1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2 // RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +zve64d \
3 // RUN: -target-feature +f -target-feature +d -disable-O0-optnone \
4 // RUN: -mvscale-min=4 -mvscale-max=4 -emit-llvm -o - %s | \
5 // RUN: opt -S -passes=sroa | FileCheck %s
7 // REQUIRES: riscv-registered-target
11 typedef __rvv_int8m1_t vint8m1_t
;
12 typedef __rvv_uint8m1_t vuint8m1_t
;
13 typedef __rvv_int16m1_t vint16m1_t
;
14 typedef __rvv_uint16m1_t vuint16m1_t
;
15 typedef __rvv_int32m1_t vint32m1_t
;
16 typedef __rvv_uint32m1_t vuint32m1_t
;
17 typedef __rvv_int64m1_t vint64m1_t
;
18 typedef __rvv_uint64m1_t vuint64m1_t
;
19 typedef __rvv_float32m1_t vfloat32m1_t
;
20 typedef __rvv_float64m1_t vfloat64m1_t
;
22 typedef vint8m1_t fixed_int8m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
23 typedef vint16m1_t fixed_int16m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
24 typedef vint32m1_t fixed_int32m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
25 typedef vint64m1_t fixed_int64m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
27 typedef vuint8m1_t fixed_uint8m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
28 typedef vuint16m1_t fixed_uint16m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
29 typedef vuint32m1_t fixed_uint32m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
30 typedef vuint64m1_t fixed_uint64m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
32 typedef vfloat32m1_t fixed_float32m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
33 typedef vfloat64m1_t fixed_float64m1_t
__attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen
)));
37 // CHECK-LABEL: @eq_i8(
39 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
40 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
41 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <32 x i8> [[A]], [[B]]
42 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
43 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
44 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
46 fixed_int8m1_t
eq_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
50 // CHECK-LABEL: @eq_i16(
52 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
53 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
54 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <16 x i16> [[A]], [[B]]
55 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
56 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
57 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
59 fixed_int16m1_t
eq_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
63 // CHECK-LABEL: @eq_i32(
65 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
66 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
67 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <8 x i32> [[A]], [[B]]
68 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
69 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
70 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
72 fixed_int32m1_t
eq_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
76 // CHECK-LABEL: @eq_i64(
78 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
79 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
80 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <4 x i64> [[A]], [[B]]
81 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
82 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
83 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
85 fixed_int64m1_t
eq_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
89 // CHECK-LABEL: @eq_u8(
91 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
92 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
93 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <32 x i8> [[A]], [[B]]
94 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
95 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
96 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
98 fixed_int8m1_t
eq_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
102 // CHECK-LABEL: @eq_u16(
103 // CHECK-NEXT: entry:
104 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
105 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
106 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <16 x i16> [[A]], [[B]]
107 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
108 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
109 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
111 fixed_int16m1_t
eq_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
115 // CHECK-LABEL: @eq_u32(
116 // CHECK-NEXT: entry:
117 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
118 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
119 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <8 x i32> [[A]], [[B]]
120 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
121 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
122 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
124 fixed_int32m1_t
eq_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
128 // CHECK-LABEL: @eq_u64(
129 // CHECK-NEXT: entry:
130 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
131 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
132 // CHECK-NEXT: [[CMP:%.*]] = icmp eq <4 x i64> [[A]], [[B]]
133 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
134 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
135 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
137 fixed_int64m1_t
eq_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
141 // CHECK-LABEL: @eq_f32(
142 // CHECK-NEXT: entry:
143 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
144 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
145 // CHECK-NEXT: [[CMP:%.*]] = fcmp oeq <8 x float> [[A]], [[B]]
146 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
147 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
148 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
150 fixed_int32m1_t
eq_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
154 // CHECK-LABEL: @eq_f64(
155 // CHECK-NEXT: entry:
156 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
157 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
158 // CHECK-NEXT: [[CMP:%.*]] = fcmp oeq <4 x double> [[A]], [[B]]
159 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
