| blob | 419f25c22eb72470e72a8e781d8fb2064743551d |
1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=aarch64 < %s | FileCheck %s
4 @A = external dso_local local_unnamed_addr global [8 x [8 x i64]], align 8
5 @B = external dso_local local_unnamed_addr global [8 x [8 x i64]], align 8
6 @C = external dso_local local_unnamed_addr global [8 x [8 x i64]], align 8
8 define dso_local void @run_test() local_unnamed_addr uwtable {
9 ; CHECK-LABEL: run_test:
10 ; CHECK: // %bb.0: // %entry
11 ; CHECK-NEXT: sub sp, sp, #192
12 ; CHECK-NEXT: .cfi_def_cfa_offset 192
13 ; CHECK-NEXT: stp d15, d14, [sp, #96] // 16-byte Folded Spill
14 ; CHECK-NEXT: stp d13, d12, [sp, #112] // 16-byte Folded Spill
15 ; CHECK-NEXT: stp d11, d10, [sp, #128] // 16-byte Folded Spill
16 ; CHECK-NEXT: stp d9, d8, [sp, #144] // 16-byte Folded Spill
17 ; CHECK-NEXT: stp x22, x21, [sp, #160] // 16-byte Folded Spill
18 ; CHECK-NEXT: stp x20, x19, [sp, #176] // 16-byte Folded Spill
19 ; CHECK-NEXT: .cfi_offset w19, -8
20 ; CHECK-NEXT: .cfi_offset w20, -16
21 ; CHECK-NEXT: .cfi_offset w21, -24
22 ; CHECK-NEXT: .cfi_offset w22, -32
23 ; CHECK-NEXT: .cfi_offset b8, -40
24 ; CHECK-NEXT: .cfi_offset b9, -48
25 ; CHECK-NEXT: .cfi_offset b10, -56
26 ; CHECK-NEXT: .cfi_offset b11, -64
27 ; CHECK-NEXT: .cfi_offset b12, -72
28 ; CHECK-NEXT: .cfi_offset b13, -80
29 ; CHECK-NEXT: .cfi_offset b14, -88
30 ; CHECK-NEXT: .cfi_offset b15, -96
31 ; CHECK-NEXT: movi v1.2d, #0000000000000000
32 ; CHECK-NEXT: // implicit-def: $q6
33 ; CHECK-NEXT: mov x8, xzr
34 ; CHECK-NEXT: mov x9, xzr
35 ; CHECK-NEXT: adrp x10, B+48
36 ; CHECK-NEXT: add x10, x10, :lo12:B+48
37 ; CHECK-NEXT: adrp x11, A
38 ; CHECK-NEXT: add x11, x11, :lo12:A
39 ; CHECK-NEXT: // kill: killed $q6
40 ; CHECK-NEXT: // implicit-def: $q6
41 ; CHECK-NEXT: // implicit-def: $q0
42 ; CHECK-NEXT: // implicit-def: $q2
43 ; CHECK-NEXT: // implicit-def: $q3
44 ; CHECK-NEXT: // implicit-def: $q4
45 ; CHECK-NEXT: // implicit-def: $q5
46 ; CHECK-NEXT: // implicit-def: $q7
47 ; CHECK-NEXT: // implicit-def: $q16
48 ; CHECK-NEXT: // implicit-def: $q17
49 ; CHECK-NEXT: // implicit-def: $q10
50 ; CHECK-NEXT: // implicit-def: $q19
51 ; CHECK-NEXT: // implicit-def: $q20
52 ; CHECK-NEXT: // implicit-def: $q21
53 ; CHECK-NEXT: // implicit-def: $q22
54 ; CHECK-NEXT: // implicit-def: $q23
55 ; CHECK-NEXT: // implicit-def: $q24
56 ; CHECK-NEXT: // implicit-def: $q25
57 ; CHECK-NEXT: // implicit-def: $q27
58 ; CHECK-NEXT: // implicit-def: $q26
59 ; CHECK-NEXT: // implicit-def: $q28
60 ; CHECK-NEXT: // implicit-def: $q30
61 ; CHECK-NEXT: // implicit-def: $q18
62 ; CHECK-NEXT: // implicit-def: $q29
63 ; CHECK-NEXT: // implicit-def: $q31
64 ; CHECK-NEXT: // implicit-def: $q12
65 ; CHECK-NEXT: // implicit-def: $q13
66 ; CHECK-NEXT: // implicit-def: $q11
67 ; CHECK-NEXT: // kill: killed $q6
68 ; CHECK-NEXT: // implicit-def: $q6
69 ; CHECK-NEXT: // kill: killed $q6
