1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
8 define <vscale x 8 x i1> @facge_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
9 ; CHECK-LABEL: facge_h:
11 ; CHECK-NEXT: facge p0.h, p0/z, z0.h, z1.h
13 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.facge.nxv8f16(<vscale x 8 x i1> %pg,
14 <vscale x 8 x half> %a,
15 <vscale x 8 x half> %b)
16 ret <vscale x 8 x i1> %out
19 define <vscale x 4 x i1> @facge_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
20 ; CHECK-LABEL: facge_s:
22 ; CHECK-NEXT: facge p0.s, p0/z, z0.s, z1.s
24 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> %pg,
25 <vscale x 4 x float> %a,
26 <vscale x 4 x float> %b)
27 ret <vscale x 4 x i1> %out
30 define <vscale x 2 x i1> @facge_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
31 ; CHECK-LABEL: facge_d:
33 ; CHECK-NEXT: facge p0.d, p0/z, z0.d, z1.d
35 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> %pg,
36 <vscale x 2 x double> %a,
37 <vscale x 2 x double> %b)
38 ret <vscale x 2 x i1> %out
45 define <vscale x 8 x i1> @facgt_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
46 ; CHECK-LABEL: facgt_h:
48 ; CHECK-NEXT: facgt p0.h, p0/z, z0.h, z1.h
50 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.facgt.nxv8f16(<vscale x 8 x i1> %pg,
51 <vscale x 8 x half> %a,
52 <vscale x 8 x half> %b)
53 ret <vscale x 8 x i1> %out
56 define <vscale x 4 x i1> @facgt_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
57 ; CHECK-LABEL: facgt_s:
59 ; CHECK-NEXT: facgt p0.s, p0/z, z0.s, z1.s
61 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> %pg,
62 <vscale x 4 x float> %a,
63 <vscale x 4 x float> %b)
64 ret <vscale x 4 x i1> %out
67 define <vscale x 2 x i1> @facgt_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
68 ; CHECK-LABEL: facgt_d:
70 ; CHECK-NEXT: facgt p0.d, p0/z, z0.d, z1.d
72 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> %pg,
73 <vscale x 2 x double> %a,
74 <vscale x 2 x double> %b)
75 ret <vscale x 2 x i1> %out
82 define <vscale x 8 x i1> @fcmeq_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
83 ; CHECK-LABEL: fcmeq_h:
85 ; CHECK-NEXT: fcmeq p0.h, p0/z, z0.h, z1.h
87 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> %pg,
88 <vscale x 8 x half> %a,
89 <vscale x 8 x half> %b)
90 ret <vscale x 8 x i1> %out
93 define <vscale x 4 x i1> @fcmeq_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
94 ; CHECK-LABEL: fcmeq_s:
96 ; CHECK-NEXT: fcmeq p0.s, p0/z, z0.s, z1.s
98 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> %pg,
99 <vscale x 4 x float> %a,
100 <vscale x 4 x float> %b)
101 ret <vscale x 4 x i1> %out
104 define <vscale x 2 x i1> @fcmeq_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
105 ; CHECK-LABEL: fcmeq_d:
107 ; CHECK-NEXT: fcmeq p0.d, p0/z, z0.d, z1.d
109 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> %pg,
110 <vscale x 2 x double> %a,
111 <vscale x 2 x double> %b)
112 ret <vscale x 2 x i1> %out
115 define <vscale x 2 x i1> @fcmeq_zero(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
116 ; CHECK-LABEL: fcmeq_zero:
118 ; CHECK-NEXT: fcmeq p0.d, p0/z, z0.d, #0.0
120 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> %pg,
121 <vscale x 2 x double> %a,
122 <vscale x 2 x double> zeroinitializer)
123 ret <vscale x 2 x i1> %out
130 define <vscale x 8 x i1> @fcmge_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
131 ; CHECK-LABEL: fcmge_h:
133 ; CHECK-NEXT: fcmge p0.h, p0/z, z0.h, z1.h
135 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> %pg,
136 <vscale x 8 x half> %a,
137 <vscale x 8 x half> %b)
138 ret <vscale x 8 x i1> %out
141 define <vscale x 4 x i1> @fcmge_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
142 ; CHECK-LABEL: fcmge_s:
144 ; CHECK-NEXT: fcmge p0.s, p0/z, z0.s, z1.s
146 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> %pg,
147 <vscale x 4 x float> %a,
148 <vscale x 4 x float> %b)
149 ret <vscale x 4 x i1> %out
152 define <vscale x 2 x i1> @fcmge_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
153 ; CHECK-LABEL: fcmge_d:
155 ; CHECK-NEXT: fcmge p0.d, p0/z, z0.d, z1.d
157 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> %pg,
158 <vscale x 2 x double> %a,
159 <vscale x 2 x double> %b)
160 ret <vscale x 2 x i1> %out
163 define <vscale x 2 x i1> @fcmge_zero(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
164 ; CHECK-LABEL: fcmge_zero:
166 ; CHECK-NEXT: fcmge p0.d, p0/z, z0.d, #0.0
168 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> %pg,
169 <vscale x 2 x double> %a,
170 <vscale x 2 x double> zeroinitializer)
171 ret <vscale x 2 x i1> %out
177 define <vscale x 8 x i1> @fcmgt_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
178 ; CHECK-LABEL: fcmgt_h:
180 ; CHECK-NEXT: fcmgt p0.h, p0/z, z0.h, z1.h
182 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> %pg,
183 <vscale x 8 x half> %a,