160 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
161 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
163 fixed_int64m1_t
eq_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {
169 // CHECK-LABEL: @neq_i8(
170 // CHECK-NEXT: entry:
171 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
172 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
173 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <32 x i8> [[A]], [[B]]
174 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
175 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
176 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
178 fixed_int8m1_t
neq_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
182 // CHECK-LABEL: @neq_i16(
183 // CHECK-NEXT: entry:
184 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
185 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
186 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <16 x i16> [[A]], [[B]]
187 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
188 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
189 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
191 fixed_int16m1_t
neq_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
195 // CHECK-LABEL: @neq_i32(
196 // CHECK-NEXT: entry:
197 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
198 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
199 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <8 x i32> [[A]], [[B]]
200 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
201 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
202 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
204 fixed_int32m1_t
neq_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
208 // CHECK-LABEL: @neq_i64(
209 // CHECK-NEXT: entry:
210 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
211 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
212 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <4 x i64> [[A]], [[B]]
213 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
214 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
215 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
217 fixed_int64m1_t
neq_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
221 // CHECK-LABEL: @neq_u8(
222 // CHECK-NEXT: entry:
223 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
224 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
225 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <32 x i8> [[A]], [[B]]
226 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
227 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
228 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
230 fixed_int8m1_t
neq_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
234 // CHECK-LABEL: @neq_u16(
235 // CHECK-NEXT: entry:
236 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
237 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
238 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <16 x i16> [[A]], [[B]]
239 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
240 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
241 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
243 fixed_int16m1_t
neq_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
247 // CHECK-LABEL: @neq_u32(
248 // CHECK-NEXT: entry:
249 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
250 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
251 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <8 x i32> [[A]], [[B]]
252 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
253 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
254 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
256 fixed_int32m1_t
neq_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
260 // CHECK-LABEL: @neq_u64(
261 // CHECK-NEXT: entry:
262 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
263 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
264 // CHECK-NEXT: [[CMP:%.*]] = icmp ne <4 x i64> [[A]], [[B]]
265 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
266 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
267 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
269 fixed_int64m1_t
neq_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
273 // CHECK-LABEL: @neq_f32(
274 // CHECK-NEXT: entry:
275 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
276 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
277 // CHECK-NEXT: [[CMP:%.*]] = fcmp une <8 x float> [[A]], [[B]]
278 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
279 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
280 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
282 fixed_int32m1_t
neq_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
286 // CHECK-LABEL: @neq_f64(
287 // CHECK-NEXT: entry:
288 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
289 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
290 // CHECK-NEXT: [[CMP:%.*]] = fcmp une <4 x double> [[A]], [[B]]
291 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
292 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
293 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
295 fixed_int64m1_t
neq_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {
301 // CHECK-LABEL: @lt_i8(
302 // CHECK-NEXT: entry:
303 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
304 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
305 // CHECK-NEXT: [[CMP:%.*]] = icmp slt <32 x i8> [[A]], [[B]]
306 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
307 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
308 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
310 fixed_int8m1_t
lt_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
314 // CHECK-LABEL: @lt_i16(
315 // CHECK-NEXT: entry:
316 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
317 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
318 // CHECK-NEXT: [[CMP:%.*]] = icmp slt <16 x i16> [[A]], [[B]]
319 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
320 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
321 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
323 fixed_int16m1_t
lt_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
327 // CHECK-LABEL: @lt_i32(
328 // CHECK-NEXT: entry:
329 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
330 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
331 // CHECK-NEXT: [[CMP:%.*]] = icmp slt <8 x i32> [[A]], [[B]]
332 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
333 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
334 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
336 fixed_int32m1_t
lt_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
340 // CHECK-LABEL: @lt_i64(
341 // CHECK-NEXT: entry:
342 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
343 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
344 // CHECK-NEXT: [[CMP:%.*]] = icmp slt <4 x i64> [[A]], [[B]]
345 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
346 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
347 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
349 fixed_int64m1_t
lt_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
353 // CHECK-LABEL: @lt_u8(
354 // CHECK-NEXT: entry:
355 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
356 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
357 // CHECK-NEXT: [[CMP:%.*]] = icmp ult <32 x i8> [[A]], [[B]]
358 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
359 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
360 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
362 fixed_int8m1_t
lt_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
366 // CHECK-LABEL: @lt_u16(
367 // CHECK-NEXT: entry:
368 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
369 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
370 // CHECK-NEXT: [[CMP:%.*]] = icmp ult <16 x i16> [[A]], [[B]]
371 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
372 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
373 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
375 fixed_int16m1_t
lt_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
379 // CHECK-LABEL: @lt_u32(
380 // CHECK-NEXT: entry:
381 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
382 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
383 // CHECK-NEXT: [[CMP:%.*]] = icmp ult <8 x i32> [[A]], [[B]]
384 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
385 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
386 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
388 fixed_int32m1_t
lt_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
392 // CHECK-LABEL: @lt_u64(
393 // CHECK-NEXT: entry:
394 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
395 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
396 // CHECK-NEXT: [[CMP:%.*]] = icmp ult <4 x i64> [[A]], [[B]]
397 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
398 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
399 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
401 fixed_int64m1_t
lt_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
405 // CHECK-LABEL: @lt_f32(
406 // CHECK-NEXT: entry:
407 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
408 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
409 // CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[A]], [[B]]
410 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
411 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
412 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
414 fixed_int32m1_t
lt_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
418 // CHECK-LABEL: @lt_f64(
419 // CHECK-NEXT: entry:
420 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
421 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
422 // CHECK-NEXT: [[CMP:%.*]] = fcmp olt <4 x double> [[A]], [[B]]
423 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
424 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
425 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
427 fixed_int64m1_t
lt_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {
433 // CHECK-LABEL: @leq_i8(
434 // CHECK-NEXT: entry:
435 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
436 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
437 // CHECK-NEXT: [[CMP:%.*]] = icmp sle <32 x i8> [[A]], [[B]]
438 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
439 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
440 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
442 fixed_int8m1_t
leq_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
446 // CHECK-LABEL: @leq_i16(
447 // CHECK-NEXT: entry:
448 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
449 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
450 // CHECK-NEXT: [[CMP:%.*]] = icmp sle <16 x i16> [[A]], [[B]]
451 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
452 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
453 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
455 fixed_int16m1_t
leq_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
459 // CHECK-LABEL: @leq_i32(
460 // CHECK-NEXT: entry:
461 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
462 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
463 // CHECK-NEXT: [[CMP:%.*]] = icmp sle <8 x i32> [[A]], [[B]]
464 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
465 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
466 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
468 fixed_int32m1_t
leq_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
472 // CHECK-LABEL: @leq_i64(
473 // CHECK-NEXT: entry:
474 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
475 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
476 // CHECK-NEXT: [[CMP:%.*]] = icmp sle <4 x i64> [[A]], [[B]]
477 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
478 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
479 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
481 fixed_int64m1_t
leq_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
485 // CHECK-LABEL: @leq_u8(
486 // CHECK-NEXT: entry:
487 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
488 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
489 // CHECK-NEXT: [[CMP:%.*]] = icmp ule <32 x i8> [[A]], [[B]]
490 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
491 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
492 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
494 fixed_int8m1_t
leq_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
