File: complex-deinterleaving-reductions-predicated-scalable.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 (325 lines) | stat: -rw-r--r-- 18,359 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
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s --mattr=+sve -o - | FileCheck %s

target triple = "aarch64"

%"class.std::complex" = type { { double, double } }

; Zero initialized reduction. The IR is generated with predicated tail folding (-prefer-predicate-over-epilogue=predicate-dont-vectorize)
;
;   complex<double> x = 0.0 + 0.0i;
;   for (int i = 0; i < 100; ++i)
;       x += a[i] * b[i];
;
define %"class.std::complex" @complex_mul_v2f64(ptr %a, ptr %b) {
; CHECK-LABEL: complex_mul_v2f64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w9, #100 // =0x64
; CHECK-NEXT:    cntd x10
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    mov x11, x10
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    rdvl x12, #2
; CHECK-NEXT:    whilelo p1.d, xzr, x9
; CHECK-NEXT:    zip2 z0.d, z1.d, z1.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z1.d
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:  .LBB0_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    add x13, x0, x8
; CHECK-NEXT:    add x14, x1, x8
; CHECK-NEXT:    zip1 p2.d, p1.d, p1.d
; CHECK-NEXT:    zip2 p3.d, p1.d, p1.d
; CHECK-NEXT:    mov z6.d, z1.d
; CHECK-NEXT:    mov z7.d, z0.d
; CHECK-NEXT:    ld1d { z2.d }, p3/z, [x13, #1, mul vl]
; CHECK-NEXT:    ld1d { z3.d }, p2/z, [x13]
; CHECK-NEXT:    ld1d { z4.d }, p3/z, [x14, #1, mul vl]
; CHECK-NEXT:    ld1d { z5.d }, p2/z, [x14]
; CHECK-NEXT:    whilelo p1.d, x11, x9
; CHECK-NEXT:    add x8, x8, x12
; CHECK-NEXT:    add x11, x11, x10
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #0
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #0
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #90
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #90
; CHECK-NEXT:    mov z0.d, p3/m, z7.d
; CHECK-NEXT:    mov z1.d, p2/m, z6.d
; CHECK-NEXT:    b.mi .LBB0_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp2 z2.d, z1.d, z0.d
; CHECK-NEXT:    uzp1 z0.d, z1.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z0.d
; CHECK-NEXT:    faddv d1, p0, z2.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %active.lane.mask.entry = tail call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 0, i64 100)
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl i64 %0, 1
  %2 = shl nuw nsw i64 %0, 5
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv35 = phi i64 [ %lsr.iv.next36, %vector.body ], [ %1, %entry ]
  %lsr.iv = phi i64 [ %lsr.iv.next, %vector.body ], [ 0, %entry ]
  %active.lane.mask = phi <vscale x 2 x i1> [ %active.lane.mask.entry, %entry ], [ %active.lane.mask.next, %vector.body ]
  %vec.phi = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %15, %vector.body ]
  %vec.phi27 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %16, %vector.body ]
  %scevgep = getelementptr i8, ptr %a, i64 %lsr.iv
  %scevgep34 = getelementptr i8, ptr %b, i64 %lsr.iv
  %interleaved.mask = tail call <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i1> %active.lane.mask)
  %wide.masked.vec = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep, i32 8, <vscale x 4 x i1> %interleaved.mask, <vscale x 4 x double> poison)
  %strided.vec = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec)
  %3 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 0
  %4 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 1
  %interleaved.mask28 = tail call <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i1> %active.lane.mask)
  %wide.masked.vec29 = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep34, i32 8, <vscale x 4 x i1> %interleaved.mask28, <vscale x 4 x double> poison)
  %strided.vec30 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec29)
  %5 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec30, 0
  %6 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec30, 1
  %7 = fmul fast <vscale x 2 x double> %6, %3
  %8 = fmul fast <vscale x 2 x double> %5, %4
  %9 = fmul fast <vscale x 2 x double> %5, %3
  %10 = fadd fast <vscale x 2 x double> %9, %vec.phi27
  %11 = fmul fast <vscale x 2 x double> %6, %4
  %12 = fsub fast <vscale x 2 x double> %10, %11
  %13 = fadd fast <vscale x 2 x double> %8, %vec.phi
  %14 = fadd fast <vscale x 2 x double> %13, %7
  %15 = select fast <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x double> %14, <vscale x 2 x double> %vec.phi
  %16 = select fast <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x double> %12, <vscale x 2 x double> %vec.phi27
  %active.lane.mask.next = tail call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 %lsr.iv35, i64 100)
  %17 = extractelement <vscale x 2 x i1> %active.lane.mask.next, i64 0
  %lsr.iv.next = add i64 %lsr.iv, %2
  %lsr.iv.next36 = add i64 %lsr.iv35, %1
  br i1 %17, label %vector.body, label %exit.block

