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 <8 x half> @test_vfmaq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
5 ; CHECK-LABEL: test_vfmaq_f16:
6 ; CHECK: @ %bb.0: @ %entry
7 ; CHECK-NEXT: vfma.f16 q0, q1, q2
10 %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %c, <8 x half> %a)
14 define arm_aapcs_vfpcc <4 x float> @test_vfmaq_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
15 ; CHECK-LABEL: test_vfmaq_f32:
16 ; CHECK: @ %bb.0: @ %entry
17 ; CHECK-NEXT: vfma.f32 q0, q1, q2
20 %0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %b, <4 x float> %c, <4 x float> %a)
24 define arm_aapcs_vfpcc <8 x half> @test_vfmaq_n_f16(<8 x half> %a, <8 x half> %b, float %c.coerce) {
25 ; CHECK-LABEL: test_vfmaq_n_f16:
26 ; CHECK: @ %bb.0: @ %entry
27 ; CHECK-NEXT: vmov r0, s8
28 ; CHECK-NEXT: vfma.f16 q0, q1, r0
31 %0 = bitcast float %c.coerce to i32
32 %tmp.0.extract.trunc = trunc i32 %0 to i16
33 %1 = bitcast i16 %tmp.0.extract.trunc to half
34 %.splatinsert = insertelement <8 x half> undef, half %1, i32 0
35 %.splat = shufflevector <8 x half> %.splatinsert, <8 x half> undef, <8 x i32> zeroinitializer
36 %2 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %.splat, <8 x half> %a)
40 define arm_aapcs_vfpcc <4 x float> @test_vfmaq_n_f32(<4 x float> %a, <4 x float> %b, float %c) {
41 ; CHECK-LABEL: test_vfmaq_n_f32:
42 ; CHECK: @ %bb.0: @ %entry
43 ; CHECK-NEXT: vmov r0, s8
44 ; CHECK-NEXT: vfma.f32 q0, q1, r0
47 %.splatinsert = insertelement <4 x float> undef, float %c, i32 0
48 %.splat = shufflevector <4 x float> %.splatinsert, <4 x float> undef, <4 x i32> zeroinitializer
49 %0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %b, <4 x float> %.splat, <4 x float> %a)
53 define arm_aapcs_vfpcc <8 x half> @test_vfmasq_n_f16(<8 x half> %a, <8 x half> %b, float %c.coerce) {
54 ; CHECK-LABEL: test_vfmasq_n_f16:
55 ; CHECK: @ %bb.0: @ %entry
56 ; CHECK-NEXT: vmov r0, s8
57 ; CHECK-NEXT: vfmas.f16 q0, q1, r0
60 %0 = bitcast float %c.coerce to i32
61 %tmp.0.extract.trunc = trunc i32 %0 to i16
62 %1 = bitcast i16 %tmp.0.extract.trunc to half
63 %.splatinsert = insertelement <8 x half> undef, half %1, i32 0
64 %.splat = shufflevector <8 x half> %.splatinsert, <8 x half> undef, <8 x i32> zeroinitializer
65 %2 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %a, <8 x half> %b, <8 x half> %.splat)
69 define arm_aapcs_vfpcc <4 x float> @test_vfmasq_n_f32(<4 x float> %a, <4 x float> %b, float %c) {
70 ; CHECK-LABEL: test_vfmasq_n_f32:
71 ; CHECK: @ %bb.0: @ %entry
72 ; CHECK-NEXT: vmov r0, s8
73 ; CHECK-NEXT: vfmas.f32 q0, q1, r0
76 %.splatinsert = insertelement <4 x float> undef, float %c, i32 0
77 %.splat = shufflevector <4 x float> %.splatinsert, <4 x float> undef, <4 x i32> zeroinitializer
78 %0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %.splat)
82 define arm_aapcs_vfpcc <8 x half> @test_vfmsq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
83 ; CHECK-LABEL: test_vfmsq_f16:
84 ; CHECK: @ %bb.0: @ %entry
85 ; CHECK-NEXT: vfms.f16 q0, q2, q1
88 %0 = fneg <8 x half> %c
89 %1 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %0, <8 x half> %a)
93 define arm_aapcs_vfpcc <4 x float> @test_vfmsq_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
94 ; CHECK-LABEL: test_vfmsq_f32:
95 ; CHECK: @ %bb.0: @ %entry
96 ; CHECK-NEXT: vfms.f32 q0, q2, q1
99 %0 = fneg <4 x float> %c
100 %1 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %b, <4 x float> %0, <4 x float> %a)
104 define arm_aapcs_vfpcc <16 x i8> @test_vmlaq_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c) {
105 ; CHECK-LABEL: test_vmlaq_n_s8:
106 ; CHECK: @ %bb.0: @ %entry
107 ; CHECK-NEXT: vmla.i8 q0, q1, r0
110 %.splatinsert = insertelement <16 x i8> undef, i8 %c, i32 0
111 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
112 %0 = mul <16 x i8> %.splat, %b
113 %1 = add <16 x i8> %0, %a
117 define arm_aapcs_vfpcc <8 x i16> @test_vmlaq_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c) {
118 ; CHECK-LABEL: test_vmlaq_n_s16:
119 ; CHECK: @ %bb.0: @ %entry