184 <vscale x 8 x half> %b)
185 ret <vscale x 8 x i1> %out
188 define <vscale x 4 x i1> @fcmgt_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
189 ; CHECK-LABEL: fcmgt_s:
191 ; CHECK-NEXT: fcmgt p0.s, p0/z, z0.s, z1.s
193 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> %pg,
194 <vscale x 4 x float> %a,
195 <vscale x 4 x float> %b)
196 ret <vscale x 4 x i1> %out
199 define <vscale x 2 x i1> @fcmgt_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
200 ; CHECK-LABEL: fcmgt_d:
202 ; CHECK-NEXT: fcmgt p0.d, p0/z, z0.d, z1.d
204 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> %pg,
205 <vscale x 2 x double> %a,
206 <vscale x 2 x double> %b)
207 ret <vscale x 2 x i1> %out
210 define <vscale x 2 x i1> @fcmgt_zero(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
211 ; CHECK-LABEL: fcmgt_zero:
213 ; CHECK-NEXT: fcmgt p0.d, p0/z, z0.d, #0.0
215 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> %pg,
216 <vscale x 2 x double> %a,
217 <vscale x 2 x double> zeroinitializer)
218 ret <vscale x 2 x i1> %out
224 define <vscale x 8 x i1> @fcmne_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
225 ; CHECK-LABEL: fcmne_h:
227 ; CHECK-NEXT: fcmne p0.h, p0/z, z0.h, z1.h
229 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> %pg,
230 <vscale x 8 x half> %a,
231 <vscale x 8 x half> %b)
232 ret <vscale x 8 x i1> %out
235 define <vscale x 4 x i1> @fcmne_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
236 ; CHECK-LABEL: fcmne_s:
238 ; CHECK-NEXT: fcmne p0.s, p0/z, z0.s, z1.s
240 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> %pg,
241 <vscale x 4 x float> %a,
242 <vscale x 4 x float> %b)
243 ret <vscale x 4 x i1> %out
246 define <vscale x 2 x i1> @fcmne_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
247 ; CHECK-LABEL: fcmne_d:
249 ; CHECK-NEXT: fcmne p0.d, p0/z, z0.d, z1.d
251 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> %pg,
252 <vscale x 2 x double> %a,
253 <vscale x 2 x double> %b)
254 ret <vscale x 2 x i1> %out
257 define <vscale x 2 x i1> @fcmne_zero(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
258 ; CHECK-LABEL: fcmne_zero:
260 ; CHECK-NEXT: fcmne p0.d, p0/z, z0.d, #0.0
262 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> %pg,
263 <vscale x 2 x double> %a,
264 <vscale x 2 x double> zeroinitializer)
265 ret <vscale x 2 x i1> %out
272 define <vscale x 8 x i1> @fcmuo_h(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %b) {
273 ; CHECK-LABEL: fcmuo_h:
275 ; CHECK-NEXT: fcmuo p0.h, p0/z, z0.h, z1.h
277 %out = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> %pg,
278 <vscale x 8 x half> %a,
279 <vscale x 8 x half> %b)
280 ret <vscale x 8 x i1> %out
283 define <vscale x 4 x i1> @fcmuo_s(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %b) {
284 ; CHECK-LABEL: fcmuo_s:
286 ; CHECK-NEXT: fcmuo p0.s, p0/z, z0.s, z1.s
288 %out = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> %pg,
289 <vscale x 4 x float> %a,
290 <vscale x 4 x float> %b)
291 ret <vscale x 4 x i1> %out
294 define <vscale x 2 x i1> @fcmuo_d(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %b) {
295 ; CHECK-LABEL: fcmuo_d:
297 ; CHECK-NEXT: fcmuo p0.d, p0/z, z0.d, z1.d
299 %out = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> %pg,
300 <vscale x 2 x double> %a,
301 <vscale x 2 x double> %b)
302 ret <vscale x 2 x i1> %out
305 declare <vscale x 8 x i1> @llvm.aarch64.sve.facge.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
306 declare <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
307 declare <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
309 declare <vscale x 8 x i1> @llvm.aarch64.sve.facgt.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
310 declare <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
311 declare <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
313 declare <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
314 declare <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
315 declare <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
317 declare <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
318 declare <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
319 declare <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
321 declare <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
322 declare <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
323 declare <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
325 declare <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
326 declare <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
327 declare <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
329 declare <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
330 declare <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
331 declare <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)