70 ; CHECK-NEXT: .LBB0_1: // %for.cond1.preheader
71 ; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
72 ; CHECK-NEXT: ldr q14, [x8]
73 ; CHECK-NEXT: mov x12, xzr
74 ; CHECK-NEXT: str q18, [sp, #32] // 16-byte Folded Spill
75 ; CHECK-NEXT: ldr x14, [x12]
76 ; CHECK-NEXT: ldr q15, [x12]
77 ; CHECK-NEXT: add x7, x11, x8
78 ; CHECK-NEXT: fmov x15, d14
79 ; CHECK-NEXT: mov x16, v14.d[1]
80 ; CHECK-NEXT: ldr q18, [sp, #64] // 16-byte Folded Reload
81 ; CHECK-NEXT: fmov x18, d15
82 ; CHECK-NEXT: mov x13, v15.d[1]
83 ; CHECK-NEXT: ldr x5, [x8]
84 ; CHECK-NEXT: ldr q14, [x10], #64
85 ; CHECK-NEXT: ldr x7, [x7, #128]
86 ; CHECK-NEXT: mul x17, x15, x14
87 ; CHECK-NEXT: mov v6.16b, v0.16b
88 ; CHECK-NEXT: mov v9.16b, v27.16b
89 ; CHECK-NEXT: mov x12, v14.d[1]
90 ; CHECK-NEXT: fmov x4, d14
91 ; CHECK-NEXT: mov v27.16b, v23.16b
92 ; CHECK-NEXT: mul x1, x16, x14
93 ; CHECK-NEXT: mov v23.16b, v19.16b
94 ; CHECK-NEXT: mov v19.16b, v7.16b
95 ; CHECK-NEXT: mov v7.16b, v2.16b
96 ; CHECK-NEXT: stp q26, q31, [sp] // 32-byte Folded Spill
97 ; CHECK-NEXT: mov v31.16b, v22.16b
98 ; CHECK-NEXT: mul x0, x18, x14
99 ; CHECK-NEXT: mov v26.16b, v10.16b
100 ; CHECK-NEXT: mov v22.16b, v5.16b
101 ; CHECK-NEXT: fmov d15, x17
102 ; CHECK-NEXT: mov v5.16b, v1.16b
103 ; CHECK-NEXT: mov v8.16b, v20.16b
104 ; CHECK-NEXT: mul x2, x13, x14
105 ; CHECK-NEXT: mov v20.16b, v16.16b
106 ; CHECK-NEXT: mov v16.16b, v3.16b
107 ; CHECK-NEXT: mov v10.16b, v21.16b
108 ; CHECK-NEXT: mov v21.16b, v17.16b
109 ; CHECK-NEXT: mov v17.16b, v4.16b
110 ; CHECK-NEXT: mov v15.d[1], x1
111 ; CHECK-NEXT: mul x3, x12, x14
112 ; CHECK-NEXT: add x8, x8, #8
113 ; CHECK-NEXT: fmov d14, x0
114 ; CHECK-NEXT: cmp x8, #64
115 ; CHECK-NEXT: add x9, x9, #1
116 ; CHECK-NEXT: mul x14, x4, x14
117 ; CHECK-NEXT: add v18.2d, v18.2d, v15.2d
118 ; CHECK-NEXT: mul x19, x15, x5
119 ; CHECK-NEXT: mov v14.d[1], x2
120 ; CHECK-NEXT: mul x15, x15, x7
121 ; CHECK-NEXT: fmov d0, x14
122 ; CHECK-NEXT: str q18, [sp, #64] // 16-byte Folded Spill
123 ; CHECK-NEXT: ldp q18, q15, [sp, #32] // 32-byte Folded Reload
124 ; CHECK-NEXT: mul x6, x16, x5
125 ; CHECK-NEXT: fmov d1, x19
126 ; CHECK-NEXT: mov v0.d[1], x3
127 ; CHECK-NEXT: mul x16, x16, x7
128 ; CHECK-NEXT: fmov d2, x15
129 ; CHECK-NEXT: add v15.2d, v15.2d, v14.2d
130 ; CHECK-NEXT: mul x21, x18, x7
131 ; CHECK-NEXT: mov v1.d[1], x6
132 ; CHECK-NEXT: mul x0, x4, x7
133 ; CHECK-NEXT: str q15, [sp, #48] // 16-byte Folded Spill
134 ; CHECK-NEXT: add v15.2d, v11.2d, v14.2d
135 ; CHECK-NEXT: mov v2.d[1], x16
136 ; CHECK-NEXT: ldr q11, [sp, #80] // 16-byte Folded Reload
137 ; CHECK-NEXT: mul x20, x13, x7
138 ; CHECK-NEXT: fmov d3, x21
139 ; CHECK-NEXT: add v11.2d, v11.2d, v0.2d
140 ; CHECK-NEXT: add v12.2d, v12.2d, v1.2d
141 ; CHECK-NEXT: mul x22, x12, x7
142 ; CHECK-NEXT: fmov d4, x0
143 ; CHECK-NEXT: add v18.2d, v18.2d, v2.2d
144 ; CHECK-NEXT: mov v2.16b, v7.16b
