1; RUN: llc -verify-machineinstrs < %s -mcpu=g5 | FileCheck %s 2target 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" 3target triple = "powerpc64-unknown-linux-gnu" 4 5define void @foo(ptr noalias nocapture %a, ptr noalias nocapture %b) #0 { 6vector.ph: 7 br label %vector.body 8 9vector.body: ; preds = %vector.body, %vector.ph 10 %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ] 11 %0 = getelementptr inbounds float, ptr %b, i64 %index 12 %wide.load = load <4 x float>, ptr %0, align 4 13 %.sum11 = or i64 %index, 4 14 %1 = getelementptr float, ptr %b, i64 %.sum11 15 %wide.load8 = load <4 x float>, ptr %1, align 4 16 %2 = fadd <4 x float> %wide.load, <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00> 17 %3 = fadd <4 x float> %wide.load8, <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00> 18 %4 = getelementptr inbounds float, ptr %a, i64 %index 19 store <4 x float> %2, ptr %4, align 4 20 %.sum12 = or i64 %index, 4 21 %5 = getelementptr float, ptr %a, i64 %.sum12 22 store <4 x float> %3, ptr %5, align 4 23 %index.next = add i64 %index, 8 24 %6 = icmp eq i64 %index.next, 16000 25 br i1 %6, label %for.end, label %vector.body 26 27; CHECK: @foo 28; CHECK-DAG: li [[C0:[0-9]+]], 0 29; CHECK-DAG: lvx [[CNST:[0-9]+]], 30; CHECK: .LBB0_1: 31; CHECK-DAG: lvsl [[MASK1:[0-9]+]], [[B1:[0-9]+]], [[C0]] 32; CHECK-DAG: add [[B3:[0-9]+]], [[B1]], [[C0]] 33; CHECK-DAG: lvx [[LD1:[0-9]+]], [[B1]], [[C0]] 34; CHECK-DAG: lvx [[LD2:[0-9]+]], [[B3]], 35; CHECK-DAG: vperm [[R1:[0-9]+]], [[LD1]], [[LD2]], [[MASK1]] 36; CHECK-DAG: vaddfp {{[0-9]+}}, [[R1]], [[CNST]] 37; CHECK: blr 38 39for.end: ; preds = %vector.body 40 ret void 41} 42 43attributes #0 = { nounwind } 44