exit.block:                                     ; preds = %vector.body
  %18 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %16)
  %19 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %15)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %18, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %19, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

; Zero initialized reduction with conditional block. The IR is generated with scalar tail folding (-prefer-predicate-over-epilogue=scalar-epilogue)
;
;   complex<double> x = 0.0 + 0.0i;
;   for (int i = 0; i < 100; ++i)
;       if (cond[i])
;           x += a[i] * b[i];
;
define %"class.std::complex" @complex_mul_predicated_v2f64(ptr %a, ptr %b, ptr %cond) {
; CHECK-LABEL: complex_mul_predicated_v2f64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    cntd x10
; CHECK-NEXT:    mov w12, #100 // =0x64
; CHECK-NEXT:    neg x11, x10
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    mov x9, xzr
; CHECK-NEXT:    and x11, x11, x12
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    rdvl x12, #2
; CHECK-NEXT:    zip2 z0.d, z1.d, z1.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z1.d
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:  .LBB1_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    ld1w { z2.d }, p0/z, [x2, x9, lsl #2]
; CHECK-NEXT:    add x13, x0, x8
; CHECK-NEXT:    add x14, x1, x8
; CHECK-NEXT:    mov z6.d, z1.d
; CHECK-NEXT:    mov z7.d, z0.d
; CHECK-NEXT:    add x9, x9, x10
; CHECK-NEXT:    add x8, x8, x12
; CHECK-NEXT:    cmpne p2.d, p0/z, z2.d, #0
; CHECK-NEXT:    zip1 p1.d, p2.d, p2.d
; CHECK-NEXT:    zip2 p2.d, p2.d, p2.d
; CHECK-NEXT:    ld1d { z2.d }, p2/z, [x13, #1, mul vl]
; CHECK-NEXT:    ld1d { z3.d }, p1/z, [x13]
; CHECK-NEXT:    ld1d { z4.d }, p2/z, [x14, #1, mul vl]
; CHECK-NEXT:    ld1d { z5.d }, p1/z, [x14]
; CHECK-NEXT:    cmp x11, x9
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #0
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #0
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #90
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #90
; CHECK-NEXT:    mov z0.d, p2/m, z7.d
; CHECK-NEXT:    mov z1.d, p1/m, z6.d
; CHECK-NEXT:    b.ne .LBB1_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp2 z2.d, z1.d, z0.d
; CHECK-NEXT:    uzp1 z0.d, z1.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z0.d
; CHECK-NEXT:    faddv d1, p0, z2.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl nuw nsw i64 %0, 1
  %n.mod.vf = urem i64 100, %1
  %n.vec = sub i64 100, %n.mod.vf
  %2 = shl nuw nsw i64 %0, 5
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv48 = phi i64 [ %lsr.iv.next, %vector.body ], [ 0, %entry ]
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.phi = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %predphi34, %vector.body ]
  %vec.phi30 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %predphi, %vector.body ]
  %3 = shl i64 %index, 2
  %scevgep47 = getelementptr i8, ptr %cond, i64 %3
  %wide.load = load <vscale x 2 x i32>, ptr %scevgep47, align 4
  %4 = icmp ne <vscale x 2 x i32> %wide.load, zeroinitializer
  %scevgep49 = getelementptr i8, ptr %a, i64 %lsr.iv48
  %scevgep50 = getelementptr i8, ptr %b, i64 %lsr.iv48
  %interleaved.mask = tail call <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1> %4, <vscale x 2 x i1> %4)
  %wide.masked.vec = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep49, i32 8, <vscale x 4 x i1> %interleaved.mask, <vscale x 4 x double> poison)
  %strided.vec = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec)
  %5 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 0
  %6 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 1
  %wide.masked.vec32 = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep50, i32 8, <vscale x 4 x i1> %interleaved.mask, <vscale x 4 x double> poison)
  %strided.vec33 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec32)
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec33, 0
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec33, 1
  %9 = fmul fast <vscale x 2 x double> %8, %5
  %10 = fmul fast <vscale x 2 x double> %7, %6
  %11 = fmul fast <vscale x 2 x double> %7, %5
  %12 = fadd fast <vscale x 2 x double> %11, %vec.phi30
  %13 = fmul fast <vscale x 2 x double> %8, %6
  %14 = fsub fast <vscale x 2 x double> %12, %13
  %15 = fadd fast <vscale x 2 x double> %10, %vec.phi
  %16 = fadd fast <vscale x 2 x double> %15, %9
  %predphi = select <vscale x 2 x i1> %4, <vscale x 2 x double> %14, <vscale x 2 x double> %vec.phi30
  %predphi34 = select <vscale x 2 x i1> %4, <vscale x 2 x double> %16, <vscale x 2 x double> %vec.phi
  %index.next = add nuw i64 %index, %1
  %lsr.iv.next = add i64 %lsr.iv48, %2
  %17 = icmp eq i64 %n.vec, %index.next
  br i1 %17, label %exit.block, label %vector.body