145 ; CHECK-NEXT: mul x14, x18, x5
146 ; CHECK-NEXT: mov v7.16b, v19.16b
147 ; CHECK-NEXT: mov v19.16b, v23.16b
148 ; CHECK-NEXT: mov v3.d[1], x20
149 ; CHECK-NEXT: mov v23.16b, v27.16b
150 ; CHECK-NEXT: mov v27.16b, v9.16b
151 ; CHECK-NEXT: mul x15, x4, x5
152 ; CHECK-NEXT: add v27.2d, v9.2d, v1.2d
153 ; CHECK-NEXT: str q11, [sp, #80] // 16-byte Folded Spill
154 ; CHECK-NEXT: mov v4.d[1], x22
155 ; CHECK-NEXT: add v19.2d, v19.2d, v1.2d
156 ; CHECK-NEXT: add v7.2d, v7.2d, v1.2d
157 ; CHECK-NEXT: mul x13, x13, x5
158 ; CHECK-NEXT: add v23.2d, v23.2d, v1.2d
159 ; CHECK-NEXT: add v1.2d, v5.2d, v1.2d
160 ; CHECK-NEXT: fmov d14, x14
161 ; CHECK-NEXT: add v30.2d, v30.2d, v3.2d
162 ; CHECK-NEXT: mov v3.16b, v16.16b
163 ; CHECK-NEXT: mul x12, x12, x5
164 ; CHECK-NEXT: mov v16.16b, v20.16b
165 ; CHECK-NEXT: mov v5.16b, v22.16b
166 ; CHECK-NEXT: fmov d0, x15
167 ; CHECK-NEXT: add v28.2d, v28.2d, v4.2d
168 ; CHECK-NEXT: mov v4.16b, v17.16b
169 ; CHECK-NEXT: mov v17.16b, v21.16b
170 ; CHECK-NEXT: mov v21.16b, v10.16b
171 ; CHECK-NEXT: mov v10.16b, v26.16b
172 ; CHECK-NEXT: mov v14.d[1], x13
173 ; CHECK-NEXT: mov v22.16b, v31.16b
174 ; CHECK-NEXT: mov v20.16b, v8.16b
175 ; CHECK-NEXT: ldp q26, q31, [sp] // 32-byte Folded Reload
176 ; CHECK-NEXT: mov v11.16b, v15.16b
177 ; CHECK-NEXT: mov v0.d[1], x12
178 ; CHECK-NEXT: add v13.2d, v13.2d, v14.2d
179 ; CHECK-NEXT: add v31.2d, v31.2d, v14.2d
180 ; CHECK-NEXT: add v26.2d, v26.2d, v14.2d
181 ; CHECK-NEXT: add v24.2d, v24.2d, v14.2d
182 ; CHECK-NEXT: add v22.2d, v22.2d, v14.2d
183 ; CHECK-NEXT: add v20.2d, v8.2d, v14.2d
184 ; CHECK-NEXT: add v10.2d, v10.2d, v14.2d
185 ; CHECK-NEXT: add v16.2d, v16.2d, v14.2d
186 ; CHECK-NEXT: add v5.2d, v5.2d, v14.2d
187 ; CHECK-NEXT: add v3.2d, v3.2d, v14.2d
188 ; CHECK-NEXT: add v2.2d, v2.2d, v14.2d
189 ; CHECK-NEXT: add v29.2d, v29.2d, v0.2d
190 ; CHECK-NEXT: add v25.2d, v25.2d, v0.2d
191 ; CHECK-NEXT: add v21.2d, v21.2d, v0.2d
192 ; CHECK-NEXT: add v17.2d, v17.2d, v0.2d
193 ; CHECK-NEXT: add v4.2d, v4.2d, v0.2d
194 ; CHECK-NEXT: add v0.2d, v6.2d, v0.2d
195 ; CHECK-NEXT: b.ne .LBB0_1
196 ; CHECK-NEXT: // %bb.2: // %for.cond.cleanup
197 ; CHECK-NEXT: ldr q6, [sp, #48] // 16-byte Folded Reload
198 ; CHECK-NEXT: adrp x8, C
199 ; CHECK-NEXT: add x8, x8, :lo12:C
200 ; CHECK-NEXT: stp q12, q31, [x8, #80]
201 ; CHECK-NEXT: ldp x20, x19, [sp, #176] // 16-byte Folded Reload
202 ; CHECK-NEXT: str q6, [x8]
203 ; CHECK-NEXT: ldr q6, [sp, #64] // 16-byte Folded Reload
204 ; CHECK-NEXT: str q29, [x8, #112]
205 ; CHECK-NEXT: ldp x22, x21, [sp, #160] // 16-byte Folded Reload
206 ; CHECK-NEXT: stp q6, q11, [x8, #16]
207 ; CHECK-NEXT: ldr q6, [sp, #80] // 16-byte Folded Reload
208 ; CHECK-NEXT: stp q18, q30, [x8, #144]
209 ; CHECK-NEXT: ldp d9, d8, [sp, #144] // 16-byte Folded Reload
210 ; CHECK-NEXT: stp q6, q13, [x8, #48]
211 ; CHECK-NEXT: ldp d13, d12, [sp, #112] // 16-byte Folded Reload
