File: sme2-intrinsics-fp-dots.ll

package info (click to toggle)
llvm-toolchain-17 1%3A17.0.6-22
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 1,799,624 kB
  • sloc: cpp: 6,428,607; ansic: 1,383,196; asm: 793,408; python: 223,504; objc: 75,364; f90: 60,502; lisp: 33,869; pascal: 15,282; sh: 9,684; perl: 7,453; ml: 4,937; awk: 3,523; makefile: 2,889; javascript: 2,149; xml: 888; fortran: 619; cs: 573
file content (274 lines) | stat: -rw-r--r-- 16,437 bytes parent folder | download | duplicates (2) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
 
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -verify-machineinstrs < %s | FileCheck %s

target triple="aarch64-linux-gnu"


; == Multi, multi (16-bit float) ==

define void @fdot_multi_za32_f16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3) #0 {
; CHECK-LABEL: fdot_multi_za32_f16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z5.d, z4.d
; CHECK-NEXT:    mov z7.d, z2.d
; CHECK-NEXT:    mov z4.d, z3.d
; CHECK-NEXT:    mov z6.d, z1.d
; CHECK-NEXT:    fdot za.s[w8, 0, vgx2], { z6.h, z7.h }, { z4.h, z5.h }
; CHECK-NEXT:    fdot za.s[w8, 7, vgx2], { z6.h, z7.h }, { z4.h, z5.h }
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3)
  ret void
}

define void @fdot_multi_za32_f16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3,
; CHECK-LABEL: fdot_multi_za32_f16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z26.d, z7.d
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z25.d, z6.d
; CHECK-NEXT:    mov z31.d, z4.d
; CHECK-NEXT:    mov z24.d, z5.d
; CHECK-NEXT:    mov z30.d, z3.d
; CHECK-NEXT:    ld1h { z27.h }, p0/z, [x1]
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z29.d, z2.d
; CHECK-NEXT:    mov z28.d, z1.d
; CHECK-NEXT:    fdot za.s[w8, 0, vgx4], { z28.h - z31.h }, { z24.h - z27.h }
; CHECK-NEXT:    fdot za.s[w8, 7, vgx4], { z28.h - z31.h }, { z24.h - z27.h }
; CHECK-NEXT:    ret
                                        <vscale x 8 x half> %zn4, <vscale x 8 x half> %zn5, <vscale x 8 x half> %zn6, <vscale x 8 x half> %zn7) #0 {
  call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3,
                                                      <vscale x 8 x half> %zn4, <vscale x 8 x half> %zn5, <vscale x 8 x half> %zn6, <vscale x 8 x half> %zn7)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3,
                                                      <vscale x 8 x half> %zn4, <vscale x 8 x half> %zn5, <vscale x 8 x half> %zn6, <vscale x 8 x half> %zn7)
  ret void
}


; == Multi, multi (16-bit bfloat) ==

define void @bfdot_multi_za32_bf16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3) #0 {
; CHECK-LABEL: bfdot_multi_za32_bf16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z5.d, z4.d
; CHECK-NEXT:    mov z7.d, z2.d
; CHECK-NEXT:    mov z4.d, z3.d
; CHECK-NEXT:    mov z6.d, z1.d
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx2], { z6.h, z7.h }, { z4.h, z5.h }
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx2], { z6.h, z7.h }, { z4.h, z5.h }
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3)
  ret void
}

define void @fdot_multi_za32_bf16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3,
; CHECK-LABEL: fdot_multi_za32_bf16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z26.d, z7.d
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z25.d, z6.d
; CHECK-NEXT:    mov z31.d, z4.d
; CHECK-NEXT:    mov z24.d, z5.d
; CHECK-NEXT:    mov z30.d, z3.d
; CHECK-NEXT:    ld1h { z27.h }, p0/z, [x1]
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z29.d, z2.d
; CHECK-NEXT:    mov z28.d, z1.d
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx4], { z28.h - z31.h }, { z24.h - z27.h }
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx4], { z28.h - z31.h }, { z24.h - z27.h }
; CHECK-NEXT:    ret
                                        <vscale x 8 x bfloat> %zn4, <vscale x 8 x bfloat> %zn5, <vscale x 8 x bfloat> %zn6, <vscale x 8 x bfloat> %zn7) #0 {
  call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3,
                                                       <vscale x 8 x bfloat> %zn4, <vscale x 8 x bfloat> %zn5, <vscale x 8 x bfloat> %zn6, <vscale x 8 x bfloat> %zn7)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3,
                                                        <vscale x 8 x bfloat> %zn4, <vscale x 8 x bfloat> %zn5, <vscale x 8 x bfloat> %zn6, <vscale x 8 x bfloat> %zn7)
  ret void
}


; == Multi, single (16-bit float) ==

define void @fdot_single_za32_f16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2) #0 {
; CHECK-LABEL: fdot_single_za32_f16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z1_z2 def $z1_z2
; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z1_z2 def $z1_z2
; CHECK-NEXT:    fdot za.s[w8, 0, vgx2], { z1.h, z2.h }, z3.h
; CHECK-NEXT:    fdot za.s[w8, 7, vgx2], { z1.h, z2.h }, z3.h
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2)
  ret void
}

define void @fdot_single_za32_f16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3, <vscale x 8 x half> %zn4) #0 {
; CHECK-LABEL: fdot_single_za32_f16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $z4 killed $z4 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    fdot za.s[w8, 0, vgx4], { z1.h - z4.h }, z5.h
; CHECK-NEXT:    fdot za.s[w8, 7, vgx4], { z1.h - z4.h }, z5.h
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3, <vscale x 8 x half> %zn4)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3, <vscale x 8 x half> %zn4)
  ret void
}