120 ; CHECK-NEXT: vmla.i16 q0, q1, r0
123 %.splatinsert = insertelement <8 x i16> undef, i16 %c, i32 0
124 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
125 %0 = mul <8 x i16> %.splat, %b
126 %1 = add <8 x i16> %0, %a
130 define arm_aapcs_vfpcc <4 x i32> @test_vmlaq_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
131 ; CHECK-LABEL: test_vmlaq_n_s32:
132 ; CHECK: @ %bb.0: @ %entry
133 ; CHECK-NEXT: vmla.i32 q0, q1, r0
136 %.splatinsert = insertelement <4 x i32> undef, i32 %c, i32 0
137 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
138 %0 = mul <4 x i32> %.splat, %b
139 %1 = add <4 x i32> %0, %a
143 define arm_aapcs_vfpcc <16 x i8> @test_vmlaq_n_u8(<16 x i8> %a, <16 x i8> %b, i8 zeroext %c) {
144 ; CHECK-LABEL: test_vmlaq_n_u8:
145 ; CHECK: @ %bb.0: @ %entry
146 ; CHECK-NEXT: vmla.i8 q0, q1, r0
149 %.splatinsert = insertelement <16 x i8> undef, i8 %c, i32 0
150 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
151 %0 = mul <16 x i8> %.splat, %b
152 %1 = add <16 x i8> %0, %a
156 define arm_aapcs_vfpcc <8 x i16> @test_vmlaq_n_u16(<8 x i16> %a, <8 x i16> %b, i16 zeroext %c) {
157 ; CHECK-LABEL: test_vmlaq_n_u16:
158 ; CHECK: @ %bb.0: @ %entry
159 ; CHECK-NEXT: vmla.i16 q0, q1, r0
162 %.splatinsert = insertelement <8 x i16> undef, i16 %c, i32 0
163 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
164 %0 = mul <8 x i16> %.splat, %b
165 %1 = add <8 x i16> %0, %a
169 define arm_aapcs_vfpcc <4 x i32> @test_vmlaq_n_u32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
170 ; CHECK-LABEL: test_vmlaq_n_u32:
171 ; CHECK: @ %bb.0: @ %entry
172 ; CHECK-NEXT: vmla.i32 q0, q1, r0
175 %.splatinsert = insertelement <4 x i32> undef, i32 %c, i32 0
176 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
177 %0 = mul <4 x i32> %.splat, %b
178 %1 = add <4 x i32> %0, %a
182 define arm_aapcs_vfpcc <16 x i8> @test_vmlasq_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c) {
183 ; CHECK-LABEL: test_vmlasq_n_s8:
184 ; CHECK: @ %bb.0: @ %entry
185 ; CHECK-NEXT: vmlas.i8 q1, q0, r0
186 ; CHECK-NEXT: vmov q0, q1
189 %0 = mul <16 x i8> %b, %a
190 %.splatinsert = insertelement <16 x i8> undef, i8 %c, i32 0
191 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
192 %1 = add <16 x i8> %.splat, %0
196 define arm_aapcs_vfpcc <8 x i16> @test_vmlasq_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c) {
197 ; CHECK-LABEL: test_vmlasq_n_s16:
198 ; CHECK: @ %bb.0: @ %entry
199 ; CHECK-NEXT: vmlas.i16 q1, q0, r0
200 ; CHECK-NEXT: vmov q0, q1
203 %0 = mul <8 x i16> %b, %a
204 %.splatinsert = insertelement <8 x i16> undef, i16 %c, i32 0
205 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
206 %1 = add <8 x i16> %.splat, %0
210 define arm_aapcs_vfpcc <4 x i32> @test_vmlasq_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
211 ; CHECK-LABEL: test_vmlasq_n_s32:
212 ; CHECK: @ %bb.0: @ %entry
213 ; CHECK-NEXT: vmlas.i32 q1, q0, r0
214 ; CHECK-NEXT: vmov q0, q1
217 %0 = mul <4 x i32> %b, %a
218 %.splatinsert = insertelement <4 x i32> undef, i32 %c, i32 0
219 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
220 %1 = add <4 x i32> %.splat, %0
224 define arm_aapcs_vfpcc <16 x i8> @test_vmlasq_n_u8(<16 x i8> %a, <16 x i8> %b, i8 zeroext %c) {
225 ; CHECK-LABEL: test_vmlasq_n_u8:
226 ; CHECK: @ %bb.0: @ %entry
227 ; CHECK-NEXT: vmlas.i8 q1, q0, r0
228 ; CHECK-NEXT: vmov q0, q1
231 %0 = mul <16 x i8> %b, %a
232 %.splatinsert = insertelement <16 x i8> undef, i8 %c, i32 0
233 %.splat = shufflevector <16 x i8> %.splatinsert, <16 x i8> undef, <16 x i32> zeroinitializer
234 %1 = add <16 x i8> %.splat, %0
238 define arm_aapcs_vfpcc <8 x i16> @test_vmlasq_n_u16(<8 x i16> %a, <8 x i16> %b, i16 zeroext %c) {
239 ; CHECK-LABEL: test_vmlasq_n_u16:
240 ; CHECK: @ %bb.0: @ %entry
241 ; CHECK-NEXT: vmlas.i16 q1, q0, r0
242 ; CHECK-NEXT: vmov q0, q1
245 %0 = mul <8 x i16> %b, %a
246 %.splatinsert = insertelement <8 x i16> undef, i16 %c, i32 0
247 %.splat = shufflevector <8 x i16> %.splatinsert, <8 x i16> undef, <8 x i32> zeroinitializer
248 %1 = add <8 x i16> %.splat, %0
