From b0abb4dc4203b903d8d0b48a952ba0a6312eeeb7 Mon Sep 17 00:00:00 2001 From: Bob Wilson Date: Tue, 11 Aug 2009 05:39:44 +0000 Subject: Use vAny type to get rid of Neon intrinsics that differed only in whether the overloaded vector types allowed floating-point or integer vector elements. Most of these operations actually depend on the element type, so bitcasting was not an option. If you include the vpadd intrinsics that I updated earlier, this gets rid of 20 intrinsics. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78646 91177308-0d34-0410-b5e6-96231b3b80d8 --- test/CodeGen/ARM/vuzp.ll | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) (limited to 'test/CodeGen/ARM/vuzp.ll') diff --git a/test/CodeGen/ARM/vuzp.ll b/test/CodeGen/ARM/vuzp.ll index 508ae14706..883e0722ab 100644 --- a/test/CodeGen/ARM/vuzp.ll +++ b/test/CodeGen/ARM/vuzp.ll @@ -15,7 +15,7 @@ define <8 x i8> @vuzpi8(<8 x i8>* %A, <8 x i8>* %B) nounwind { ;CHECK: vuzp.8 %tmp1 = load <8 x i8>* %A %tmp2 = load <8 x i8>* %B - %tmp3 = call %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vuzpi.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2) + %tmp3 = call %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vuzp.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v8qi2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v8qi2 %tmp3, 1 %tmp6 = add <8 x i8> %tmp4, %tmp5 @@ -27,7 +27,7 @@ define <4 x i16> @vuzpi16(<4 x i16>* %A, <4 x i16>* %B) nounwind { ;CHECK: vuzp.16 %tmp1 = load <4 x i16>* %A %tmp2 = load <4 x i16>* %B - %tmp3 = call %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vuzpi.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2) + %tmp3 = call %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vuzp.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v4hi2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v4hi2 %tmp3, 1 %tmp6 = add <4 x i16> %tmp4, %tmp5 @@ -39,7 +39,7 @@ define <2 x i32> @vuzpi32(<2 x i32>* %A, <2 x i32>* %B) nounwind { ;CHECK: vuzp.32 %tmp1 = load <2 x i32>* %A %tmp2 = load <2 x i32>* %B - %tmp3 = call %struct.__builtin_neon_v2si2 @llvm.arm.neon.vuzpi.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2) + %tmp3 = call %struct.__builtin_neon_v2si2 @llvm.arm.neon.vuzp.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v2si2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v2si2 %tmp3, 1 %tmp6 = add <2 x i32> %tmp4, %tmp5 @@ -51,7 +51,7 @@ define <2 x float> @vuzpf(<2 x float>* %A, <2 x float>* %B) nounwind { ;CHECK: vuzp.32 %tmp1 = load <2 x float>* %A %tmp2 = load <2 x float>* %B - %tmp3 = call %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vuzpf.v2f32(<2 x float> %tmp1, <2 x float> %tmp2) + %tmp3 = call %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vuzp.v2f32(<2 x float> %tmp1, <2 x float> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v2sf2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v2sf2 %tmp3, 1 %tmp6 = add <2 x float> %tmp4, %tmp5 @@ -63,7 +63,7 @@ define <16 x i8> @vuzpQi8(<16 x i8>* %A, <16 x i8>* %B) nounwind { ;CHECK: vuzp.8 %tmp1 = load <16 x i8>* %A %tmp2 = load <16 x i8>* %B - %tmp3 = call %struct.__builtin_neon_v16qi2 @llvm.arm.neon.vuzpi.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2) + %tmp3 = call %struct.__builtin_neon_v16qi2 @llvm.arm.neon.vuzp.