-; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
+; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s
define <8 x i8> @vaddi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK-LABEL: vaddi8:
ret <4 x float> %tmp3
}
-define <8 x i8> @vaddhni16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
-;CHECK-LABEL: vaddhni16:
-;CHECK: vaddhn.i16
- %tmp1 = load <8 x i16>* %A
- %tmp2 = load <8 x i16>* %B
- %tmp3 = call <8 x i8> @llvm.arm.neon.vaddhn.v8i8(<8 x i16> %tmp1, <8 x i16> %tmp2)
- ret <8 x i8> %tmp3
-}
-
-define <4 x i16> @vaddhni32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
-;CHECK-LABEL: vaddhni32:
-;CHECK: vaddhn.i32
- %tmp1 = load <4 x i32>* %A
- %tmp2 = load <4 x i32>* %B
- %tmp3 = call <4 x i16> @llvm.arm.neon.vaddhn.v4i16(<4 x i32> %tmp1, <4 x i32> %tmp2)
- ret <4 x i16> %tmp3
-}
-
-define <2 x i32> @vaddhni64(<2 x i64>* %A, <2 x i64>* %B) nounwind {
-;CHECK-LABEL: vaddhni64:
-;CHECK: vaddhn.i64
- %tmp1 = load <2 x i64>* %A
- %tmp2 = load <2 x i64>* %B
- %tmp3 = call <2 x i32> @llvm.arm.neon.vaddhn.v2i32(<2 x i64> %tmp1, <2 x i64> %tmp2)
- ret <2 x i32> %tmp3
-}
-
-declare <8 x i8> @llvm.arm.neon.vaddhn.v8i8(<8 x i16>, <8 x i16>) nounwind readnone
-declare <4 x i16> @llvm.arm.neon.vaddhn.v4i16(<4 x i32>, <4 x i32>) nounwind readnone
-declare <2 x i32> @llvm.arm.neon.vaddhn.v2i32(<2 x i64>, <2 x i64>) nounwind readnone
-
define <8 x i8> @vraddhni16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
;CHECK-LABEL: vraddhni16:
;CHECK: vraddhn.i16
declare <4 x i16> @llvm.arm.neon.vraddhn.v4i16(<4 x i32>, <4 x i32>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vraddhn.v2i32(<2 x i64>, <2 x i64>) nounwind readnone
+define <8 x i8> @vaddhni16_natural(<8 x i16> %A, <8 x i16> %B) nounwind {
+; CHECK-LABEL: vaddhni16_natural:
+; CHECK: vaddhn.i16
+ %sum = add <8 x i16> %A, %B
+ %shift = lshr <8 x i16> %sum, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
+ %trunc = trunc <8 x i16> %shift to <8 x i8>
+ ret <8 x i8> %trunc
+}
+
+define <4 x i16> @vaddhni32_natural(<4 x i32> %A, <4 x i32> %B) nounwind {
+; CHECK-LABEL: vaddhni32_natural:
+; CHECK: vaddhn.i32
+ %sum = add <4 x i32> %A, %B
+ %shift = lshr <4 x i32> %sum, <i32 16, i32 16, i32 16, i32 16>
+ %trunc = trunc <4 x i32> %shift to <4 x i16>
+ ret <4 x i16> %trunc
+}
+
+define <2 x i32> @vaddhni64_natural(<2 x i64> %A, <2 x i64> %B) nounwind {
+; CHECK-LABEL: vaddhni64_natural:
+; CHECK: vaddhn.i64
+ %sum = add <2 x i64> %A, %B
+ %shift = lshr <2 x i64> %sum, <i64 32, i64 32>
+ %trunc = trunc <2 x i64> %shift to <2 x i32>
+ ret <2 x i32> %trunc
+}
+
define <8 x i16> @vaddls8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK-LABEL: vaddls8:
;CHECK: vaddl.s8