252 define arm_aapcs_vfpcc <4 x i32> @test_vmlasq_n_u32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
253 ; CHECK-LABEL: test_vmlasq_n_u32:
254 ; CHECK: @ %bb.0: @ %entry
255 ; CHECK-NEXT: vmlas.i32 q1, q0, r0
256 ; CHECK-NEXT: vmov q0, q1
259 %0 = mul <4 x i32> %b, %a
260 %.splatinsert = insertelement <4 x i32> undef, i32 %c, i32 0
261 %.splat = shufflevector <4 x i32> %.splatinsert, <4 x i32> undef, <4 x i32> zeroinitializer
262 %1 = add <4 x i32> %.splat, %0
266 define arm_aapcs_vfpcc <16 x i8> @test_vqdmlahq_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c) {
267 ; CHECK-LABEL: test_vqdmlahq_n_s8:
268 ; CHECK: @ %bb.0: @ %entry
269 ; CHECK-NEXT: vqdmlah.s8 q0, q1, r0
272 %0 = zext i8 %c to i32
273 %1 = tail call <16 x i8> @llvm.arm.mve.vqdmlah.v16i8(<16 x i8> %a, <16 x i8> %b, i32 %0)
277 define arm_aapcs_vfpcc <8 x i16> @test_vqdmlahq_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c) {
278 ; CHECK-LABEL: test_vqdmlahq_n_s16:
279 ; CHECK: @ %bb.0: @ %entry
280 ; CHECK-NEXT: vqdmlah.s16 q0, q1, r0
283 %0 = zext i16 %c to i32
284 %1 = tail call <8 x i16> @llvm.arm.mve.vqdmlah.v8i16(<8 x i16> %a, <8 x i16> %b, i32 %0)
288 define arm_aapcs_vfpcc <4 x i32> @test_vqdmlahq_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
289 ; CHECK-LABEL: test_vqdmlahq_n_s32:
290 ; CHECK: @ %bb.0: @ %entry
291 ; CHECK-NEXT: vqdmlah.s32 q0, q1, r0
294 %0 = tail call <4 x i32> @llvm.arm.mve.vqdmlah.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %c)
298 define arm_aapcs_vfpcc <16 x i8> @test_vqdmlashq_n_s8(<16 x i8> %m1, <16 x i8> %m2, i8 signext %add) {
299 ; CHECK-LABEL: test_vqdmlashq_n_s8:
300 ; CHECK: @ %bb.0: @ %entry
301 ; CHECK-NEXT: vqdmlash.s8 q0, q1, r0
304 %0 = zext i8 %add to i32
305 %1 = tail call <16 x i8> @llvm.arm.mve.vqdmlash.v16i8(<16 x i8> %m1, <16 x i8> %m2, i32 %0)
309 define arm_aapcs_vfpcc <8 x i16> @test_vqdmlashq_n_s16(<8 x i16> %m1, <8 x i16> %m2, i16 signext %add) {
310 ; CHECK-LABEL: test_vqdmlashq_n_s16:
311 ; CHECK: @ %bb.0: @ %entry
312 ; CHECK-NEXT: vqdmlash.s16 q0, q1, r0
315 %0 = zext i16 %add to i32
316 %1 = tail call <8 x i16> @llvm.arm.mve.vqdmlash.v8i16(<8 x i16> %m1, <8 x i16> %m2, i32 %0)
320 define arm_aapcs_vfpcc <4 x i32> @test_vqdmlashq_n_s32(<4 x i32> %m1, <4 x i32> %m2, i32 %add) {
321 ; CHECK-LABEL: test_vqdmlashq_n_s32:
322 ; CHECK: @ %bb.0: @ %entry
323 ; CHECK-NEXT: vqdmlash.s32 q0, q1, r0
326 %0 = tail call <4 x i32> @llvm.arm.mve.vqdmlash.v4i32(<4 x i32> %m1, <4 x i32> %m2, i32 %add)
330 define arm_aapcs_vfpcc <16 x i8> @test_vqrdmlahq_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c) {
331 ; CHECK-LABEL: test_vqrdmlahq_n_s8:
332 ; CHECK: @ %bb.0: @ %entry
333 ; CHECK-NEXT: vqrdmlah.s8 q0, q1, r0
336 %0 = zext i8 %c to i32
337 %1 = tail call <16 x i8> @llvm.arm.mve.vqrdmlah.v16i8(<16 x i8> %a, <16 x i8> %b, i32 %0)
341 define arm_aapcs_vfpcc <8 x i16> @test_vqrdmlahq_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c) {
342 ; CHECK-LABEL: test_vqrdmlahq_n_s16:
343 ; CHECK: @ %bb.0: @ %entry
344 ; CHECK-NEXT: vqrdmlah.s16 q0, q1, r0
347 %0 = zext i16 %c to i32
348 %1 = tail call <8 x i16> @llvm.arm.mve.vqrdmlah.v8i16(<8 x i16> %a, <8 x i16> %b, i32 %0)
352 define arm_aapcs_vfpcc <4 x i32> @test_vqrdmlahq_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
353 ; CHECK-LABEL: test_vqrdmlahq_n_s32:
354 ; CHECK: @ %bb.0: @ %entry
355 ; CHECK-NEXT: vqrdmlah.s32 q0, q1, r0
358 %0 = tail call <4 x i32> @llvm.arm.mve.vqrdmlah.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %c)
362 define arm_aapcs_vfpcc <16 x i8> @test_vqrdmlashq_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c) {
363 ; CHECK-LABEL: test_vqrdmlashq_n_s8:
364 ; CHECK: @ %bb.0: @ %entry
365 ; CHECK-NEXT: vqrdmlash.s8 q0, q1, r0
368 %0 = zext i8 %c to i32
369 %1 = tail call <16 x i8> @llvm.arm.mve.vqrdmlash.v16i8(<16 x i8> %a, <16 x i8> %b, i32 %0)
373 define arm_aapcs_vfpcc <8 x i16> @test_vqrdmlashq_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c) {
374 ; CHECK-LABEL: test_vqrdmlashq_n_s16:
375 ; CHECK: @ %bb.0: @ %entry
