1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=x86_64-- | FileCheck %s
3; PR1103
4
5target datalayout = "e-p:64:64"
6@i6000 = global [128 x i64] zeroinitializer, align 16
7
8
9define void @foo(i32* %a0, i32* %a1, i32* %a2, i32* %a3, i32* %a4, i32* %a5) {
10; CHECK-LABEL: foo:
11; CHECK:       # %bb.0: # %b
12; CHECK-NEXT:    pushq %rbp
13; CHECK-NEXT:    .cfi_def_cfa_offset 16
14; CHECK-NEXT:    .cfi_offset %rbp, -16
15; CHECK-NEXT:    movq %rsp, %rbp
16; CHECK-NEXT:    .cfi_def_cfa_register %rbp
17; CHECK-NEXT:    movslq (%rdi), %rax
18; CHECK-NEXT:    movslq (%rsi), %r8
19; CHECK-NEXT:    movslq (%rdx), %r10
20; CHECK-NEXT:    movl (%rcx), %edi
21; CHECK-NEXT:    movslq (%r9), %rcx
22; CHECK-NEXT:    movq %rsp, %rdx
23; CHECK-NEXT:    subl %eax, %r8d
24; CHECK-NEXT:    movslq %r8d, %rsi
25; CHECK-NEXT:    js .LBB0_1
26; CHECK-NEXT:  # %bb.11: # %b63
27; CHECK-NEXT:    testq %rsi, %rsi
28; CHECK-NEXT:    js .LBB0_14
29; CHECK-NEXT:  # %bb.12:
30; CHECK-NEXT:    xorl %eax, %eax
31; CHECK-NEXT:    .p2align 4, 0x90
32; CHECK-NEXT:  .LBB0_13: # %a25b
33; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
34; CHECK-NEXT:    testb %al, %al
35; CHECK-NEXT:    je .LBB0_13
36; CHECK-NEXT:  .LBB0_14: # %b85
37; CHECK-NEXT:    movb $1, %al
38; CHECK-NEXT:    testb %al, %al
39; CHECK-NEXT:    jne .LBB0_1
40; CHECK-NEXT:  # %bb.15:
41; CHECK-NEXT:    xorl %eax, %eax
42; CHECK-NEXT:    .p2align 4, 0x90
43; CHECK-NEXT:  .LBB0_16: # %a25b140
44; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
45; CHECK-NEXT:    testb %al, %al
46; CHECK-NEXT:    je .LBB0_16
47; CHECK-NEXT:  .LBB0_1: # %a29b
48; CHECK-NEXT:    cmpl %r10d, %edi
49; CHECK-NEXT:    js .LBB0_10
50; CHECK-NEXT:  # %bb.2: # %b158
51; CHECK-NEXT:    xorl %edi, %edi
52; CHECK-NEXT:    xorps %xmm0, %xmm0
53; CHECK-NEXT:    movb $1, %r10b
54; CHECK-NEXT:    jmp .LBB0_3
55; CHECK-NEXT:    .p2align 4, 0x90
56; CHECK-NEXT:  .LBB0_9: # %b1606
57; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
58; CHECK-NEXT:    testb %dil, %dil
59; CHECK-NEXT:    je .LBB0_10
60; CHECK-NEXT:  .LBB0_3: # %a29b173
61; CHECK-NEXT:    # =>This Loop Header: Depth=1
62; CHECK-NEXT:    # Child Loop BB0_37 Depth 2
63; CHECK-NEXT:    # Child Loop BB0_19 Depth 2
64; CHECK-NEXT:    # Child Loop BB0_20 Depth 2
65; CHECK-NEXT:    # Child Loop BB0_21 Depth 3
66; CHECK-NEXT:    # Child Loop BB0_23 Depth 2
67; CHECK-NEXT:    # Child Loop BB0_24 Depth 3
68; CHECK-NEXT:    # Child Loop BB0_26 Depth 2
69; CHECK-NEXT:    # Child Loop BB0_38 Depth 3
70; CHECK-NEXT:    # Child Loop BB0_29 Depth 3
71; CHECK-NEXT:    # Child Loop BB0_30 Depth 2
72; CHECK-NEXT:    # Child Loop BB0_39 Depth 3
73; CHECK-NEXT:    # Child Loop BB0_33 Depth 3
74; CHECK-NEXT:    # Child Loop BB0_34 Depth 2
75; CHECK-NEXT:    # Child Loop BB0_36 Depth 2
76; CHECK-NEXT:    testl %r8d, %r8d
77; CHECK-NEXT:    js .LBB0_4
78; CHECK-NEXT:  # %bb.17: # %b179
79; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
80; CHECK-NEXT:    testq %rsi, %rsi
81; CHECK-NEXT:    js .LBB0_18
82; CHECK-NEXT:    .p2align 4, 0x90
83; CHECK-NEXT:  .LBB0_37: # %a30b
84; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
85; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
86; CHECK-NEXT:    testb %dil, %dil