212 ; CHECK-NEXT: stp q28, q26, [x8, #176]
213 ; CHECK-NEXT: ldp d15, d14, [sp, #96] // 16-byte Folded Reload
214 ; CHECK-NEXT: stp q19, q10, [x8, #336]
215 ; CHECK-NEXT: ldp d11, d10, [sp, #128] // 16-byte Folded Reload
216 ; CHECK-NEXT: str q27, [x8, #208]
217 ; CHECK-NEXT: stp q25, q24, [x8, #240]
218 ; CHECK-NEXT: stp q23, q22, [x8, #272]
219 ; CHECK-NEXT: stp q21, q20, [x8, #304]
220 ; CHECK-NEXT: stp q17, q16, [x8, #368]
221 ; CHECK-NEXT: stp q7, q5, [x8, #400]
222 ; CHECK-NEXT: stp q4, q3, [x8, #432]
223 ; CHECK-NEXT: stp q1, q2, [x8, #464]
224 ; CHECK-NEXT: str q0, [x8, #496]
225 ; CHECK-NEXT: add sp, sp, #192
226 ; CHECK-NEXT: .cfi_def_cfa_offset 0
227 ; CHECK-NEXT: .cfi_restore w19
228 ; CHECK-NEXT: .cfi_restore w20
229 ; CHECK-NEXT: .cfi_restore w21
230 ; CHECK-NEXT: .cfi_restore w22
231 ; CHECK-NEXT: .cfi_restore b8
232 ; CHECK-NEXT: .cfi_restore b9
233 ; CHECK-NEXT: .cfi_restore b10
234 ; CHECK-NEXT: .cfi_restore b11
235 ; CHECK-NEXT: .cfi_restore b12
236 ; CHECK-NEXT: .cfi_restore b13
237 ; CHECK-NEXT: .cfi_restore b14
238 ; CHECK-NEXT: .cfi_restore b15
239 ; CHECK-NEXT: ret
240 ; CH`ECK-NEXT: .cfi_offset b9, -16
241 entry:
242 br label %for.cond1.preheader
244 for.cond1.preheader: ; preds = %for.cond1.preheader, %entry
245 %0 = phi <2 x i64> [ undef, %entry ], [ %118, %for.cond1.preheader ]
246 %1 = phi <2 x i64> [ undef, %entry ], [ %116, %for.cond1.preheader ]
247 %2 = phi <2 x i64> [ zeroinitializer, %entry ], [ %114, %for.cond1.preheader ]
248 %3 = phi <2 x i64> [ undef, %entry ], [ %112, %for.cond1.preheader ]
249 %4 = phi <2 x i64> [ undef, %entry ], [ %107, %for.cond1.preheader ]
250 %5 = phi <2 x i64> [ undef, %entry ], [ %105, %for.cond1.preheader ]
251 %6 = phi <2 x i64> [ undef, %entry ], [ %103, %for.cond1.preheader ]
252 %7 = phi <2 x i64> [ undef, %entry ], [ %101, %for.cond1.preheader ]
253 %8 = phi <2 x i64> [ undef, %entry ], [ %96, %for.cond1.preheader ]
254 %9 = phi <2 x i64> [ undef, %entry ], [ %94, %for.cond1.preheader ]
255 %10 = phi <2 x i64> [ undef, %entry ], [ %92, %for.cond1.preheader ]
256 %11 = phi <2 x i64> [ undef, %entry ], [ %90, %for.cond1.preheader ]
257 %12 = phi <2 x i64> [ undef, %entry ], [ %85, %for.cond1.preheader ]
258 %13 = phi <2 x i64> [ undef, %entry ], [ %83, %for.cond1.preheader ]
259 %14 = phi <2 x i64> [ undef, %entry ], [ %81, %for.cond1.preheader ]
260 %15 = phi <2 x i64> [ undef, %entry ], [ %79, %for.cond1.preheader ]
261 %16 = phi <2 x i64> [ undef, %entry ], [ %74, %for.cond1.preheader ]
262 %17 = phi <2 x i64> [ undef, %entry ], [ %72, %for.cond1.preheader ]
263 %18 = phi <2 x i64> [ undef, %entry ], [ %70, %for.cond1.preheader ]
264 %19 = phi <2 x i64> [ undef, %entry ], [ %65, %for.cond1.preheader ]
265 %20 = phi <2 x i64> [ undef, %entry ], [ %63, %for.cond1.preheader ]
266 %21 = phi <2 x i64> [ undef, %entry ], [ %61, %for.cond1.preheader ]
