mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-25 12:12:47 +01:00
6db76aaf10
This is still a work in progress but most of the NEON instruction set is supported. llvm-svn: 73919
49 lines
1.6 KiB
LLVM
49 lines
1.6 KiB
LLVM
; RUN: llvm-as < %s | llc -march=arm -mattr=+neon > %t
|
|
; RUN: grep {vcls\\.s8} %t | count 2
|
|
; RUN: grep {vcls\\.s16} %t | count 2
|
|
; RUN: grep {vcls\\.s32} %t | count 2
|
|
|
|
define <8 x i8> @vclss8(<8 x i8>* %A) nounwind {
|
|
%tmp1 = load <8 x i8>* %A
|
|
%tmp2 = call <8 x i8> @llvm.arm.neon.vcls.v8i8(<8 x i8> %tmp1)
|
|
ret <8 x i8> %tmp2
|
|
}
|
|
|
|
define <4 x i16> @vclss16(<4 x i16>* %A) nounwind {
|
|
%tmp1 = load <4 x i16>* %A
|
|
%tmp2 = call <4 x i16> @llvm.arm.neon.vcls.v4i16(<4 x i16> %tmp1)
|
|
ret <4 x i16> %tmp2
|
|
}
|
|
|
|
define <2 x i32> @vclss32(<2 x i32>* %A) nounwind {
|
|
%tmp1 = load <2 x i32>* %A
|
|
%tmp2 = call <2 x i32> @llvm.arm.neon.vcls.v2i32(<2 x i32> %tmp1)
|
|
ret <2 x i32> %tmp2
|
|
}
|
|
|
|
define <16 x i8> @vclsQs8(<16 x i8>* %A) nounwind {
|
|
%tmp1 = load <16 x i8>* %A
|
|
%tmp2 = call <16 x i8> @llvm.arm.neon.vcls.v16i8(<16 x i8> %tmp1)
|
|
ret <16 x i8> %tmp2
|
|
}
|
|
|
|
define <8 x i16> @vclsQs16(<8 x i16>* %A) nounwind {
|
|
%tmp1 = load <8 x i16>* %A
|
|
%tmp2 = call <8 x i16> @llvm.arm.neon.vcls.v8i16(<8 x i16> %tmp1)
|
|
ret <8 x i16> %tmp2
|
|
}
|
|
|
|
define <4 x i32> @vclsQs32(<4 x i32>* %A) nounwind {
|
|
%tmp1 = load <4 x i32>* %A
|
|
%tmp2 = call <4 x i32> @llvm.arm.neon.vcls.v4i32(<4 x i32> %tmp1)
|
|
ret <4 x i32> %tmp2
|
|
}
|
|
|
|
declare <8 x i8> @llvm.arm.neon.vcls.v8i8(<8 x i8>) nounwind readnone
|
|
declare <4 x i16> @llvm.arm.neon.vcls.v4i16(<4 x i16>) nounwind readnone
|
|
declare <2 x i32> @llvm.arm.neon.vcls.v2i32(<2 x i32>) nounwind readnone
|
|
|
|
declare <16 x i8> @llvm.arm.neon.vcls.v16i8(<16 x i8>) nounwind readnone
|
|
declare <8 x i16> @llvm.arm.neon.vcls.v8i16(<8 x i16>) nounwind readnone
|
|
declare <4 x i32> @llvm.arm.neon.vcls.v4i32(<4 x i32>) nounwind readnone
|