87; CHECK-NEXT:    je .LBB0_37
88; CHECK-NEXT:  .LBB0_18: # %b188
89; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
90; CHECK-NEXT:    testb %r10b, %r10b
91; CHECK-NEXT:    jne .LBB0_4
92; CHECK-NEXT:    .p2align 4, 0x90
93; CHECK-NEXT:  .LBB0_19: # %a30b294
94; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
95; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
96; CHECK-NEXT:    testb %dil, %dil
97; CHECK-NEXT:    je .LBB0_19
98; CHECK-NEXT:  .LBB0_4: # %a33b
99; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
100; CHECK-NEXT:    movl %ecx, %eax
101; CHECK-NEXT:    orl %r8d, %eax
102; CHECK-NEXT:    movl %eax, %r9d
103; CHECK-NEXT:    shrl $31, %r9d
104; CHECK-NEXT:    testl %eax, %eax
105; CHECK-NEXT:    jns .LBB0_20
106; CHECK-NEXT:  .LBB0_5: # %a50b
107; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
108; CHECK-NEXT:    movl %r8d, %eax
109; CHECK-NEXT:    orl %ecx, %eax
110; CHECK-NEXT:    movl %eax, %r11d
111; CHECK-NEXT:    shrl $31, %r11d
112; CHECK-NEXT:    testl %eax, %eax
113; CHECK-NEXT:    jns .LBB0_26
114; CHECK-NEXT:  .LBB0_6: # %a57b
115; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
116; CHECK-NEXT:    testb %r9b, %r9b
117; CHECK-NEXT:    je .LBB0_30
118; CHECK-NEXT:  .LBB0_7: # %a66b
119; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
120; CHECK-NEXT:    testb %r11b, %r11b
121; CHECK-NEXT:    jne .LBB0_8
122; CHECK-NEXT:    .p2align 4, 0x90
123; CHECK-NEXT:  .LBB0_34: # %a74b
124; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
125; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
126; CHECK-NEXT:    testb %dil, %dil
127; CHECK-NEXT:    jne .LBB0_34
128; CHECK-NEXT:  # %bb.35: # %b1582
129; CHECK-NEXT:    # in Loop: Header=BB0_34 Depth=2
130; CHECK-NEXT:    testb %dil, %dil
131; CHECK-NEXT:    jne .LBB0_34
132; CHECK-NEXT:  .LBB0_8: # %a93b
133; CHECK-NEXT:    # in Loop: Header=BB0_3 Depth=1
134; CHECK-NEXT:    testl %r8d, %r8d
135; CHECK-NEXT:    js .LBB0_9
136; CHECK-NEXT:    .p2align 4, 0x90
137; CHECK-NEXT:  .LBB0_36: # %a97b
138; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
139; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
140; CHECK-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
141; CHECK-NEXT:    addss %xmm0, %xmm1
142; CHECK-NEXT:    addss %xmm0, %xmm1
143; CHECK-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
144; CHECK-NEXT:    addss %xmm0, %xmm2
145; CHECK-NEXT:    addss %xmm1, %xmm2
146; CHECK-NEXT:    movss %xmm2, {{.*}}(%rip)
147; CHECK-NEXT:    testb %dil, %dil
148; CHECK-NEXT:    jne .LBB0_36
149; CHECK-NEXT:    jmp .LBB0_9
150; CHECK-NEXT:    .p2align 4, 0x90
151; CHECK-NEXT:  .LBB0_22: # %b463
152; CHECK-NEXT:    # in Loop: Header=BB0_20 Depth=2
153; CHECK-NEXT:    testb %dil, %dil
154; CHECK-NEXT:    je .LBB0_23
155; CHECK-NEXT:  .LBB0_20: # %b341
156; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
157; CHECK-NEXT:    # => This Loop Header: Depth=2
158; CHECK-NEXT:    # Child Loop BB0_21 Depth 3
159; CHECK-NEXT:    testq %rcx, %rcx
160; CHECK-NEXT:    js .LBB0_22
161; CHECK-NEXT:    .p2align 4, 0x90
162; CHECK-NEXT:  .LBB0_21: # %a35b
163; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
164; CHECK-NEXT:    # Parent Loop BB0_20 Depth=2
165; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
166; CHECK-NEXT:    testb %dil, %dil
167; CHECK-NEXT:    je .LBB0_21
168; CHECK-NEXT:    jmp .LBB0_22
169; CHECK-NEXT:    .p2align 4, 0x90
170; CHECK-NEXT:  .LBB0_28: # %b1016
171; CHECK-NEXT:    # in Loop: Header=BB0_26 Depth=2
172; CHECK-NEXT:    testq %rcx, %rcx
173; CHECK-NEXT:    jle .LBB0_6
174; CHECK-NEXT:  .LBB0_26: # %b858
175; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
176; CHECK-NEXT:    # => This Loop Header: Depth=2
177; CHECK-NEXT:    # Child Loop BB0_38 Depth 3
178; CHECK-NEXT:    # Child Loop BB0_29 Depth 3
179; CHECK-NEXT:    testq %rsi, %rsi
180; CHECK-NEXT:    js .LBB0_27
181; CHECK-NEXT:    .p2align 4, 0x90
182; CHECK-NEXT:  .LBB0_38: # %a53b
183; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
184; CHECK-NEXT:    # Parent Loop BB0_26 Depth=2
185; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
186; CHECK-NEXT:    testb %dil, %dil
187; CHECK-NEXT:    je .LBB0_38
188; CHECK-NEXT:  .LBB0_27: # %b879
189; CHECK-NEXT:    # in Loop: Header=BB0_26 Depth=2
190; CHECK-NEXT:    testb %r10b, %r10b
191; CHECK-NEXT:    jne .LBB0_28
192; CHECK-NEXT:    .p2align 4, 0x90
193; CHECK-NEXT:  .LBB0_29: # %a53b1019
194; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
195; CHECK-NEXT:    # Parent Loop BB0_26 Depth=2
196; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
197; CHECK-NEXT:    testq %rsi, %rsi
198; CHECK-NEXT:    jle .LBB0_29
199; CHECK-NEXT:    jmp .LBB0_28
200; CHECK-NEXT:    .p2align 4, 0x90
201; CHECK-NEXT:  .LBB0_32: # %b1263
202; CHECK-NEXT:    # in Loop: Header=BB0_30 Depth=2
203; CHECK-NEXT:    testq %rsi, %rsi
204; CHECK-NEXT:    jle .LBB0_7
205; CHECK-NEXT:  .LBB0_30: # %b1117
206; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
207; CHECK-NEXT:    # => This Loop Header: Depth=2
208; CHECK-NEXT:    # Child Loop BB0_39 Depth 3
209; CHECK-NEXT:    # Child Loop BB0_33 Depth 3
210; CHECK-NEXT:    testq %rcx, %rcx
211; CHECK-NEXT:    js .LBB0_31
212; CHECK-NEXT:    .p2align 4, 0x90
213; CHECK-NEXT:  .LBB0_39: # %a63b
214; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
215; CHECK-NEXT:    # Parent Loop BB0_30 Depth=2
216; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
217; CHECK-NEXT:    testq %rcx, %rcx
218; CHECK-NEXT:    jle .LBB0_39
219; CHECK-NEXT:  .LBB0_31: # %b1139
220; CHECK-NEXT:    # in Loop: Header=BB0_30 Depth=2
221; CHECK-NEXT:    testq %rcx, %rcx
222; CHECK-NEXT:    jle .LBB0_32
223; CHECK-NEXT:    .p2align 4, 0x90
224; CHECK-NEXT:  .LBB0_33: # %a63b1266
225; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
226; CHECK-NEXT:    # Parent Loop BB0_30 Depth=2
227; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
228; CHECK-NEXT:    testq %rcx, %rcx
229; CHECK-NEXT:    jle .LBB0_33
230; CHECK-NEXT:    jmp .LBB0_32
231; CHECK-NEXT:    .p2align 4, 0x90
232; CHECK-NEXT:  .LBB0_25: # %b712
233; CHECK-NEXT:    # in Loop: Header=BB0_23 Depth=2
234; CHECK-NEXT:    testb %dil, %dil
235; CHECK-NEXT:    je .LBB0_5
236; CHECK-NEXT:  .LBB0_23: # %b535
237; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
238; CHECK-NEXT:    # => This Loop Header: Depth=2
239; CHECK-NEXT:    # Child Loop BB0_24 Depth 3
240; CHECK-NEXT:    testq %rsi, %rsi