376 ; CHECK-NEXT: vqrdmlash.s16 q0, q1, r0
379 %0 = zext i16 %c to i32
380 %1 = tail call <8 x i16> @llvm.arm.mve.vqrdmlash.v8i16(<8 x i16> %a, <8 x i16> %b, i32 %0)
384 define arm_aapcs_vfpcc <4 x i32> @test_vqrdmlashq_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c) {
385 ; CHECK-LABEL: test_vqrdmlashq_n_s32:
386 ; CHECK: @ %bb.0: @ %entry
387 ; CHECK-NEXT: vqrdmlash.s32 q0, q1, r0
390 %0 = tail call <4 x i32> @llvm.arm.mve.vqrdmlash.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %c)
394 define arm_aapcs_vfpcc <8 x half> @test_vfmaq_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
395 ; CHECK-LABEL: test_vfmaq_m_f16:
396 ; CHECK: @ %bb.0: @ %entry
397 ; CHECK-NEXT: vmsr p0, r0
399 ; CHECK-NEXT: vfmat.f16 q0, q1, q2
402 %0 = zext i16 %p to i32
403 %1 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
404 %2 = tail call <8 x half> @llvm.arm.mve.fma.predicated.v8f16.v8i1(<8 x half> %b, <8 x half> %c, <8 x half> %a, <8 x i1> %1)
408 define arm_aapcs_vfpcc <4 x float> @test_vfmaq_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
409 ; CHECK-LABEL: test_vfmaq_m_f32:
410 ; CHECK: @ %bb.0: @ %entry
411 ; CHECK-NEXT: vmsr p0, r0
413 ; CHECK-NEXT: vfmat.f32 q0, q1, q2
416 %0 = zext i16 %p to i32
417 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
418 %2 = tail call <4 x float> @llvm.arm.mve.fma.predicated.v4f32.v4i1(<4 x float> %b, <4 x float> %c, <4 x float> %a, <4 x i1> %1)
422 define arm_aapcs_vfpcc <8 x half> @test_vfmaq_m_n_f16(<8 x half> %a, <8 x half> %b, float %c.coerce, i16 zeroext %p) {
423 ; CHECK-LABEL: test_vfmaq_m_n_f16:
424 ; CHECK: @ %bb.0: @ %entry
425 ; CHECK-NEXT: vmov r1, s8
426 ; CHECK-NEXT: vmsr p0, r0
428 ; CHECK-NEXT: vfmat.f16 q0, q1, r1
431 %0 = bitcast float %c.coerce to i32
432 %tmp.0.extract.trunc = trunc i32 %0 to i16
433 %1 = bitcast i16 %tmp.0.extract.trunc to half
434 %.splatinsert = insertelement <8 x half> undef, half %1, i32 0
435 %.splat = shufflevector <8 x half> %.splatinsert, <8 x half> undef, <8 x i32> zeroinitializer
436 %2 = zext i16 %p to i32
437 %3 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %2)
438 %4 = tail call <8 x half> @llvm.arm.mve.fma.predicated.v8f16.v8i1(<8 x half> %b, <8 x half> %.splat, <8 x half> %a, <8 x i1> %3)
442 define arm_aapcs_vfpcc <4 x float> @test_vfmaq_m_n_f32(<4 x float> %a, <4 x float> %b, float %c, i16 zeroext %p) {
443 ; CHECK-LABEL: test_vfmaq_m_n_f32:
444 ; CHECK: @ %bb.0: @ %entry
445 ; CHECK-NEXT: vmov r1, s8
446 ; CHECK-NEXT: vmsr p0, r0
448 ; CHECK-NEXT: vfmat.f32 q0, q1, r1
451 %.splatinsert = insertelement <4 x float> undef, float %c, i32 0
452 %.splat = shufflevector <4 x float> %.splatinsert, <4 x float> undef, <4 x i32> zeroinitializer
453 %0 = zext i16 %p to i32
454 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
455 %2 = tail call <4 x float> @llvm.arm.mve.fma.predicated.v4f32.v4i1(<4 x float> %b, <4 x float> %.splat, <4 x float> %a, <4 x i1> %1)
459 define arm_aapcs_vfpcc <8 x half> @test_vfmasq_m_n_f16(<8 x half> %a, <8 x half> %b, float %c.coerce, i16 zeroext %p) {
460 ; CHECK-LABEL: test_vfmasq_m_n_f16:
461 ; CHECK: @ %bb.0: @ %entry
462 ; CHECK-NEXT: vmov r1, s8
463 ; CHECK-NEXT: vmsr p0, r0
465 ; CHECK-NEXT: vfmast.f16 q0, q1, r1
468 %0 = bitcast float %c.coerce to i32
469 %tmp.0.extract.trunc = trunc i32 %0 to i16
470 %1 = bitcast i16 %tmp.0.extract.trunc to half
471 %.splatinsert = insertelement <8 x half> undef, half %1, i32 0
472 %.splat = shufflevector <8 x half> %.splatinsert, <8 x half> undef, <8 x i32> zeroinitializer
473 %2 = zext i16 %p to i32
474 %3 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %2)
475 %4 = tail call <8 x half> @llvm.arm.mve.fma.predicated.v8f16.v8i1(<8 x half> %a, <8 x half> %b, <8 x half> %.splat, <8 x i1> %3)
479 define arm_aapcs_vfpcc <4 x float> @test_vfmasq_m_n_f32(<4 x float> %a, <4 x float> %b, float %c, i16 zeroext %p) {
480 ; CHECK-LABEL: test_vfmasq_m_n_f32:
481 ; CHECK: @ %bb.0: @ %entry
482 ; CHECK-NEXT: vmov r1, s8
483 ; CHECK-NEXT: vmsr p0, r0
485 ; CHECK-NEXT: vfmast.f32 q0, q1, r1