; == Multi, single (16-bit bfloat) ==

define void @bfdot_single_za32_bf16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2) #0 {
; CHECK-LABEL: bfdot_single_za32_bf16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z1_z2 def $z1_z2
; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z1_z2 def $z1_z2
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx2], { z1.h, z2.h }, z3.h
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx2], { z1.h, z2.h }, z3.h
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2)
  ret void
}

define void @bfdot_single_za32_bf16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3, <vscale x 8 x bfloat> %zn4) #0 {
; CHECK-LABEL: bfdot_single_za32_bf16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $z4 killed $z4 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z1_z2_z3_z4 def $z1_z2_z3_z4
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx4], { z1.h - z4.h }, z5.h
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx4], { z1.h - z4.h }, z5.h
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3, <vscale x 8 x bfloat> %zn4)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3, <vscale x 8 x bfloat> %zn4)
  ret void
}


; == Multi, indexed (16-bit float) ==

define void @fdot_lane_za32_f16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2) #0 {
; CHECK-LABEL: fdot_lane_za32_f16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    fdot za.s[w8, 0, vgx2], { z4.h, z5.h }, z3.h[3]
; CHECK-NEXT:    fdot za.s[w8, 7, vgx2], { z4.h, z5.h }, z3.h[3]
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, i32 3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, i32 3)
  ret void
}

define void @fdot_lane_za32_f16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3, <vscale x 8 x half> %zn4) #0 {
; CHECK-LABEL: fdot_lane_za32_f16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z27.d, z4.d
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z26.d, z3.d
; CHECK-NEXT:    mov z25.d, z2.d
; CHECK-NEXT:    mov z24.d, z1.d
; CHECK-NEXT:    fdot za.s[w8, 0, vgx4], { z24.h - z27.h }, z5.h[3]
; CHECK-NEXT:    fdot za.s[w8, 7, vgx4], { z24.h - z27.h }, z5.h[3]
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8f16(i32 %slice, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3,
                                                           <vscale x 8 x half> %zn4, i32 3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8f16(i32 %slice2, <vscale x 8 x half> %zn0, <vscale x 8 x half> %zn1, <vscale x 8 x half> %zn2, <vscale x 8 x half> %zn3,
                                                           <vscale x 8 x half> %zn4, i32 3)
  ret void
}


; == Multi, indexed (16-bit bfloat) ==

define void @bfdot_lane_za32_bf16_vg1x2(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2) #0 {
; CHECK-LABEL: bfdot_lane_za32_bf16_vg1x2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx2], { z4.h, z5.h }, z3.h[3]
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx2], { z4.h, z5.h }, z3.h[3]
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, i32 3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, i32 3)
  ret void
}

define void @bfdot_lane_za32_bf16_vg1x4(i32 %slice, <vscale x 16 x i8> %unused, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3, <vscale x 8 x bfloat> %zn4) #0 {
; CHECK-LABEL: bfdot_lane_za32_bf16_vg1x4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z27.d, z4.d
; CHECK-NEXT:    mov w8, w0
; CHECK-NEXT:    mov z26.d, z3.d
; CHECK-NEXT:    mov z25.d, z2.d
; CHECK-NEXT:    mov z24.d, z1.d
; CHECK-NEXT:    bfdot za.s[w8, 0, vgx4], { z24.h - z27.h }, z5.h[3]
; CHECK-NEXT:    bfdot za.s[w8, 7, vgx4], { z24.h - z27.h }, z5.h[3]
; CHECK-NEXT:    ret
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3,
                                                             <vscale x 8 x bfloat> %zn4, i32 3)
  %slice2 = add i32 %slice, 7
  call void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8bf16(i32 %slice2, <vscale x 8 x bfloat> %zn0, <vscale x 8 x bfloat> %zn1, <vscale x 8 x bfloat> %zn2, <vscale x 8 x bfloat> %zn3,
                                                             <vscale x 8 x bfloat> %zn4, i32 3)
  ret void
}


attributes #0 = { nounwind "target-features"="+sme2" }


; == Multi, multi (16-bit float)

declare void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>)
declare void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>,
                                                       <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>)

; == Multi, multi (16-bit bfloat)

declare void @llvm.aarch64.sme.fdot.za32.vg1x2.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
declare void @llvm.aarch64.sme.fdot.za32.vg1x4.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>,
                                                         <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)

; == Multi, single (16-bit float)

declare void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>)
declare void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>)

; == Multi, single (16-bit bfloat)

declare void @llvm.aarch64.sme.fdot.single.za32.vg1x2.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
declare void @llvm.aarch64.sme.fdot.single.za32.vg1x4.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)

; == Multi, indexed (16-bit float)

declare void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, i32)
declare void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, i32)

; == Multi, indexed (16-bit bfloat)

declare void @llvm.aarch64.sme.fdot.lane.za32.vg1x2.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32)
declare void @llvm.aarch64.sme.fdot.lane.za32.vg1x4.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32)