241; CHECK-NEXT:    js .LBB0_25
242; CHECK-NEXT:    .p2align 4, 0x90
243; CHECK-NEXT:  .LBB0_24: # %a45b
244; CHECK-NEXT:    # Parent Loop BB0_3 Depth=1
245; CHECK-NEXT:    # Parent Loop BB0_23 Depth=2
246; CHECK-NEXT:    # => This Inner Loop Header: Depth=3
247; CHECK-NEXT:    testb %dil, %dil
248; CHECK-NEXT:    je .LBB0_24
249; CHECK-NEXT:    jmp .LBB0_25
250; CHECK-NEXT:  .LBB0_10: # %a109b
251; CHECK-NEXT:    movq %rbp, %rsp
252; CHECK-NEXT:    popq %rbp
253; CHECK-NEXT:    .cfi_def_cfa %rsp, 8
254; CHECK-NEXT:    retq
255b:
256	%r = load i32, i32* %a0
257	%r2 = load i32, i32* %a1
258	%r4 = load i32, i32* %a2
259	%r6 = load i32, i32* %a3
260	%r8 = load i32, i32* %a4
261	%r14 = load i32, i32* %a5
262	%rx = sext i32 %r2 to i64
263	%r9 = sext i32 %r to i64
264	%r11 = add i64 %rx, 0
265	%ras = icmp slt i64 %r11, 0
266	%r12 = select i1 %ras, i64 0, i64 %r11
267	%r16 = sext i32 %r14 to i64
268	%r17 = sext i32 %r8 to i64
269	%r18 = sub i64 %r16, 0
270	%r19 = add i64 %r18, 0
271	%r20 = icmp slt i64 %r19, 0
272	%r19h = add i64 %r18, 0
273	%r22 = select i1 %r20, i64 1, i64 %r19h
274	%r23 = mul i64 %r22, 0
275	%r23a = trunc i64 %r23 to i32
276	%r24 = shl i32 %r23a, 0
277	%r25 = add i32 %r24, 0
278	%ras2 = alloca i8, i32 %r25, align 16
279	%r28 = getelementptr i8, i8* %ras2, i32 0
280	%r38 = shl i64 %r12, 0
281	%s2013 = add i64 %r38, 0
282	%c22012 = getelementptr i8, i8* %ras2, i64 %s2013
283	%r42 = shl i64 %r12, 0
284	%s2011 = add i64 %r42, 16
285	%c22010 = getelementptr i8, i8* %ras2, i64 %s2011
286	%r50 = add i64 %r16, 0
287	%r51 = icmp slt i64 %r50, 0
288	%r50sh = shl i64 %r50, 0
289	%r50j = add i64 %r50sh, 0
290	%r54 = select i1 %r51, i64 0, i64 %r50j
291	%r56 = mul i64 %r54, %r12
292	%r28s = add i64 %r56, 16
293	%c2 = getelementptr i8, i8* %ras2, i64 %r28s
294	%r60 = sub i32 %r2, %r
295	%r61 = icmp slt i32 %r60, 0
296	br i1 %r61, label %a29b, label %b63
297a29b:
298	%r155 = sub i32 %r6, %r4
299	%r156 = icmp slt i32 %r155, 0
300	br i1 %r156, label %a109b, label %b158
301b63:
302	%r66 = sext i32 %r60 to i64
303	%r67 = add i64 %r66, 0
304	%r76 = mul i64 %r17, 0
305	%r82 = add i64 %r76, 0
306	%r84 = icmp slt i64 %r67, 0
307	br i1 %r84, label %b85, label %a25b
308b85:
309	%e641 = phi i64 [ 0, %b63 ], [ %r129, %a25b ]
310	%r137 = icmp slt i64 %e641, 0
311	br i1 %r137, label %a25b140q, label %a29b
312a25b140q:
313	br label %a25b140
314a25b:
315	%w1989 = phi i64 [ 0, %b63 ], [ %v1990, %a25b ]
316	%e642 = shl i64 %w1989, 0
317	%r129 = add i64 %e642, 0
318	%r132 = add i64 %e642, 0
319	%r134 = icmp slt i64 %r132, 0
320	%v1990 = add i64 %w1989, 0
321	br i1 %r134, label %b85, label %a25b
322a25b140:
323	%w1982 = phi i64 [ 0, %a25b140q ], [ %v1983, %a25b140 ]
324	%r145 = add i64 %r82, 0
325	%v1983 = add i64 %w1982, 0
326	%u1987 = icmp slt i64 %v1983, 0
327	br i1 %u1987, label %a29b, label %a25b140
328b158:
329	%r161 = sext i32 %r to i64
330	%r163 = sext i32 %r4 to i64
331	br label %a29b173
332a29b173:
333	%w1964 = phi i64 [ 0, %b158 ], [ %v1973, %b1606 ]
334	%b1974 = mul i64 %r163, 0
335	%b1975 = add i64 %r161, 0
336	%b1976 = mul i64 %w1964, 0
337	%b1977 = add i64 %b1976, 0
338	%s761 = bitcast i64 %b1977 to i64
339	%b1980 = mul i64 %w1964, 0
340	%s661 = add i64 %b1980, 0