488 %.splatinsert = insertelement <4 x float> undef, float %c, i32 0
489 %.splat = shufflevector <4 x float> %.splatinsert, <4 x float> undef, <4 x i32> zeroinitializer
490 %0 = zext i16 %p to i32
491 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
492 %2 = tail call <4 x float> @llvm.arm.mve.fma.predicated.v4f32.v4i1(<4 x float> %a, <4 x float> %b, <4 x float> %.splat, <4 x i1> %1)
496 define arm_aapcs_vfpcc <8 x half> @test_vfmsq_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
497 ; CHECK-LABEL: test_vfmsq_m_f16:
498 ; CHECK: @ %bb.0: @ %entry
499 ; CHECK-NEXT: vmsr p0, r0
501 ; CHECK-NEXT: vfmst.f16 q0, q1, q2
504 %0 = fneg <8 x half> %c
505 %1 = zext i16 %p to i32
506 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
507 %3 = tail call <8 x half> @llvm.arm.mve.fma.predicated.v8f16.v8i1(<8 x half> %b, <8 x half> %0, <8 x half> %a, <8 x i1> %2)
511 define arm_aapcs_vfpcc <4 x float> @test_vfmsq_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
512 ; CHECK-LABEL: test_vfmsq_m_f32:
513 ; CHECK: @ %bb.0: @ %entry
514 ; CHECK-NEXT: vmsr p0, r0
516 ; CHECK-NEXT: vfmst.f32 q0, q1, q2
519 %0 = fneg <4 x float> %c
520 %1 = zext i16 %p to i32
521 %2 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %1)
522 %3 = tail call <4 x float> @llvm.arm.mve.fma.predicated.v4f32.v4i1(<4 x float> %b, <4 x float> %0, <4 x float> %a, <4 x i1> %2)
526 define arm_aapcs_vfpcc <16 x i8> @test_vmlaq_m_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c, i16 zeroext %p) {
527 ; CHECK-LABEL: test_vmlaq_m_n_s8:
528 ; CHECK: @ %bb.0: @ %entry
529 ; CHECK-NEXT: vmsr p0, r1
531 ; CHECK-NEXT: vmlat.i8 q0, q1, r0
534 %0 = zext i8 %c to i32
535 %1 = zext i16 %p to i32
536 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
537 %3 = tail call <16 x i8> @llvm.arm.mve.vmla.n.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
541 define arm_aapcs_vfpcc <8 x i16> @test_vmlaq_m_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c, i16 zeroext %p) {
542 ; CHECK-LABEL: test_vmlaq_m_n_s16:
543 ; CHECK: @ %bb.0: @ %entry
544 ; CHECK-NEXT: vmsr p0, r1
546 ; CHECK-NEXT: vmlat.i16 q0, q1, r0
549 %0 = zext i16 %c to i32
550 %1 = zext i16 %p to i32
551 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
552 %3 = tail call <8 x i16> @llvm.arm.mve.vmla.n.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
556 define arm_aapcs_vfpcc <4 x i32> @test_vmlaq_m_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
557 ; CHECK-LABEL: test_vmlaq_m_n_s32:
558 ; CHECK: @ %bb.0: @ %entry
559 ; CHECK-NEXT: vmsr p0, r1
561 ; CHECK-NEXT: vmlat.i32 q0, q1, r0
564 %0 = zext i16 %p to i32
565 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
566 %2 = tail call <4 x i32> @llvm.arm.mve.vmla.n.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
570 define arm_aapcs_vfpcc <16 x i8> @test_vmlaq_m_n_u8(<16 x i8> %a, <16 x i8> %b, i8 zeroext %c, i16 zeroext %p) {
571 ; CHECK-LABEL: test_vmlaq_m_n_u8:
572 ; CHECK: @ %bb.0: @ %entry
573 ; CHECK-NEXT: vmsr p0, r1
575 ; CHECK-NEXT: vmlat.i8 q0, q1, r0
578 %0 = zext i8 %c to i32
579 %1 = zext i16 %p to i32
580 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
581 %3 = tail call <16 x i8> @llvm.arm.mve.vmla.n.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
585 define arm_aapcs_vfpcc <8 x i16> @test_vmlaq_m_n_u16(<8 x i16> %a, <8 x i16> %b, i16 zeroext %c, i16 zeroext %p) {
586 ; CHECK-LABEL: test_vmlaq_m_n_u16:
587 ; CHECK: @ %bb.0: @ %entry
588 ; CHECK-NEXT: vmsr p0, r1
590 ; CHECK-NEXT: vmlat.i16 q0, q1, r0
593 %0 = zext i16 %c to i32
594 %1 = zext i16 %p to i32
595 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
596 %3 = tail call <8 x i16> @llvm.arm.mve.vmla.n.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
600 define arm_aapcs_vfpcc <4 x i32> @test_vmlaq_m_n_u32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
601 ; CHECK-LABEL: test_vmlaq_m_n_u32:
602 ; CHECK: @ %bb.0: @ %entry
603 ; CHECK-NEXT: vmsr p0, r1
605 ; CHECK-NEXT: vmlat.i32 q0, q1, r0
608 %0 = zext i16 %p to i32
609 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
