mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-30 23:42:52 +01:00
6db76aaf10
This is still a work in progress but most of the NEON instruction set is supported. llvm-svn: 73919
65 lines
2.5 KiB
LLVM
65 lines
2.5 KiB
LLVM
; RUN: llvm-as < %s | llc -march=arm -mattr=+neon > %t
|
|
; RUN: grep {vcvt\\.s32\\.f32} %t | count 2
|
|
; RUN: grep {vcvt\\.u32\\.f32} %t | count 2
|
|
; RUN: grep {vcvt\\.f32\\.s32} %t | count 2
|
|
; RUN: grep {vcvt\\.f32\\.u32} %t | count 2
|
|
|
|
define <2 x i32> @vcvt_f32tos32(<2 x float>* %A) nounwind {
|
|
%tmp1 = load <2 x float>* %A
|
|
%tmp2 = call <2 x i32> @llvm.arm.neon.vcvtfp2fxs.v2i32.v2f32(<2 x float> %tmp1, i32 1)
|
|
ret <2 x i32> %tmp2
|
|
}
|
|
|
|
define <2 x i32> @vcvt_f32tou32(<2 x float>* %A) nounwind {
|
|
%tmp1 = load <2 x float>* %A
|
|
%tmp2 = call <2 x i32> @llvm.arm.neon.vcvtfp2fxu.v2i32.v2f32(<2 x float> %tmp1, i32 1)
|
|
ret <2 x i32> %tmp2
|
|
}
|
|
|
|
define <2 x float> @vcvt_s32tof32(<2 x i32>* %A) nounwind {
|
|
%tmp1 = load <2 x i32>* %A
|
|
%tmp2 = call <2 x float> @llvm.arm.neon.vcvtfxs2fp.v2f32.v2i32(<2 x i32> %tmp1, i32 1)
|
|
ret <2 x float> %tmp2
|
|
}
|
|
|
|
define <2 x float> @vcvt_u32tof32(<2 x i32>* %A) nounwind {
|
|
%tmp1 = load <2 x i32>* %A
|
|
%tmp2 = call <2 x float> @llvm.arm.neon.vcvtfxu2fp.v2f32.v2i32(<2 x i32> %tmp1, i32 1)
|
|
ret <2 x float> %tmp2
|
|
}
|
|
|
|
declare <2 x i32> @llvm.arm.neon.vcvtfp2fxs.v2i32.v2f32(<2 x float>, i32) nounwind readnone
|
|
declare <2 x i32> @llvm.arm.neon.vcvtfp2fxu.v2i32.v2f32(<2 x float>, i32) nounwind readnone
|
|
declare <2 x float> @llvm.arm.neon.vcvtfxs2fp.v2f32.v2i32(<2 x i32>, i32) nounwind readnone
|
|
declare <2 x float> @llvm.arm.neon.vcvtfxu2fp.v2f32.v2i32(<2 x i32>, i32) nounwind readnone
|
|
|
|
define <4 x i32> @vcvtQ_f32tos32(<4 x float>* %A) nounwind {
|
|
%tmp1 = load <4 x float>* %A
|
|
%tmp2 = call <4 x i32> @llvm.arm.neon.vcvtfp2fxs.v4i32.v4f32(<4 x float> %tmp1, i32 1)
|
|
ret <4 x i32> %tmp2
|
|
}
|
|
|
|
define <4 x i32> @vcvtQ_f32tou32(<4 x float>* %A) nounwind {
|
|
%tmp1 = load <4 x float>* %A
|
|
%tmp2 = call <4 x i32> @llvm.arm.neon.vcvtfp2fxu.v4i32.v4f32(<4 x float> %tmp1, i32 1)
|
|
ret <4 x i32> %tmp2
|
|
}
|
|
|
|
define <4 x float> @vcvtQ_s32tof32(<4 x i32>* %A) nounwind {
|
|
%tmp1 = load <4 x i32>* %A
|
|
%tmp2 = call <4 x float> @llvm.arm.neon.vcvtfxs2fp.v4f32.v4i32(<4 x i32> %tmp1, i32 1)
|
|
ret <4 x float> %tmp2
|
|
}
|
|
|
|
define <4 x float> @vcvtQ_u32tof32(<4 x i32>* %A) nounwind {
|
|
%tmp1 = load <4 x i32>* %A
|
|
%tmp2 = call <4 x float> @llvm.arm.neon.vcvtfxu2fp.v4f32.v4i32(<4 x i32> %tmp1, i32 1)
|
|
ret <4 x float> %tmp2
|
|
}
|
|
|
|
declare <4 x i32> @llvm.arm.neon.vcvtfp2fxs.v4i32.v4f32(<4 x float>, i32) nounwind readnone
|
|
declare <4 x i32> @llvm.arm.neon.vcvtfp2fxu.v4i32.v4f32(<4 x float>, i32) nounwind readnone
|
|
declare <4 x float> @llvm.arm.neon.vcvtfxs2fp.v4f32.v4i32(<4 x i32>, i32) nounwind readnone
|
|
declare <4 x float> @llvm.arm.neon.vcvtfxu2fp.v4f32.v4i32(<4 x i32>, i32) nounwind readnone
|
|
|