341	br i1 %r61, label %a33b, label %b179
342a33b:
343	%r328 = icmp slt i32 %r14, 0
344	%r335 = or i1 %r328, %r61
345	br i1 %r335, label %a50b, label %b341
346b179:
347	%r182 = sext i32 %r60 to i64
348	%r183 = add i64 %r182, 0
349	%r187 = icmp slt i64 %r183, 0
350	br i1 %r187, label %b188, label %a30b
351b188:
352	%e653 = phi i64 [ 0, %b179 ], [ %r283, %a30b ]
353	%r291 = icmp slt i64 %e653, 0
354	br i1 %r291, label %a30b294q, label %a33b
355a30b294q:
356	br label %a30b294
357a30b:
358	%w = phi i64 [ 0, %b179 ], [ %v, %a30b ]
359	%b2 = shl i64 %w, 0
360	%r283 = add i64 %b2, 0
361	%r286 = add i64 %b2, 0
362	%r288 = icmp slt i64 %r286, 0
363	%v = add i64 %w, 0
364	br i1 %r288, label %b188, label %a30b
365a30b294:
366	%w1847 = phi i64 [ 0, %a30b294q ], [ %v1848, %a30b294 ]
367	%v1848 = add i64 %w1847, 0
368	%u = icmp slt i64 %v1848, 0
369	br i1 %u, label %a33b, label %a30b294
370a50b:
371	%r814 = add i32 %r14, 0
372	%r815 = icmp slt i32 %r814, 0
373	%r817 = or i1 %r61, %r815
374	br i1 %r817, label %a57b, label %b820
375b341:
376	%w1874 = phi i64 [ 0, %a33b ], [ %v1880, %b463 ]
377	%d753 = bitcast i64 %w1874 to i64
378	%r343 = add i64 %s661, 0
379	%r346 = add i64 %r343, 0
380	%r347 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r346
381	%r348 = load float, float* %r347
382	%r352 = add i64 %r343, 0
383	%r353 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r352
384	%r354 = load float, float* %r353
385	%r362 = load float, float* bitcast ([128 x i64]* @i6000 to float*)
386	%r363 = fadd float 0.000000e+00, %r362
387	%r370 = load float, float* bitcast ([128 x i64]* @i6000 to float*)
388	%r376 = icmp slt i64 %r16, 0
389	br i1 %r376, label %b377, label %a35b
390b377:
391	%d753p = phi i64 [ %d753, %b341 ], [ %r411, %a35b ]
392	%s761p = phi i64 [ %s761, %b341 ], [ 322, %a35b ]
393	%e784 = phi i64 [ 0, %b341 ], [ %r454, %a35b ]
394	%s794 = add i64 %d753p, 0
395	%r462 = icmp slt i64 %e784, 0
396	br i1 %r462, label %a35b465, label %b463
397a35b:
398	%w1865 = phi i64 [ 0, %b341 ], [ %v1866, %a35b ]
399	%e785 = shl i64 %w1865, 0
400	%b1877 = mul i64 %w1865, 0
401	%s795 = add i64 %b1877, 0
402	%r399 = fadd float %r354, 0.000000e+00
403	%r402 = fadd float %r370, 0.000000e+00
404	%r403 = fadd float %r348, 0.000000e+00
405	%r411 = add i64 %s795, 0
406	%r431 = fadd float %r362, 0.000000e+00
407	%r454 = add i64 %e785, 0
408	%r457 = add i64 %e785, 0
409	%r459 = icmp slt i64 %r457, 0
410	%v1866 = add i64 %w1865, 0
411	br i1 %r459, label %b377, label %a35b
412b463:
413	%r506 = add i64 %d753, 0
414	%r511 = sext i32 %r60 to i64
415	%r512 = add i64 %r511, 0
416	%r513 = icmp slt i64 %r506, 0
417	%v1880 = add i64 %w1874, 0
418	br i1 %r513, label %b341, label %b514
419a35b465:
420	%r469 = add i64 %s794, 0
421	br label %b463
422b514:
423	%r525 = mul i64 %r17, 0
424	%r533 = add i64 %r525, 0
425	br label %b535
426b535:
427	%w1855 = phi i64 [ 0, %b514 ], [ %v1856, %b712 ]
428	%s923 = phi i64 [ 0, %b514 ], [ %r799, %b712 ]
429	%s933 = phi i64 [ %r533, %b514 ], [ %r795, %b712 ]
430	%r538 = add i64 %w1855, 0
431	%r539 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r538
432	%r540 = load float, float* %r539
433	%r551 = load float, float* bitcast ([128 x i64]* @i6000 to float*)
434	%r562 = sub i64 %s933, 0