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v16qi2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v16qi2 %tmp3, 1 %tmp6 = add <16 x i8> %tmp4, %tmp5 @@ -75,7 +75,7 @@ define <8 x i16> @vuzpQi16(<8 x i16>* %A, <8 x i16>* %B) nounwind { ;CHECK: vuzp.16 %tmp1 = load <8 x i16>* %A %tmp2 = load <8 x i16>* %B - %tmp3 = call %struct.__builtin_neon_v8hi2 @llvm.arm.neon.vuzpi.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2) + %tmp3 = call %struct.__builtin_neon_v8hi2 @llvm.arm.neon.vuzp.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v8hi2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v8hi2 %tmp3, 1 %tmp6 = add <8 x i16> %tmp4, %tmp5 @@ -87,7 +87,7 @@ define <4 x i32> @vuzpQi32(<4 x i32>* %A, <4 x i32>* %B) nounwind { ;CHECK: vuzp.32 %tmp1 = load <4 x i32>* %A %tmp2 = load <4 x i32>* %B - %tmp3 = call %struct.__builtin_neon_v4si2 @llvm.arm.neon.vuzpi.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2) + %tmp3 = call %struct.__builtin_neon_v4si2 @llvm.arm.neon.vuzp.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v4si2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v4si2 %tmp3, 1 %tmp6 = add <4 x i32> %tmp4, %tmp5 @@ -99,19 +99,19 @@ define <4 x float> @vuzpQf(<4 x float>* %A, <4 x float>* %B) nounwind { ;CHECK: vuzp.32 %tmp1 = load <4 x float>* %A %tmp2 = load <4 x float>* %B - %tmp3 = call %struct.__builtin_neon_v4sf2 @llvm.arm.neon.vuzpf.v4f32(<4 x float> %tmp1, <4 x float> %tmp2) + %tmp3 = call %struct.__builtin_neon_v4sf2 @llvm.arm.neon.vuzp.v4f32(<4 x float> %tmp1, <4 x float> %tmp2) %tmp4 = extractvalue %struct.__builtin_neon_v4sf2 %tmp3, 0 %tmp5 = extractvalue %struct.__builtin_neon_v4sf2 %tmp3, 1 %tmp6 = add <4 x float> %tmp4, %tmp5 ret <4 x float> %tmp6 } -declare %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vuzpi.v8i8(<8 x i8>, <8 x i8>) nounwind readnone -declare %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vuzpi.v4i16(<4 x i16>, <4 x i16>) nounwind readnone -declare %struct.__builtin_neon_v2si2 @llvm.arm.neon.vuzpi.v2i32(<2 x i32>, <2 x i32>) nounwind readnone -declare %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vuzpf.v2f32(<2 x float>, <2 x float>) nounwind readnone +declare %struct.__builtin_neon_v8qi2 @llvm.arm.neon.vuzp.v8i8(<8 x i8>, <8 x i8>) nounwind readnone +declare %struct.__builtin_neon_v4hi2 @llvm.arm.neon.vuzp.v4i16(<4 x i16>, <4 x i16>) nounwind readnone +declare %struct.__builtin_neon_v2si2 @llvm.arm.neon.vuzp.v2i32(<2 x i32>, <2 x i32>) nounwind readnone +declare %struct.__builtin_neon_v2sf2 @llvm.arm.neon.vuzp.v2f32(<2 x float>, <2 x float>) nounwind readnone -declare %struct.__builtin_neon_v16qi2 @llvm.arm.neon.vuzpi.v16i8(<16 x i8>, <16 x i8>) nounwind readnone -declare %struct.__builtin_neon_v8hi2 @llvm.arm.neon.vuzpi.v8i16(<8 x i16>, <8 x i16>) nounwind readnone -declare %struct.__builtin_neon_v4si2 @llvm.arm.neon.vuzpi.v4i32(<4 x i32>, <4 x i32>) nounwind readnone -declare %struct.__builtin_neon_v4sf2 @llvm.arm.neon.vuzpf.v4f32(<4 x float>, <4 x float>) nounwind readnone +declare %struct.__builtin_neon_v16qi2 @llvm.arm.neon.vuzp.v16i8(<16 x i8>, <16 x i8>) nounwind readnone +declare %struct.__builtin_neon_v8hi2 @llvm.arm.neon.vuzp.v8i16(<8 x i16>, <8 x i16>) nounwind readnone +declare %struct.__builtin_neon_v4si2 @llvm.arm.neon.vuzp.v4i32(<4 x i32>, <4 x i32>) nounwind readnone +declare %struct.__builtin_neon_v4sf2 @llvm.arm.neon.vuzp.v4f32(<4 x float>, <4 x float>) nounwind readnone -- cgit v1.2.3