1 ; RUN: llc -verify-machineinstrs < %s -mcpu=g5 | FileCheck %s
2 target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-f128:128:128-v128:128:128-n32:64"
3 target triple = "powerpc64-unknown-linux-gnu"
5 define void @foo(float* noalias nocapture %a, float* noalias nocapture %b) #0 {
9 vector.body: ; preds = %vector.body, %vector.ph
10 %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
11 %0 = getelementptr inbounds float, float* %b, i64 %index
12 %1 = bitcast float* %0 to <4 x float>*
13 %wide.load = load <4 x float>, <4 x float>* %1, align 4
14 %.sum11 = or i64 %index, 4
15 %2 = getelementptr float, float* %b, i64 %.sum11
16 %3 = bitcast float* %2 to <4 x float>*
17 %wide.load8 = load <4 x float>, <4 x float>* %3, align 4
18 %4 = fadd <4 x float> %wide.load, <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>
19 %5 = fadd <4 x float> %wide.load8, <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>
20 %6 = getelementptr inbounds float, float* %a, i64 %index
21 %7 = bitcast float* %6 to <4 x float>*
22 store <4 x float> %4, <4 x float>* %7, align 4
23 %.sum12 = or i64 %index, 4
24 %8 = getelementptr float, float* %a, i64 %.sum12
25 %9 = bitcast float* %8 to <4 x float>*
26 store <4 x float> %5, <4 x float>* %9, align 4
27 %index.next = add i64 %index, 8
28 %10 = icmp eq i64 %index.next, 16000
29 br i1 %10, label %for.end, label %vector.body
32 ; CHECK-DAG: li [[C16:[0-9]+]], 16
33 ; CHECK-DAG: lvx [[CNST:[0-9]+]],
35 ; CHECK-DAG: lvx [[LD1:[0-9]+]], 0, [[C0:[0-9]+]]
36 ; CHECK-DAG: lvx [[LD2:[0-9]+]], [[C0]], [[C16]]
37 ; CHECK-DAG: lvsl [[MASK1:[0-9]+]], 0, [[C0]]
38 ; CHECK-DAG: vperm [[VR1:[0-9]+]], [[LD1]], [[LD2]], [[MASK1]]
39 ; CHECK-DAG: vaddfp {{[0-9]+}}, [[VR1]], [[CNST]]
42 for.end: ; preds = %vector.body
46 attributes #0 = { nounwind }