exit.block:                                     ; preds = %vector.body
  %18 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %predphi)
  %19 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %predphi34)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %18, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %19, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

; Zero initialized reduction with conditional block. The IR is generated with scalar tail folding (-predicate-over-epilogue=predicate-dont-vectorize)
;
;   complex<double> x = 0.0 + 0.0i;
;   for (int i = 0; i < 100; ++i)
;       if (cond[i])
;           x += a[i] * b[i];
;
define %"class.std::complex" @complex_mul_predicated_x2_v2f64(ptr %a, ptr %b, ptr %cond) {
; CHECK-LABEL: complex_mul_predicated_x2_v2f64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w10, #100 // =0x64
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    mov x9, xzr
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    cntd x11
; CHECK-NEXT:    rdvl x12, #2
; CHECK-NEXT:    whilelo p1.d, xzr, x10
; CHECK-NEXT:    zip2 z0.d, z1.d, z1.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z1.d
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:  .LBB2_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    ld1w { z2.d }, p1/z, [x2, x9, lsl #2]
; CHECK-NEXT:    add x13, x0, x8
; CHECK-NEXT:    add x14, x1, x8
; CHECK-NEXT:    mov z6.d, z1.d
; CHECK-NEXT:    mov z7.d, z0.d
; CHECK-NEXT:    add x9, x9, x11
; CHECK-NEXT:    add x8, x8, x12
; CHECK-NEXT:    cmpne p1.d, p1/z, z2.d, #0
; CHECK-NEXT:    zip1 p2.d, p1.d, p1.d
; CHECK-NEXT:    zip2 p3.d, p1.d, p1.d
; CHECK-NEXT:    ld1d { z2.d }, p3/z, [x13, #1, mul vl]
; CHECK-NEXT:    ld1d { z3.d }, p2/z, [x13]
; CHECK-NEXT:    ld1d { z4.d }, p3/z, [x14, #1, mul vl]
; CHECK-NEXT:    ld1d { z5.d }, p2/z, [x14]
; CHECK-NEXT:    whilelo p1.d, x9, x10
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #0
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #0
; CHECK-NEXT:    fcmla z6.d, p0/m, z5.d, z3.d, #90
; CHECK-NEXT:    fcmla z7.d, p0/m, z4.d, z2.d, #90
; CHECK-NEXT:    mov z0.d, p3/m, z7.d
; CHECK-NEXT:    mov z1.d, p2/m, z6.d
; CHECK-NEXT:    b.mi .LBB2_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp2 z2.d, z1.d, z0.d
; CHECK-NEXT:    uzp1 z0.d, z1.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z0.d
; CHECK-NEXT:    faddv d1, p0, z2.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %active.lane.mask.entry = tail call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 0, i64 100)
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl i64 %0, 1
  %2 = shl nuw nsw i64 %0, 5
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv = phi i64 [ %lsr.iv.next, %vector.body ], [ 0, %entry ]
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %active.lane.mask = phi <vscale x 2 x i1> [ %active.lane.mask.entry, %entry ], [ %active.lane.mask.next, %vector.body ]
  %vec.phi = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %19, %vector.body ]
  %vec.phi30 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %21, %vector.body ]
  %3 = shl i64 %index, 2
  %scevgep = getelementptr i8, ptr %cond, i64 %3
  %wide.masked.load = tail call <vscale x 2 x i32> @llvm.masked.load.nxv2i32.p0(ptr %scevgep, i32 4, <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i32> poison)