435	%r564 = icmp slt i64 %r512, 0
436	br i1 %r564, label %b565, label %a45b
437b565:
438	%e944 = phi i64 [ 0, %b535 ], [ %r703, %a45b ]
439	%r711 = icmp slt i64 %e944, 0
440	br i1 %r711, label %a45b714, label %b712
441a45b:
442	%w1852 = phi i64 [ 0, %b535 ], [ %v1853, %a45b ]
443	%e945 = shl i64 %w1852, 0
444	%r609 = add i64 %r562, 0
445	%r703 = add i64 %e945, 0
446	%r706 = add i64 %e945, 0
447	%r708 = icmp slt i64 %r706, 0
448	%v1853 = add i64 %w1852, 0
449	br i1 %r708, label %b565, label %a45b
450b712:
451	%r795 = add i64 %rx, 0
452	%r799 = add i64 %s923, 0
453	%r802 = add i64 %w1855, 0
454	%r807 = icmp slt i64 %r802, 0
455	%v1856 = add i64 %w1855, 0
456	br i1 %r807, label %b535, label %a50b
457a45b714:
458	%r717 = add i64 %e944, 0
459	%r720 = add i64 %r717, 0
460	%r721 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r720
461	%r722 = load float, float* %r721
462	%r726 = add i64 %r717, 0
463	%r727 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r726
464	%r728 = load float, float* %r727
465	%r732 = add i64 %r717, 0
466	%r733 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r732
467	%r734 = load float, float* %r733
468	%r738 = add i64 %r717, 0
469	%r739 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r738
470	%r740 = load float, float* %r739
471	%r744 = add i64 %r717, 0
472	%r745 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r744
473	%r746 = load float, float* %r745
474	%r750 = add i64 %r717, 0
475	%r751 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r750
476	%r752 = load float, float* %r751
477	%r753 = fadd float %r752, %r746
478	%r754 = fadd float %r728, %r722
479	%r755 = fadd float %r734, %r754
480	%r756 = fadd float %r755, %r740
481	%r757 = fadd float %r753, %r756
482	%r759 = fadd float %r757, %r540
483	%r770 = add i64 %r717, 0
484	%r771 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r770
485	%r772 = load float, float* %r771
486	%r776 = add i64 %r717, 0
487	%r777 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r776
488	%r778 = load float, float* %r777
489	%r781 = fadd float %r363, %r772
490	%r782 = fadd float %r781, %r778
491	%r783 = fadd float %r551, %r782
492	br label %b712
493a57b:
494	br i1 %r335, label %a66b, label %b1086
495b820:
496	%r823 = sext i32 %r2 to i64
497	%r834 = sext i32 %r8 to i64
498	%r844 = add i64 %r16, 0
499	%r846 = sext i32 %r60 to i64
500	%r847 = add i64 %r846, 0
501	%r851 = load float, float* bitcast ([128 x i64]* @i6000 to float*)
502	%r856 = sub i64 %rx, 0
503	br label %b858
504b858:
505	%w1891 = phi i64 [ 0, %b820 ], [ %v1892, %b1016 ]
506	%s1193 = phi i64 [ 0, %b820 ], [ %r1068, %b1016 ]
507	%b1894 = mul i64 %r834, 0
508	%b1896 = shl i64 %r823, 0
509	%b1902 = mul i64 %w1891, 0
510	%s1173 = add i64 %b1902, 0
511	%r859 = add i64 %r856, 0
512	%r862 = add i64 %w1891, 0
513	%r863 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r862
514	%r864 = load float, float* %r863
515	%r868 = add i64 %w1891, 0
516	%r869 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r868
517	%r870 = load float, float* %r869
518	%r873 = sub i64 %r859, 0
519	%r876 = sub i64 %s1173, 0
520	%r878 = icmp slt i64 %r847, 0
521	br i1 %r878, label %b879, label %a53b
522b879:
523	%e1204 = phi i64 [ 0, %b858 ], [ %r1007, %a53b ]
