1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+avx | FileCheck %s --check-prefix=AVX
3 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+avx2 | FileCheck %s --check-prefix=AVX
5 ; Prefer a blend instruction to a vinsert128 instruction because blends
6 ; are simpler (no lane changes) and therefore will have equal or better
9 define <8 x float> @castA(<4 x float> %m) nounwind uwtable readnone ssp {
12 ; AVX-NEXT: vmovaps %xmm0, %xmm0
14 %shuffle.i = shufflevector <4 x float> %m, <4 x float> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 4, i32 4, i32 4>
15 ret <8 x float> %shuffle.i
18 define <4 x double> @castB(<2 x double> %m) nounwind uwtable readnone ssp {
21 ; AVX-NEXT: vmovaps %xmm0, %xmm0
23 %shuffle.i = shufflevector <2 x double> %m, <2 x double> zeroinitializer, <4 x i32> <i32 0, i32 1, i32 2, i32 2>
24 ret <4 x double> %shuffle.i
27 ; AVX2 is needed for integer types.
29 define <4 x i64> @castC(<2 x i64> %m) nounwind uwtable readnone ssp {
32 ; AVX-NEXT: vmovaps %xmm0, %xmm0
34 %shuffle.i = shufflevector <2 x i64> %m, <2 x i64> zeroinitializer, <4 x i32> <i32 0, i32 1, i32 2, i32 2>
35 ret <4 x i64> %shuffle.i
38 ; The next three tests don't need any shuffling. There may or may not be a
39 ; vzeroupper before the return, so just check for the absence of shuffles.
41 define <4 x float> @castD(<8 x float> %m) nounwind uwtable readnone ssp {
44 ; AVX-NEXT: ## kill: def $xmm0 killed $xmm0 killed $ymm0
45 ; AVX-NEXT: vzeroupper
47 %shuffle.i = shufflevector <8 x float> %m, <8 x float> %m, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
48 ret <4 x float> %shuffle.i
51 define <2 x i64> @castE(<4 x i64> %m) nounwind uwtable readnone ssp {
54 ; AVX-NEXT: ## kill: def $xmm0 killed $xmm0 killed $ymm0
55 ; AVX-NEXT: vzeroupper
57 %shuffle.i = shufflevector <4 x i64> %m, <4 x i64> %m, <2 x i32> <i32 0, i32 1>
58 ret <2 x i64> %shuffle.i
61 define <2 x double> @castF(<4 x double> %m) nounwind uwtable readnone ssp {
64 ; AVX-NEXT: ## kill: def $xmm0 killed $xmm0 killed $ymm0
65 ; AVX-NEXT: vzeroupper
67 %shuffle.i = shufflevector <4 x double> %m, <4 x double> %m, <2 x i32> <i32 0, i32 1>
68 ret <2 x double> %shuffle.i