8 x i16> %a, i16 signext %b, i16 zeroext %p) {
195 ; CHECK-LABEL: test_vhsubq_x_n_s16:
196 ; CHECK: @ %bb.0: @ %entry
197 ; CHECK-NEXT: vmsr p0, r1
199 ; CHECK-NEXT: vhsubt.s16 q0, q0, r0
202 %.splatinsert = insertelement <8 x i16> undef, i16 %b, i32 0
203 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
204 %0 = zext i16 %p to i32
205 %1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
206 %2 = call <8 x i16> @llvm.arm.mve.hsub.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %.splat, i32 0, <8 x i1> %1, <8 x i16> undef)
210 define arm_aapcs_vfpcc <4 x i32> @test_vhsubq_x_n_u32(<4 x i32> %a, i32 %b, i16 zeroext %p) {
211 ; CHECK-LABEL: test_vhsubq_x_n_u32:
212 ; CHECK: @ %bb.0: @ %entry
213 ; CHECK-NEXT: vmsr p0, r1
215 ; CHECK-NEXT: vhsubt.u32 q0, q0, r0
218 %.splatinsert = insertelement <4 x i32> undef, i32 %b, i32 0
219 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
220 %0 = zext i16 %p to i32
221 %1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
222 %2 = call <4 x i32> @llvm.arm.mve.hsub.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %.splat, i32 1, <4 x i1> %1, <4 x i32> undef)