1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-26 04:32:44 +01:00
llvm-mirror/test/CodeGen/AArch64/arm64-vext_reverse.ll
Mikhail Maltsev 63df72fb38 [AArch64] Optimize instruction selection for certain vector shuffles
This patch adds code to recognize vector shuffles which can be
represented as VDUP (splat) of a vector lane with of a different
(wider) type than the original vector lane type.

For example:
    shufflevector <4 x i16> %v, <4 x i16> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
is essentially:
    shufflevector <2 x i32> %v, <2 x i32> undef, <2 x i32> <i32 0, i32 0>

Such patterns are generated by the SelectionDAG machinery in some cases
(see DAGCombiner::visitBITCAST in DAGCombiner.cpp, the "Remove double
bitcasts from shuffles" part).

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D86225
2020-08-27 11:06:49 +01:00

173 lines
5.4 KiB
LLVM

; RUN: llc -mtriple=arm64-linux-gnuabi < %s | FileCheck %s
; The following tests is to check the correctness of reversing input operand
; of vext by enumerating all cases of using two undefs in shuffle masks.
define <4 x i16> @vext_6701_0(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_0:
; CHECK: ext v0.8b, v1.8b, v0.8b, #4
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_12(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_12:
; CHECK: dup v0.2s, v0.s[0]
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 undef, i32 0, i32 1>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_13(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_13:
; CHECK: ext v0.8b, v1.8b, v0.8b, #4
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 7, i32 undef, i32 1>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_14(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_14:
; CHECK: ext v0.8b, v1.8b, v0.8b, #4
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 7, i32 0, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_23(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_23:
; CHECK: ext v0.8b, v1.8b, v0.8b, #4
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 6, i32 undef, i32 undef, i32 1>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_24(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_24:
; CHECK: ext v0.8b, v1.8b, v0.8b, #4
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 6, i32 undef, i32 0, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_6701_34(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_6701_34:
; CHECK: dup v0.2s, v1.s[1]
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 6, i32 7, i32 undef, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_0(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_0:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 5, i32 6, i32 7, i32 0>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_12(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_12:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 undef, i32 7, i32 0>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_13(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_13:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 6, i32 undef, i32 0>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_14(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_14:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 6, i32 7, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_23(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_23:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 5, i32 undef, i32 undef, i32 0>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_24(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_24:
; CHECK: rev32 v0.4h, v1.4h
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 5, i32 undef, i32 7, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_5670_34(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_5670_34:
; CHECK: ext v0.8b, v1.8b, v0.8b, #2
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 5, i32 6, i32 undef, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_0(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_0:
; CHECK: ext v0.8b, v1.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 7, i32 0, i32 1, i32 2>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_12(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_12:
; CHECK: ext v0.8b, v0.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 undef, i32 1, i32 2>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_13(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_13:
; CHECK: rev32 v0.4h, v0.4h
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 0, i32 undef, i32 2>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_14(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_14:
; CHECK: ext v0.8b, v0.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 undef, i32 0, i32 1, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_23(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_23:
; CHECK: ext v0.8b, v1.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 7, i32 undef, i32 undef, i32 2>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_24(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_24:
; CHECK: ext v0.8b, v1.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 7, i32 undef, i32 1, i32 undef>
ret <4 x i16> %x
}
define <4 x i16> @vext_7012_34(<4 x i16> %a1, <4 x i16> %a2) {
entry:
; CHECK-LABEL: vext_7012_34:
; CHECK: ext v0.8b, v1.8b, v0.8b, #6
%x = shufflevector <4 x i16> %a1, <4 x i16> %a2, <4 x i32> <i32 7, i32 0, i32 undef, i32 undef>
ret <4 x i16> %x
}