498 // CHECK-LABEL: @leq_u16(
499 // CHECK-NEXT: entry:
500 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
501 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
502 // CHECK-NEXT: [[CMP:%.*]] = icmp ule <16 x i16> [[A]], [[B]]
503 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
504 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
505 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
507 fixed_int16m1_t
leq_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
511 // CHECK-LABEL: @leq_u32(
512 // CHECK-NEXT: entry:
513 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
514 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
515 // CHECK-NEXT: [[CMP:%.*]] = icmp ule <8 x i32> [[A]], [[B]]
516 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
517 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
518 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
520 fixed_int32m1_t
leq_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
524 // CHECK-LABEL: @leq_u64(
525 // CHECK-NEXT: entry:
526 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
527 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
528 // CHECK-NEXT: [[CMP:%.*]] = icmp ule <4 x i64> [[A]], [[B]]
529 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
530 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
531 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
533 fixed_int64m1_t
leq_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
537 // CHECK-LABEL: @leq_f32(
538 // CHECK-NEXT: entry:
539 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
540 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
541 // CHECK-NEXT: [[CMP:%.*]] = fcmp ole <8 x float> [[A]], [[B]]
542 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
543 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
544 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
546 fixed_int32m1_t
leq_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
550 // CHECK-LABEL: @leq_f64(
551 // CHECK-NEXT: entry:
552 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
553 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
554 // CHECK-NEXT: [[CMP:%.*]] = fcmp ole <4 x double> [[A]], [[B]]
555 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
556 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
557 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
559 fixed_int64m1_t
leq_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {
565 // CHECK-LABEL: @gt_i8(
566 // CHECK-NEXT: entry:
567 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
568 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
569 // CHECK-NEXT: [[CMP:%.*]] = icmp sgt <32 x i8> [[A]], [[B]]
570 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
571 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
572 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
574 fixed_int8m1_t
gt_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
578 // CHECK-LABEL: @gt_i16(
579 // CHECK-NEXT: entry:
580 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
581 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
582 // CHECK-NEXT: [[CMP:%.*]] = icmp sgt <16 x i16> [[A]], [[B]]
583 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
584 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
585 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
587 fixed_int16m1_t
gt_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
591 // CHECK-LABEL: @gt_i32(
592 // CHECK-NEXT: entry:
593 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
594 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
595 // CHECK-NEXT: [[CMP:%.*]] = icmp sgt <8 x i32> [[A]], [[B]]
596 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
597 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
598 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
600 fixed_int32m1_t
gt_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
604 // CHECK-LABEL: @gt_i64(
605 // CHECK-NEXT: entry:
606 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
607 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
608 // CHECK-NEXT: [[CMP:%.*]] = icmp sgt <4 x i64> [[A]], [[B]]
609 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
610 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
611 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
613 fixed_int64m1_t
gt_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
617 // CHECK-LABEL: @gt_u8(
618 // CHECK-NEXT: entry:
619 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
620 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
621 // CHECK-NEXT: [[CMP:%.*]] = icmp ugt <32 x i8> [[A]], [[B]]
622 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
623 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
624 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
626 fixed_int8m1_t
gt_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
630 // CHECK-LABEL: @gt_u16(
631 // CHECK-NEXT: entry:
632 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
633 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
634 // CHECK-NEXT: [[CMP:%.*]] = icmp ugt <16 x i16> [[A]], [[B]]
635 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
636 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
637 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
639 fixed_int16m1_t
gt_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
643 // CHECK-LABEL: @gt_u32(
644 // CHECK-NEXT: entry:
645 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
646 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
647 // CHECK-NEXT: [[CMP:%.*]] = icmp ugt <8 x i32> [[A]], [[B]]
648 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
649 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
650 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
652 fixed_int32m1_t
gt_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
656 // CHECK-LABEL: @gt_u64(
657 // CHECK-NEXT: entry:
658 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
659 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
660 // CHECK-NEXT: [[CMP:%.*]] = icmp ugt <4 x i64> [[A]], [[B]]
661 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
662 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
663 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