267 %22 = phi <2 x i64> [ undef, %entry ], [ %56, %for.cond1.preheader ]
268 %23 = phi <2 x i64> [ undef, %entry ], [ %54, %for.cond1.preheader ]
269 %24 = phi <2 x i64> [ undef, %entry ], [ %52, %for.cond1.preheader ]
270 %25 = phi <2 x i64> [ undef, %entry ], [ %50, %for.cond1.preheader ]
271 %26 = phi <2 x i64> [ undef, %entry ], [ %45, %for.cond1.preheader ]
272 %27 = phi <2 x i64> [ undef, %entry ], [ %43, %for.cond1.preheader ]
273 %28 = phi <2 x i64> [ undef, %entry ], [ %41, %for.cond1.preheader ]
274 %29 = phi <2 x i64> [ undef, %entry ], [ %39, %for.cond1.preheader ]
275 %indvars.iv40 = phi i64 [ 0, %entry ], [ %indvars.iv.next41, %for.cond1.preheader ]
276 %30 = load <2 x i64>, <2 x i64>* null, align 8
277 %31 = load <2 x i64>, <2 x i64>* undef, align 8
278 %arrayidx14.4.phi.trans.insert = getelementptr inbounds [8 x [8 x i64]], [8 x [8 x i64]]* @B, i64 0, i64 %indvars.iv40, i64 4
279 %32 = load <2 x i64>, <2 x i64>* null, align 8
280 %arrayidx14.6.phi.trans.insert = getelementptr inbounds [8 x [8 x i64]], [8 x [8 x i64]]* @B, i64 0, i64 %indvars.iv40, i64 6
281 %33 = bitcast i64* %arrayidx14.6.phi.trans.insert to <2 x i64>*
282 %34 = load <2 x i64>, <2 x i64>* %33, align 8
283 %35 = load i64, i64* null, align 8
284 %36 = insertelement <2 x i64> undef, i64 %35, i32 0
285 %37 = shufflevector <2 x i64> %36, <2 x i64> undef, <2 x i32> zeroinitializer
286 %38 = mul nsw <2 x i64> %30, %37
287 %39 = add nsw <2 x i64> %29, %38
288 %40 = mul nsw <2 x i64> %31, %37
289 %41 = add nsw <2 x i64> %28, %40
290 %42 = mul nsw <2 x i64> %32, %37
291 %43 = add nsw <2 x i64> %27, %42
292 %44 = mul nsw <2 x i64> %34, %37
293 %45 = add nsw <2 x i64> %26, %44
294 %46 = load i64, i64* undef, align 8
295 %47 = insertelement <2 x i64> undef, i64 %46, i32 0
296 %48 = shufflevector <2 x i64> %47, <2 x i64> undef, <2 x i32> zeroinitializer
297 %49 = mul nsw <2 x i64> %30, %48
298 %50 = add nsw <2 x i64> %25, %49
299 %51 = mul nsw <2 x i64> %31, %48
300 %52 = add nsw <2 x i64> %24, %51
301 %53 = mul nsw <2 x i64> %32, %48
302 %54 = add nsw <2 x i64> %23, %53
303 %55 = mul nsw <2 x i64> %34, %48
304 %56 = add nsw <2 x i64> %22, %55
305 %arrayidx10.2 = getelementptr inbounds [8 x [8 x i64]], [8 x [8 x i64]]* @A, i64 0, i64 2, i64 %indvars.iv40
306 %57 = load i64, i64* %arrayidx10.2, align 8
307 %58 = insertelement <2 x i64> undef, i64 %57, i32 0
308 %59 = shufflevector <2 x i64> %58, <2 x i64> undef, <2 x i32> zeroinitializer
309 %60 = mul nsw <2 x i64> %31, %59
310 %61 = add nsw <2 x i64> %21, %60
311 %62 = mul nsw <2 x i64> %32, %59
312 %63 = add nsw <2 x i64> %20, %62
313 %64 = mul nsw <2 x i64> %34, %59
314 %65 = add nsw <2 x i64> %19, %64
315 %66 = load i64, i64* undef, align 8
316 %67 = insertelement <2 x i64> undef, i64 %66, i32 0
317 %68 = shufflevector <2 x i64> %67, <2 x i64> undef, <2 x i32> zeroinitializer
318 %69 = mul nsw <2 x i64> %30, %68
