xref: /llvm-project/llvm/test/CodeGen/X86/vec_shuf-insert.ll (revision 33c283adfd0b157af7edb2bcde065248ca3b8c22)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=x86_64-unknown-linux -mcpu=corei7-avx | FileCheck %s
3
4; These tests check that an insert_subvector which replaces one of the halves
5; of a concat_vectors is optimized into a single vinsertf128.
6
7
8declare <8 x float> @llvm.x86.avx.vinsertf128.ps.256(<8 x float>, <4 x float>, i8)
9
10define <8 x float> @lower_half(<4 x float> %v1, <4 x float> %v2, <4 x float> %v3) {
11; CHECK-LABEL: lower_half:
12; CHECK:       # %bb.0:
13; CHECK-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
14; CHECK-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm0
15; CHECK-NEXT:    retq
16  %1 = shufflevector <4 x float> %v1, <4 x float> %v2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
17  %2 = tail call <8 x float> @llvm.x86.avx.vinsertf128.ps.256(<8 x float> %1, <4 x float> %v3, i8 0)
18  ret <8 x float> %2
19
20}
21
22define <8 x float> @upper_half(<4 x float> %v1, <4 x float> %v2, <4 x float> %v3) {
23; CHECK-LABEL: upper_half:
24; CHECK:       # %bb.0:
25; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
26; CHECK-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
27; CHECK-NEXT:    retq
28  %1 = shufflevector <4 x float> %v1, <4 x float> %v2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
29  %2 = tail call <8 x float> @llvm.x86.avx.vinsertf128.ps.256(<8 x float> %1, <4 x float> %v3, i8 1)
30  ret <8 x float> %2
31
32}
33