610 %2 = tail call <4 x i32> @llvm.arm.mve.vmla.n.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
614 define arm_aapcs_vfpcc <16 x i8> @test_vmlasq_m_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c, i16 zeroext %p) {
615 ; CHECK-LABEL: test_vmlasq_m_n_s8:
616 ; CHECK: @ %bb.0: @ %entry
617 ; CHECK-NEXT: vmsr p0, r1
619 ; CHECK-NEXT: vmlast.i8 q0, q1, r0
622 %0 = zext i8 %c to i32
623 %1 = zext i16 %p to i32
624 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
625 %3 = tail call <16 x i8> @llvm.arm.mve.vmlas.n.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
629 define arm_aapcs_vfpcc <8 x i16> @test_vmlasq_m_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c, i16 zeroext %p) {
630 ; CHECK-LABEL: test_vmlasq_m_n_s16:
631 ; CHECK: @ %bb.0: @ %entry
632 ; CHECK-NEXT: vmsr p0, r1
634 ; CHECK-NEXT: vmlast.i16 q0, q1, r0
637 %0 = zext i16 %c to i32
638 %1 = zext i16 %p to i32
639 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
640 %3 = tail call <8 x i16> @llvm.arm.mve.vmlas.n.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
644 define arm_aapcs_vfpcc <4 x i32> @test_vmlasq_m_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
645 ; CHECK-LABEL: test_vmlasq_m_n_s32:
646 ; CHECK: @ %bb.0: @ %entry
647 ; CHECK-NEXT: vmsr p0, r1
649 ; CHECK-NEXT: vmlast.i32 q0, q1, r0
652 %0 = zext i16 %p to i32
653 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
654 %2 = tail call <4 x i32> @llvm.arm.mve.vmlas.n.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
658 define arm_aapcs_vfpcc <16 x i8> @test_vmlasq_m_n_u8(<16 x i8> %a, <16 x i8> %b, i8 zeroext %c, i16 zeroext %p) {
659 ; CHECK-LABEL: test_vmlasq_m_n_u8:
660 ; CHECK: @ %bb.0: @ %entry
661 ; CHECK-NEXT: vmsr p0, r1
663 ; CHECK-NEXT: vmlast.i8 q0, q1, r0
666 %0 = zext i8 %c to i32
667 %1 = zext i16 %p to i32
668 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
669 %3 = tail call <16 x i8> @llvm.arm.mve.vmlas.n.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
673 define arm_aapcs_vfpcc <8 x i16> @test_vmlasq_m_n_u16(<8 x i16> %a, <8 x i16> %b, i16 zeroext %c, i16 zeroext %p) {
674 ; CHECK-LABEL: test_vmlasq_m_n_u16:
675 ; CHECK: @ %bb.0: @ %entry
676 ; CHECK-NEXT: vmsr p0, r1
678 ; CHECK-NEXT: vmlast.i16 q0, q1, r0
681 %0 = zext i16 %c to i32
682 %1 = zext i16 %p to i32
683 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
684 %3 = tail call <8 x i16> @llvm.arm.mve.vmlas.n.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
688 define arm_aapcs_vfpcc <4 x i32> @test_vmlasq_m_n_u32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
689 ; CHECK-LABEL: test_vmlasq_m_n_u32:
690 ; CHECK: @ %bb.0: @ %entry
691 ; CHECK-NEXT: vmsr p0, r1
693 ; CHECK-NEXT: vmlast.i32 q0, q1, r0
696 %0 = zext i16 %p to i32
697 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
698 %2 = tail call <4 x i32> @llvm.arm.mve.vmlas.n.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
702 define arm_aapcs_vfpcc <16 x i8> @test_vqdmlahq_m_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c, i16 zeroext %p) {
703 ; CHECK-LABEL: test_vqdmlahq_m_n_s8:
704 ; CHECK: @ %bb.0: @ %entry
705 ; CHECK-NEXT: vmsr p0, r1
707 ; CHECK-NEXT: vqdmlaht.s8 q0, q1, r0
710 %0 = zext i8 %c to i32
711 %1 = zext i16 %p to i32
712 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
713 %3 = tail call <16 x i8> @llvm.arm.mve.vqdmlah.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
717 define arm_aapcs_vfpcc <8 x i16> @test_vqdmlahq_m_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c, i16 zeroext %p) {
718 ; CHECK-LABEL: test_vqdmlahq_m_n_s16:
719 ; CHECK: @ %bb.0: @ %entry
720 ; CHECK-NEXT: vmsr p0, r1
722 ; CHECK-NEXT: vqdmlaht.s16 q0, q1, r0
725 %0 = zext i16 %c to i32
726 %1 = zext i16 %p to i32
727 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
728 %3 = tail call <8 x i16> @llvm.arm.mve.vqdmlah.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
732 define arm_aapcs_vfpcc <4 x i32> @test_vqdmlahq_m_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
733 ; CHECK-LABEL: test_vqdmlahq_m_n_s32:
734 ; CHECK: @ %bb.0: @ %entry
735 ; CHECK-NEXT: vmsr p0, r1
737 ; CHECK-NEXT: vqdmlaht.s32 q0, q1, r0
740 %0 = zext i16 %p to i32
741 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
742 %2 = tail call <4 x i32> @llvm.arm.mve.vqdmlah.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
746 define arm_aapcs_vfpcc <16 x i8> @test_vqdmlashq_m_n_s8(<16 x i8> %m1, <16 x i8> %m2, i8 signext %add, i16 zeroext %p) {
747 ; CHECK-LABEL: test_vqdmlashq_m_n_s8:
748 ; CHECK: @ %bb.0: @ %entry
749 ; CHECK-NEXT: vmsr p0, r1
751 ; CHECK-NEXT: vqdmlasht.s8 q0, q1, r0
754 %0 = zext i8 %add to i32
755 %1 = zext i16 %p to i32
756 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
757 %3 = tail call <16 x i8> @llvm.arm.mve.vqdmlash.predicated.v16i8.v16i1(<16 x i8> %m1, <16 x i8> %m2, i32 %0, <16 x i1> %2)
761 define arm_aapcs_vfpcc <8 x i16> @test_vqdmlashq_m_n_s16(<8 x i16> %m1, <8 x i16> %m2, i16 signext %add, i16 zeroext %p) {
762 ; CHECK-LABEL: test_vqdmlashq_m_n_s16:
763 ; CHECK: @ %bb.0: @ %entry
764 ; CHECK-NEXT: vmsr p0, r1
766 ; CHECK-NEXT: vqdmlasht.s16 q0, q1, r0
769 %0 = zext i16 %add to i32
770 %1 = zext i16 %p to i32
771 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
772 %3 = tail call <8 x i16> @llvm.arm.mve.vqdmlash.predicated.v8i16.v8i1(<8 x i16> %m1, <8 x i16> %m2, i32 %0, <8 x i1> %2)
776 define arm_aapcs_vfpcc <4 x i32> @test_vqdmlashq_m_n_s32(<4 x i32> %m1, <4 x i32> %m2, i32 %add, i16 zeroext %p) {
777 ; CHECK-LABEL: test_vqdmlashq_m_n_s32:
778 ; CHECK: @ %bb.0: @ %entry
779 ; CHECK-NEXT: vmsr p0, r1
781 ; CHECK-NEXT: vqdmlasht.s32 q0, q1, r0
784 %0 = zext i16 %p to i32
785 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
786 %2 = tail call <4 x i32> @llvm.arm.mve.vqdmlash.predicated.v4i32.v4i1(<4 x i32> %m1, <4 x i32> %m2, i32 %add, <4 x i1> %1)
790 define arm_aapcs_vfpcc <16 x i8> @test_vqrdmlahq_m_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c, i16 zeroext %p) {
791 ; CHECK-LABEL: test_vqrdmlahq_m_n_s8:
792 ; CHECK: @ %bb.0: @ %entry
793 ; CHECK-NEXT: vmsr p0, r1
795 ; CHECK-NEXT: vqrdmlaht.s8 q0, q1, r0
798 %0 = zext i8 %c to i32
799 %1 = zext i16 %p to i32
800 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
801 %3 = tail call <16 x i8> @llvm.arm.mve.vqrdmlah.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
805 define arm_aapcs_vfpcc <8 x i16> @test_vqrdmlahq_m_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c, i16 zeroext %p) {
806 ; CHECK-LABEL: test_vqrdmlahq_m_n_s16:
807 ; CHECK: @ %bb.0: @ %entry
808 ; CHECK-NEXT: vmsr p0, r1
810 ; CHECK-NEXT: vqrdmlaht.s16 q0, q1, r0
813 %0 = zext i16 %c to i32
814 %1 = zext i16 %p to i32
815 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
816 %3 = tail call <8 x i16> @llvm.arm.mve.vqrdmlah.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
820 define arm_aapcs_vfpcc <4 x i32> @test_vqrdmlahq_m_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
821 ; CHECK-LABEL: test_vqrdmlahq_m_n_s32:
822 ; CHECK: @ %bb.0: @ %entry
823 ; CHECK-NEXT: vmsr p0, r1
825 ; CHECK-NEXT: vqrdmlaht.s32 q0, q1, r0
828 %0 = zext i16 %p to i32
829 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
830 %2 = tail call <4 x i32> @llvm.arm.mve.vqrdmlah.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
834 define arm_aapcs_vfpcc <16 x i8> @test_vqrdmlashq_m_n_s8(<16 x i8> %a, <16 x i8> %b, i8 signext %c, i16 zeroext %p) {
835 ; CHECK-LABEL: test_vqrdmlashq_m_n_s8:
836 ; CHECK: @ %bb.0: @ %entry
837 ; CHECK-NEXT: vmsr p0, r1
839 ; CHECK-NEXT: vqrdmlasht.s8 q0, q1, r0
842 %0 = zext i8 %c to i32
843 %1 = zext i16 %p to i32
844 %2 = tail call <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32 %1)
845 %3 = tail call <16 x i8> @llvm.arm.mve.vqrdmlash.predicated.v16i8.v16i1(<16 x i8> %a, <16 x i8> %b, i32 %0, <16 x i1> %2)
849 define arm_aapcs_vfpcc <8 x i16> @test_vqrdmlashq_m_n_s16(<8 x i16> %a, <8 x i16> %b, i16 signext %c, i16 zeroext %p) {
850 ; CHECK-LABEL: test_vqrdmlashq_m_n_s16:
851 ; CHECK: @ %bb.0: @ %entry
852 ; CHECK-NEXT: vmsr p0, r1
854 ; CHECK-NEXT: vqrdmlasht.s16 q0, q1, r0