665 fixed_int64m1_t
gt_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
669 // CHECK-LABEL: @gt_f32(
670 // CHECK-NEXT: entry:
671 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
672 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
673 // CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <8 x float> [[A]], [[B]]
674 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
675 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
676 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
678 fixed_int32m1_t
gt_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
682 // CHECK-LABEL: @gt_f64(
683 // CHECK-NEXT: entry:
684 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
685 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
686 // CHECK-NEXT: [[CMP:%.*]] = fcmp ogt <4 x double> [[A]], [[B]]
687 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
688 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
689 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
691 fixed_int64m1_t
gt_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {
697 // CHECK-LABEL: @geq_i8(
698 // CHECK-NEXT: entry:
699 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
700 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
701 // CHECK-NEXT: [[CMP:%.*]] = icmp sge <32 x i8> [[A]], [[B]]
702 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
703 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
704 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
706 fixed_int8m1_t
geq_i8(fixed_int8m1_t a
, fixed_int8m1_t b
) {
710 // CHECK-LABEL: @geq_i16(
711 // CHECK-NEXT: entry:
712 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
713 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
714 // CHECK-NEXT: [[CMP:%.*]] = icmp sge <16 x i16> [[A]], [[B]]
715 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
716 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
717 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
719 fixed_int16m1_t
geq_i16(fixed_int16m1_t a
, fixed_int16m1_t b
) {
723 // CHECK-LABEL: @geq_i32(
724 // CHECK-NEXT: entry:
725 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
726 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
727 // CHECK-NEXT: [[CMP:%.*]] = icmp sge <8 x i32> [[A]], [[B]]
728 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
729 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
730 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
732 fixed_int32m1_t
geq_i32(fixed_int32m1_t a
, fixed_int32m1_t b
) {
736 // CHECK-LABEL: @geq_i64(
737 // CHECK-NEXT: entry:
738 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
739 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
740 // CHECK-NEXT: [[CMP:%.*]] = icmp sge <4 x i64> [[A]], [[B]]
741 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
742 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
743 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
745 fixed_int64m1_t
geq_i64(fixed_int64m1_t a
, fixed_int64m1_t b
) {
749 // CHECK-LABEL: @geq_u8(
750 // CHECK-NEXT: entry:
751 // CHECK-NEXT: [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)
752 // CHECK-NEXT: [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)
753 // CHECK-NEXT: [[CMP:%.*]] = icmp uge <32 x i8> [[A]], [[B]]
754 // CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
755 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> undef, <32 x i8> [[SEXT]], i64 0)
756 // CHECK-NEXT: ret <vscale x 8 x i8> [[CAST_SCALABLE]]
758 fixed_int8m1_t
geq_u8(fixed_uint8m1_t a
, fixed_uint8m1_t b
) {
762 // CHECK-LABEL: @geq_u16(
763 // CHECK-NEXT: entry:
764 // CHECK-NEXT: [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)
765 // CHECK-NEXT: [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)
766 // CHECK-NEXT: [[CMP:%.*]] = icmp uge <16 x i16> [[A]], [[B]]
767 // CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
768 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> undef, <16 x i16> [[SEXT]], i64 0)
769 // CHECK-NEXT: ret <vscale x 4 x i16> [[CAST_SCALABLE]]
771 fixed_int16m1_t
geq_u16(fixed_uint16m1_t a
, fixed_uint16m1_t b
) {
775 // CHECK-LABEL: @geq_u32(
776 // CHECK-NEXT: entry:
777 // CHECK-NEXT: [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)
778 // CHECK-NEXT: [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)
779 // CHECK-NEXT: [[CMP:%.*]] = icmp uge <8 x i32> [[A]], [[B]]
780 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
781 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
782 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
784 fixed_int32m1_t
geq_u32(fixed_uint32m1_t a
, fixed_uint32m1_t b
) {
788 // CHECK-LABEL: @geq_u64(
789 // CHECK-NEXT: entry:
790 // CHECK-NEXT: [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)
791 // CHECK-NEXT: [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)
792 // CHECK-NEXT: [[CMP:%.*]] = icmp uge <4 x i64> [[A]], [[B]]
793 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
794 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
795 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
797 fixed_int64m1_t
geq_u64(fixed_uint64m1_t a
, fixed_uint64m1_t b
) {
801 // CHECK-LABEL: @geq_f32(
802 // CHECK-NEXT: entry:
803 // CHECK-NEXT: [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)
804 // CHECK-NEXT: [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)
805 // CHECK-NEXT: [[CMP:%.*]] = fcmp oge <8 x float> [[A]], [[B]]
806 // CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
807 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[SEXT]], i64 0)
808 // CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
810 fixed_int32m1_t
geq_f32(fixed_float32m1_t a
, fixed_float32m1_t b
) {
814 // CHECK-LABEL: @geq_f64(
815 // CHECK-NEXT: entry:
816 // CHECK-NEXT: [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)
817 // CHECK-NEXT: [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)
818 // CHECK-NEXT: [[CMP:%.*]] = fcmp oge <4 x double> [[A]], [[B]]
819 // CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
820 // CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[SEXT]], i64 0)
821 // CHECK-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
823 fixed_int64m1_t
geq_f64(fixed_float64m1_t a
, fixed_float64m1_t b
) {