524	%r1015 = icmp slt i64 %e1204, 0
525	br i1 %r1015, label %a53b1019q, label %b1016
526a53b1019q:
527	%b1888 = sub i64 %r846, 0
528	%b1889 = add i64 %b1888, 0
529	br label %a53b1019
530a53b:
531	%w1881 = phi i64 [ 0, %b858 ], [ %v1882, %a53b ]
532	%e1205 = shl i64 %w1881, 0
533	%r1007 = add i64 %e1205, 0
534	%r1010 = add i64 %e1205, 0
535	%r1012 = icmp slt i64 %r1010, 0
536	%v1882 = add i64 %w1881, 0
537	br i1 %r1012, label %b879, label %a53b
538b1016:
539	%r1068 = add i64 %s1193, 0
540	%r1071 = add i64 %w1891, 0
541	%r1073 = icmp slt i64 %r1071, %r844
542	%v1892 = add i64 %w1891, 0
543	br i1 %r1073, label %b858, label %a57b
544a53b1019:
545	%w1885 = phi i64 [ 0, %a53b1019q ], [ %v1886, %a53b1019 ]
546	%r1022 = add i64 %r876, 0
547	%r1024 = bitcast i8* %c2 to float*
548	%r1025 = add i64 %r1022, 0
549	%r1026 = getelementptr float, float* %r1024, i64 %r1025
550	%r1027 = load float, float* %r1026
551	%r1032 = add i64 %r873, 0
552	%r1033 = add i64 %r1032, 0
553	%r1034 = getelementptr float, float* %r1024, i64 %r1033
554	%r1035 = load float, float* %r1034
555	%r1037 = bitcast i8* %c22010 to float*
556	%r1040 = getelementptr float, float* %r1037, i64 %r1025
557	%r1044 = fadd float %r864, %r1035
558	%r1046 = fadd float %r870, %r1027
559	%r1047 = fadd float %r1044, %r1046
560	%r1048 = fadd float %r851, %r1047
561	%v1886 = add i64 %w1885, 0
562	%u1890 = icmp slt i64 %v1886, %b1889
563	br i1 %u1890, label %b1016, label %a53b1019
564a66b:
565	br i1 %r817, label %a93b, label %b1321
566b1086:
567	%r1089 = sext i32 %r2 to i64
568	%r1090 = add i64 %rx, 0
569	%r1096 = mul i64 %r9, 0
570	%r1101 = sext i32 %r8 to i64
571	%r1104 = add i64 %r1096, 0
572	%r1108 = sub i64 %r1104, 0
573	%r1110 = sext i32 %r60 to i64
574	%r1111 = add i64 %r1110, 0
575	%r1113 = sext i32 %r14 to i64
576	%r1114 = add i64 %r16, 0
577	br label %b1117
578b1117:
579	%w1915 = phi i64 [ 0, %b1086 ], [ %v1957, %b1263 ]
580	%d1353 = bitcast i64 %w1915 to i64
581	%r1120 = add i64 %s661, 0
582	%r1121 = add i64 %r1120, 0
583	%r1122 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r1121
584	%r1123 = load float, float* %r1122
585	%r1132 = bitcast i8* %c22012 to float*
586	%r1134 = getelementptr float, float* %r1132, i64 %w1915
587	%r1135 = load float, float* %r1134
588	%r1136 = fadd float %r1123, %r1135
589	%r1138 = icmp slt i64 %r1114, 0
590	br i1 %r1138, label %b1139, label %a63b
591b1139:
592	%e1364 = phi i64 [ 0, %b1117 ], [ %r1254, %a63b ]
593	%p1998 = phi i64 [ %s761, %b1117 ], [ %r1216, %a63b ]
594	%r1108p = phi i64 [ %r1108, %b1117 ], [ %r1219, %a63b ]
595	%p2004 = phi i64 [ %d1353, %b1117 ], [ %r1090, %a63b ]
596	%s1374 = phi i64 [ 0, %b1117 ], [ %r1251, %a63b ]
597	%s1384 = add i64 %r1108p, 0
598	%s1394 = add i64 %p1998, 0
599	%r1262 = icmp slt i64 %e1364, %r1114
600	br i1 %r1262, label %a63b1266q, label %b1263
601a63b1266q:
602	%b1947 = sub i64 %r1113, 0
603	%b1948 = add i64 %b1947, 0
604	br label %a63b1266
605a63b:
606	%w1904 = phi i64 [ 0, %b1117 ], [ %v1905, %a63b ]
607	%s1375 = phi i64 [ 0, %b1117 ], [ %r1251, %a63b ]
608	%b1906 = add i64 %r1089, 0
609	%b1907 = mul i64 %r1101, 0
610	%b1929 = mul i64 %w1904, 0
611	%s1395 = add i64 %b1929, 0
612	%e1365 = shl i64 %w1904, 0
613	%r1163 = add i64 %r1090, 0
614	%r1167 = add i64 %s1375, 0