857 %0 = zext i16 %c to i32
858 %1 = zext i16 %p to i32
859 %2 = tail call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %1)
860 %3 = tail call <8 x i16> @llvm.arm.mve.vqrdmlash.predicated.v8i16.v8i1(<8 x i16> %a, <8 x i16> %b, i32 %0, <8 x i1> %2)
864 define arm_aapcs_vfpcc <4 x i32> @test_vqrdmlashq_m_n_s32(<4 x i32> %a, <4 x i32> %b, i32 %c, i16 zeroext %p) {
865 ; CHECK-LABEL: test_vqrdmlashq_m_n_s32:
866 ; CHECK: @ %bb.0: @ %entry
867 ; CHECK-NEXT: vmsr p0, r1
869 ; CHECK-NEXT: vqrdmlasht.s32 q0, q1, r0
872 %0 = zext i16 %p to i32
873 %1 = tail call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
874 %2 = tail call <4 x i32> @llvm.arm.mve.vqrdmlash.predicated.v4i32.v4i1(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %1)
878 declare <16 x i1> @llvm.arm.mve.pred.i2v.v16i1(i32)
879 declare <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32)
880 declare <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32)
882 declare <8 x half> @llvm.fma.v8f16(<8 x half>, <8 x half>, <8 x half>)
883 declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
884 declare <8 x half> @llvm.arm.mve.fma.predicated.v8f16.v8i1(<8 x half>, <8 x half>, <8 x half>, <8 x i1>)
885 declare <4 x float> @llvm.arm.mve.fma.predicated.v4f32.v4i1(<4 x float>, <4 x float>, <4 x float>, <4 x i1>)
886 declare <16 x i8> @llvm.arm.mve.vmla.n.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
887 declare <8 x i16> @llvm.arm.mve.vmla.n.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
888 declare <4 x i32> @llvm.arm.mve.vmla.n.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)
889 declare <16 x i8> @llvm.arm.mve.vmlas.n.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
890 declare <8 x i16> @llvm.arm.mve.vmlas.n.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
891 declare <4 x i32> @llvm.arm.mve.vmlas.n.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)
892 declare <16 x i8> @llvm.arm.mve.vqdmlah.v16i8(<16 x i8>, <16 x i8>, i32)
893 declare <8 x i16> @llvm.arm.mve.vqdmlah.v8i16(<8 x i16>, <8 x i16>, i32)
894 declare <4 x i32> @llvm.arm.mve.vqdmlah.v4i32(<4 x i32>, <4 x i32>, i32)
895 declare <16 x i8> @llvm.arm.mve.vqdmlash.v16i8(<16 x i8>, <16 x i8>, i32)
896 declare <8 x i16> @llvm.arm.mve.vqdmlash.v8i16(<8 x i16>, <8 x i16>, i32)
897 declare <4 x i32> @llvm.arm.mve.vqdmlash.v4i32(<4 x i32>, <4 x i32>, i32)
898 declare <16 x i8> @llvm.arm.mve.vqrdmlah.v16i8(<16 x i8>, <16 x i8>, i32)
899 declare <8 x i16> @llvm.arm.mve.vqrdmlah.v8i16(<8 x i16>, <8 x i16>, i32)
900 declare <4 x i32> @llvm.arm.mve.vqrdmlah.v4i32(<4 x i32>, <4 x i32>, i32)
901 declare <16 x i8> @llvm.arm.mve.vqrdmlash.v16i8(<16 x i8>, <16 x i8>, i32)
902 declare <8 x i16> @llvm.arm.mve.vqrdmlash.v8i16(<8 x i16>, <8 x i16>, i32)
903 declare <4 x i32> @llvm.arm.mve.vqrdmlash.v4i32(<4 x i32>, <4 x i32>, i32)
904 declare <16 x i8> @llvm.arm.mve.vqdmlah.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
905 declare <8 x i16> @llvm.arm.mve.vqdmlah.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
906 declare <4 x i32> @llvm.arm.mve.vqdmlah.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)
907 declare <16 x i8> @llvm.arm.mve.vqdmlash.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
908 declare <8 x i16> @llvm.arm.mve.vqdmlash.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
909 declare <4 x i32> @llvm.arm.mve.vqdmlash.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)
910 declare <16 x i8> @llvm.arm.mve.vqrdmlah.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
911 declare <8 x i16> @llvm.arm.mve.vqrdmlah.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
912 declare <4 x i32> @llvm.arm.mve.vqrdmlah.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)
913 declare <16 x i8> @llvm.arm.mve.vqrdmlash.predicated.v16i8.v16i1(<16 x i8>, <16 x i8>, i32, <16 x i1>)
914 declare <8 x i16> @llvm.arm.mve.vqrdmlash.predicated.v8i16.v8i1(<8 x i16>, <8 x i16>, i32, <8 x i1>)
915 declare <4 x i32> @llvm.arm.mve.vqrdmlash.predicated.v4i32.v4i1(<4 x i32>, <4 x i32>, i32, <4 x i1>)