319 %70 = add nsw <2 x i64> %18, %69
320 %71 = mul nsw <2 x i64> %31, %68
321 %72 = add nsw <2 x i64> %17, %71
322 %73 = mul nsw <2 x i64> %34, %68
323 %74 = add nsw <2 x i64> %16, %73
324 %75 = load i64, i64* undef, align 8
325 %76 = insertelement <2 x i64> undef, i64 %75, i32 0
326 %77 = shufflevector <2 x i64> %76, <2 x i64> undef, <2 x i32> zeroinitializer
327 %78 = mul nsw <2 x i64> %30, %77
328 %79 = add nsw <2 x i64> %15, %78
329 %80 = mul nsw <2 x i64> %31, %77
330 %81 = add nsw <2 x i64> %14, %80
331 %82 = mul nsw <2 x i64> %32, %77
332 %83 = add nsw <2 x i64> %13, %82
333 %84 = mul nsw <2 x i64> %34, %77
334 %85 = add nsw <2 x i64> %12, %84
335 %86 = load i64, i64* undef, align 8
336 %87 = insertelement <2 x i64> undef, i64 %86, i32 0
337 %88 = shufflevector <2 x i64> %87, <2 x i64> undef, <2 x i32> zeroinitializer
338 %89 = mul nsw <2 x i64> %30, %88
339 %90 = add nsw <2 x i64> %11, %89
340 %91 = mul nsw <2 x i64> %31, %88
341 %92 = add nsw <2 x i64> %10, %91
342 %93 = mul nsw <2 x i64> %32, %88
343 %94 = add nsw <2 x i64> %9, %93
344 %95 = mul nsw <2 x i64> %34, %88
345 %96 = add nsw <2 x i64> %8, %95
346 %97 = load i64, i64* undef, align 8
347 %98 = insertelement <2 x i64> undef, i64 %97, i32 0
348 %99 = shufflevector <2 x i64> %98, <2 x i64> undef, <2 x i32> zeroinitializer
349 %100 = mul nsw <2 x i64> %30, %99
350 %101 = add nsw <2 x i64> %7, %100
351 %102 = mul nsw <2 x i64> %31, %99
352 %103 = add nsw <2 x i64> %6, %102
353 %104 = mul nsw <2 x i64> %32, %99
354 %105 = add nsw <2 x i64> %5, %104
355 %106 = mul nsw <2 x i64> %34, %99
356 %107 = add nsw <2 x i64> %4, %106
357 %108 = load i64, i64* undef, align 8
358 %109 = insertelement <2 x i64> undef, i64 %108, i32 0
359 %110 = shufflevector <2 x i64> %109, <2 x i64> undef, <2 x i32> zeroinitializer
360 %111 = mul nsw <2 x i64> %30, %110
361 %112 = add nsw <2 x i64> %3, %111
362 %113 = mul nsw <2 x i64> %31, %110
363 %114 = add nsw <2 x i64> %2, %113
364 %115 = mul nsw <2 x i64> %32, %110
365 %116 = add nsw <2 x i64> %1, %115
366 %117 = mul nsw <2 x i64> %34, %110
367 %118 = add nsw <2 x i64> %0, %117
368 %indvars.iv.next41 = add nuw nsw i64 %indvars.iv40, 1
369 %exitcond42 = icmp eq i64 %indvars.iv.next41, 8
370 br i1 %exitcond42, label %for.cond.cleanup, label %for.cond1.preheader
372 for.cond.cleanup: ; preds = %for.cond1.preheader
373 store <2 x i64> %39, <2 x i64>* bitcast ([8 x [8 x i64]]* @C to <2 x i64>*), align 8
374 store <2 x i64> %41, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 0, i64 2) to <2 x i64>*), align 8
375 store <2 x i64> %43, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 0, i64 4) to <2 x i64>*), align 8
376 store <2 x i64> %45, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 0, i64 6) to <2 x i64>*), align 8
377 store <2 x i64> %50, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 1, i64 0) to <2 x i64>*), align 8