615	%r1191 = add i64 %r1163, 0
616	%r1195 = add i64 %r1167, 0
617	%r1216 = add i64 %s1395, 0
618	%r1219 = add i64 %r1191, 0
619	%r1223 = add i64 %r1195, 0
620	%r1251 = add i64 %r1223, 0
621	%r1254 = add i64 %e1365, 0
622	%r1257 = add i64 %e1365, 0
623	%r1259 = icmp slt i64 %r1257, %r1114
624	%v1905 = add i64 %w1904, 0
625	br i1 %r1259, label %b1139, label %a63b
626b1263:
627	%r1306 = add i64 %d1353, 0
628	%r1308 = icmp slt i64 %r1306, %r1111
629	%v1957 = add i64 %w1915, 0
630	br i1 %r1308, label %b1117, label %a66b
631a63b1266:
632	%w1944 = phi i64 [ 0, %a63b1266q ], [ %v1945, %a63b1266 ]
633	%s1377 = phi i64 [ %s1374, %a63b1266q ], [ %r1297, %a63b1266 ]
634	%r1282 = fadd float %r1136, 0.000000e+00
635	%r1297 = add i64 %s1377, 0
636	%v1945 = add i64 %w1944, 0
637	%u1949 = icmp slt i64 %v1945, %b1948
638	br i1 %u1949, label %b1263, label %a63b1266
639a93b:
640	br i1 %r61, label %b1606, label %a97b
641b1321:
642	%r1331 = mul i64 %r17, 0
643	%r1339 = add i64 %r1331, 0
644	br label %b1342
645b1342:
646	%w1960 = phi i64 [ 0, %b1321 ], [ %v1961, %b1582 ]
647	%s1523 = phi i64 [ %r1339, %b1321 ], [ %r1587, %b1582 ]
648	%s1563 = phi i64 [ 0, %b1321 ], [ %r1591, %b1582 ]
649	%d1533 = bitcast i64 %w1960 to i64
650	%b1968 = mul i64 %w1960, 0
651	%s1543 = add i64 %b1968, 0
652	%r1345 = add i64 %s1523, 0
653	%r1348 = sub i64 %r1345, 0
654	%r1352 = add i64 %s1523, 0
655	%r1355 = sub i64 %r1352, 0
656	%r1370 = add i64 %d1533, 0
657	%r1371 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r1370
658	%r1372 = load float, float* %r1371
659	br label %a74b
660a74b:
661	%w1958 = phi i64 [ 0, %b1342 ], [ %v1959, %a74b ]
662	%r1379 = add i64 %s1543, 0
663	%r1403 = add i64 %r1355, 0
664	%r1422 = add i64 %r1348, 0
665	%r1526 = fadd float %r1372, 0.000000e+00
666	%r1573 = add i64 %w1958, 0
667	%r1581 = icmp slt i64 %r1573, 0
668	%v1959 = add i64 %w1958, 0
669	br i1 %r1581, label %a74b, label %b1582
670b1582:
671	%r1587 = add i64 %rx, 0
672	%r1591 = add i64 %s1563, 0
673	%r1596 = add i64 %d1533, 0
674	%r1601 = icmp slt i64 %r1596, 0
675	%v1961 = add i64 %w1960, 0
676	br i1 %r1601, label %b1342, label %a93b
677b1606:
678	%r1833 = add i64 %w1964, 0
679	%r1840 = icmp slt i64 %r1833, 0
680	%v1973 = add i64 %w1964, 0
681	br i1 %r1840, label %a29b173, label %a109b
682a97b:
683	%w1970 = phi i64 [ 0, %a93b ], [ %v1971, %a97b ]
684	%r1613 = add i64 %w1964, 0
685	%r1614 = mul i64 %r1613, 0
686	%r1622 = add i64 %r1614, 0
687	%r1754 = bitcast i8* %r28 to float*
688	%r1756 = getelementptr float, float* %r1754, i64 %w1970
689	%r1757 = load float, float* %r1756
690	%r1761 = add i64 %r1622, 0
691	%r1762 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r1761
692	%r1763 = load float, float* %r1762
693	%r1767 = add i64 %r1622, 0
694	%r1768 = getelementptr float, float* bitcast ([128 x i64]* @i6000 to float*), i64 %r1767
695	%r1772 = fadd float %r1763, 0.000000e+00
696	%r1773 = fadd float %r1772, 0.000000e+00
697	%r1809 = fadd float %r1757, 0.000000e+00
698	%r1810 = fadd float %r1773, %r1809
699	store float %r1810, float* %r1768
700	%r1818 = add i64 %w1970, 0
701	%r1826 = icmp slt i64 %r1818, 0
702	%v1971 = add i64 %w1970, 0
703	br i1 %r1826, label %a97b, label %b1606
704a109b:
705	ret void
706}
707