  %4 = icmp ne <vscale x 2 x i32> %wide.masked.load, zeroinitializer
  %scevgep38 = getelementptr i8, ptr %a, i64 %lsr.iv
  %scevgep39 = getelementptr i8, ptr %b, i64 %lsr.iv
  %5 = select <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i1> %4, <vscale x 2 x i1> zeroinitializer
  %interleaved.mask = tail call <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1> %5, <vscale x 2 x i1> %5)
  %wide.masked.vec = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep38, i32 8, <vscale x 4 x i1> %interleaved.mask, <vscale x 4 x double> poison)
  %strided.vec = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec)
  %6 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 0
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 1
  %interleaved.mask31 = tail call <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1> %5, <vscale x 2 x i1> %5)
  %wide.masked.vec32 = tail call <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr %scevgep39, i32 8, <vscale x 4 x i1> %interleaved.mask31, <vscale x 4 x double> poison)
  %strided.vec33 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.masked.vec32)
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec33, 0
  %9 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec33, 1
  %10 = fmul fast <vscale x 2 x double> %9, %6
  %11 = fmul fast <vscale x 2 x double> %8, %7
  %12 = fmul fast <vscale x 2 x double> %8, %6
  %13 = fadd fast <vscale x 2 x double> %12, %vec.phi30
  %14 = fmul fast <vscale x 2 x double> %9, %7
  %15 = fsub fast <vscale x 2 x double> %13, %14
  %16 = fadd fast <vscale x 2 x double> %11, %vec.phi
  %17 = fadd fast <vscale x 2 x double> %16, %10
  %18 = select <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i1> %4, <vscale x 2 x i1> zeroinitializer
  %19 = select fast <vscale x 2 x i1> %18, <vscale x 2 x double> %17, <vscale x 2 x double> %vec.phi
  %20 = select <vscale x 2 x i1> %active.lane.mask, <vscale x 2 x i1> %4, <vscale x 2 x i1> zeroinitializer
  %21 = select fast <vscale x 2 x i1> %20, <vscale x 2 x double> %15, <vscale x 2 x double> %vec.phi30
  %index.next = add i64 %index, %1
  %22 = add i64 %1, %index
  %active.lane.mask.next = tail call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 %22, i64 100)
  %23 = extractelement <vscale x 2 x i1> %active.lane.mask.next, i64 0
  %lsr.iv.next = add i64 %lsr.iv, %2
  br i1 %23, label %vector.body, label %exit.block

exit.block:                                     ; preds = %vector.body
  %24 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %21)
  %25 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %19)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %24, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %25, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

declare i64 @llvm.vscale.i64()
declare <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64, i64)
declare <vscale x 2 x i32> @llvm.masked.load.nxv2i32.p0(ptr nocapture, i32 immarg, <vscale x 2 x i1>, <vscale x 2 x i32>)
declare <vscale x 4 x double> @llvm.masked.load.nxv4f64.p0(ptr nocapture, i32 immarg, <vscale x 4 x i1>, <vscale x 4 x double>)
declare <vscale x 4 x i1> @llvm.experimental.vector.interleave2.nxv4i1(<vscale x 2 x i1>, <vscale x 2 x i1>)
declare { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double>)
declare double @llvm.vector.reduce.fadd.nxv2f64(double, <vscale x 2 x double>)