1; RUN: llc < %s -march=arm64 -aarch64-neon-syntax=apple | FileCheck %s 2 3define <8 x i8> @vuzpi8(<8 x i8>* %A, <8 x i8>* %B) nounwind { 4;CHECK-LABEL: vuzpi8: 5;CHECK: uzp1.8b 6;CHECK: uzp2.8b 7;CHECK-NEXT: add.8b 8 %tmp1 = load <8 x i8>* %A 9 %tmp2 = load <8 x i8>* %B 10 %tmp3 = shufflevector <8 x i8> %tmp1, <8 x i8> %tmp2, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14> 11 %tmp4 = shufflevector <8 x i8> %tmp1, <8 x i8> %tmp2, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15> 12 %tmp5 = add <8 x i8> %tmp3, %tmp4 13 ret <8 x i8> %tmp5 14} 15 16define <4 x i16> @vuzpi16(<4 x i16>* %A, <4 x i16>* %B) nounwind { 17;CHECK-LABEL: vuzpi16: 18;CHECK: uzp1.4h 19;CHECK: uzp2.4h 20;CHECK-NEXT: add.4h 21 %tmp1 = load <4 x i16>* %A 22 %tmp2 = load <4 x i16>* %B 23 %tmp3 = shufflevector <4 x i16> %tmp1, <4 x i16> %tmp2, <4 x i32> <i32 0, i32 2, i32 4, i32 6> 24 %tmp4 = shufflevector <4 x i16> %tmp1, <4 x i16> %tmp2, <4 x i32> <i32 1, i32 3, i32 5, i32 7> 25 %tmp5 = add <4 x i16> %tmp3, %tmp4 26 ret <4 x i16> %tmp5 27} 28 29define <16 x i8> @vuzpQi8(<16 x i8>* %A, <16 x i8>* %B) nounwind { 30;CHECK-LABEL: vuzpQi8: 31;CHECK: uzp1.16b 32;CHECK: uzp2.16b 33;CHECK-NEXT: add.16b 34 %tmp1 = load <16 x i8>* %A 35 %tmp2 = load <16 x i8>* %B 36 %tmp3 = shufflevector <16 x i8> %tmp1, <16 x i8> %tmp2, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30> 37 %tmp4 = shufflevector <16 x i8> %tmp1, <16 x i8> %tmp2, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31> 38 %tmp5 = add <16 x i8> %tmp3, %tmp4 39 ret <16 x i8> %tmp5 40} 41 42define <8 x i16> @vuzpQi16(<8 x i16>* %A, <8 x i16>* %B) nounwind { 43;CHECK-LABEL: vuzpQi16: 44;CHECK: uzp1.8h 45;CHECK: uzp2.8h 46;CHECK-NEXT: add.8h 47 %tmp1 = load <8 x i16>* %A 48 %tmp2 = load <8 x i16>* %B 49 %tmp3 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14> 50 %tmp4 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15> 51 %tmp5 = add <8 x i16> %tmp3, %tmp4 52 ret <8 x i16> %tmp5 53} 54 55define <4 x i32> @vuzpQi32(<4 x i32>* %A, <4 x i32>* %B) nounwind { 56;CHECK-LABEL: vuzpQi32: 57;CHECK: uzp1.4s 58;CHECK: uzp2.4s 59;CHECK-NEXT: add.4s 60 %tmp1 = load <4 x i32>* %A 61 %tmp2 = load <4 x i32>* %B 62 %tmp3 = shufflevector <4 x i32> %tmp1, <4 x i32> %tmp2, <4 x i32> <i32 0, i32 2, i32 4, i32 6> 63 %tmp4 = shufflevector <4 x i32> %tmp1, <4 x i32> %tmp2, <4 x i32> <i32 1, i32 3, i32 5, i32 7> 64 %tmp5 = add <4 x i32> %tmp3, %tmp4 65 ret <4 x i32> %tmp5 66} 67 68define <4 x float> @vuzpQf(<4 x float>* %A, <4 x float>* %B) nounwind { 69;CHECK-LABEL: vuzpQf: 70;CHECK: uzp1.4s 71;CHECK: uzp2.4s 72;CHECK-NEXT: fadd.4s 73 %tmp1 = load <4 x float>* %A 74 %tmp2 = load <4 x float>* %B 75 %tmp3 = shufflevector <4 x float> %tmp1, <4 x float> %tmp2, <4 x i32> <i32 0, i32 2, i32 4, i32 6> 76 %tmp4 = shufflevector <4 x float> %tmp1, <4 x float> %tmp2, <4 x i32> <i32 1, i32 3, i32 5, i32 7> 77 %tmp5 = fadd <4 x float> %tmp3, %tmp4 78 ret <4 x float> %tmp5 79} 80 81; Undef shuffle indices should not prevent matching to VUZP: 82 83define <8 x i8> @vuzpi8_undef(<8 x i8>* %A, <8 x i8>* %B) nounwind { 84;CHECK-LABEL: vuzpi8_undef: 85;CHECK: uzp1.8b 86;CHECK: uzp2.8b 87;CHECK-NEXT: add.8b 88 %tmp1 = load <8 x i8>* %A 89 %tmp2 = load <8 x i8>* %B 90 %tmp3 = shufflevector <8 x i8> %tmp1, <8 x i8> %tmp2, <8 x i32> <i32 0, i32 2, i32 undef, i32 undef, i32 8, i32 10, i32 12, i32 14> 91 %tmp4 = shufflevector <8 x i8> %tmp1, <8 x i8> %tmp2, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 undef, i32 undef, i32 13, i32 15> 92 %tmp5 = add <8 x i8> %tmp3, %tmp4 93 ret <8 x i8> %tmp5 94} 95 96define <8 x i16> @vuzpQi16_undef(<8 x i16>* %A, <8 x i16>* %B) nounwind { 97;CHECK-LABEL: vuzpQi16_undef: 98;CHECK: uzp1.8h 99;CHECK: uzp2.8h 100;CHECK-NEXT: add.8h 101 %tmp1 = load <8 x i16>* %A 102 %tmp2 = load <8 x i16>* %B 103 %tmp3 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 0, i32 undef, i32 4, i32 undef, i32 8, i32 10, i32 12, i32 14> 104 %tmp4 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> <i32 1, i32 3, i32 5, i32 undef, i32 undef, i32 11, i32 13, i32 15> 105 %tmp5 = add <8 x i16> %tmp3, %tmp4 106 ret <8 x i16> %tmp5 107} 108