1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-linux-gnu -mcpu=haswell | FileCheck %s
4 define void @WhileWithLoopInvariantOperation.21() {
5 ; CHECK-LABEL: WhileWithLoopInvariantOperation.21:
6 ; CHECK: # %bb.0: # %while.1.body.preheader
7 ; CHECK-NEXT: movq (%rax), %rax
8 ; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0
9 ; CHECK-NEXT: vmovaps %xmm0, 32(%rax)
10 ; CHECK-NEXT: vmovsd {{.*#+}} xmm0 = [4294967295,4294967295,0,0]
11 ; CHECK-NEXT: vmaskmovps %ymm0, %ymm0, (%rax)
12 while.1.body.preheader:
13 %0 = load ptr, ptr undef, align 8, !invariant.load !0, !dereferenceable !1, !align !2
14 %1 = getelementptr inbounds i8, ptr %0, i64 32
15 tail call void @llvm.memset.p0.i64(ptr nonnull align 16 dereferenceable(16) %1, i8 0, i64 16, i1 false)
16 %2 = tail call <8 x float> @llvm.masked.load.v8f32.p0(ptr undef, i32 4, <8 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <8 x float> <float undef, float undef, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00>)
17 %3 = tail call <16 x float> @llvm.masked.load.v16f32.p0(ptr undef, i32 4, <16 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x float> <float undef, float undef, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00>)
18 tail call void @llvm.masked.store.v16f32.p0(<16 x float> undef, ptr nonnull undef, i32 4, <16 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>)
22 declare void @llvm.memset.p0.i64(ptr nocapture writeonly, i8, i64, i1 immarg)
24 declare <8 x float> @llvm.masked.load.v8f32.p0(ptr, i32 immarg, <8 x i1>, <8 x float>)
26 declare <16 x float> @llvm.masked.load.v16f32.p0(ptr, i32 immarg, <16 x i1>, <16 x float>)
28 declare void @llvm.masked.store.v16f32.p0(<16 x float>, ptr, i32 immarg, <16 x i1>)