378 store <2 x i64> %52, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 1, i64 2) to <2 x i64>*), align 8
379 store <2 x i64> %54, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 1, i64 4) to <2 x i64>*), align 8
380 store <2 x i64> %56, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 1, i64 6) to <2 x i64>*), align 8
381 store <2 x i64> %61, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 2, i64 2) to <2 x i64>*), align 8
382 store <2 x i64> %63, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 2, i64 4) to <2 x i64>*), align 8
383 store <2 x i64> %65, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 2, i64 6) to <2 x i64>*), align 8
384 store <2 x i64> %70, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 3, i64 0) to <2 x i64>*), align 8
385 store <2 x i64> %72, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 3, i64 2) to <2 x i64>*), align 8
386 store <2 x i64> %74, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 3, i64 6) to <2 x i64>*), align 8
387 store <2 x i64> %79, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 4, i64 0) to <2 x i64>*), align 8
388 store <2 x i64> %81, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 4, i64 2) to <2 x i64>*), align 8
389 store <2 x i64> %83, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 4, i64 4) to <2 x i64>*), align 8
390 store <2 x i64> %85, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 4, i64 6) to <2 x i64>*), align 8
391 store <2 x i64> %90, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 5, i64 0) to <2 x i64>*), align 8
392 store <2 x i64> %92, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 5, i64 2) to <2 x i64>*), align 8
393 store <2 x i64> %94, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 5, i64 4) to <2 x i64>*), align 8
394 store <2 x i64> %96, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 5, i64 6) to <2 x i64>*), align 8
395 store <2 x i64> %101, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 6, i64 0) to <2 x i64>*), align 8
396 store <2 x i64> %103, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 6, i64 2) to <2 x i64>*), align 8
397 store <2 x i64> %105, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 6, i64 4) to <2 x i64>*), align 8
398 store <2 x i64> %107, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 6, i64 6) to <2 x i64>*), align 8
399 store <2 x i64> %112, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 7, i64 0) to <2 x i64>*), align 8
400 store <2 x i64> %114, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 7, i64 2) to <2 x i64>*), align 8
401 store <2 x i64> %116, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 7, i64 4) to <2 x i64>*), align 8
402 store <2 x i64> %118, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x [8 x i64]], [8 x [8 x i64]]* @C, i64 0, i64 7, i64 6) to <2 x i64>*), align 8
403 ret void
404 }