225 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py2; RUN: llc < %s -mtriple=aarch64 --mattr=+complxnum,+neon,+fullfp16 -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD3; RUN: llc < %s -mtriple=aarch64 --mattr=+complxnum,+neon,+fullfp16,+sve -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD4; RUN: llc < %s -mtriple=aarch64 --mattr=+complxnum,+neon,+fullfp16,+sve2 -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD5; RUN: llc < %s -mtriple=aarch64 --global-isel --global-isel-abort=2 --mattr=+complxnum,+neon,+fullfp16 -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI6; RUN: llc < %s -mtriple=aarch64 --global-isel --global-isel-abort=2 --mattr=+complxnum,+neon,+fullfp16,+sve -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI7; RUN: llc < %s -mtriple=aarch64 --global-isel --global-isel-abort=2 --mattr=+complxnum,+neon,+fullfp16,+sve2 -o - 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI8 9; CHECK-GI: warning: Instruction selection used fallback path for complex_add_v16f1610; CHECK-GI-NEXT: warning: Instruction selection used fallback path for complex_add_v32f1611; CHECK-GI-NEXT: warning: Instruction selection used fallback path for complex_add_v16f16_with_intrinsic12 13; Expected to not transform14define <2 x half> @complex_add_v2f16(<2 x half> %a, <2 x half> %b) {15; CHECK-SD-LABEL: complex_add_v2f16:16; CHECK-SD: // %bb.0: // %entry17; CHECK-SD-NEXT: // kill: def $d1 killed $d1 def $q118; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q019; CHECK-SD-NEXT: mov h2, v0.h[1]20; CHECK-SD-NEXT: mov h3, v1.h[1]21; CHECK-SD-NEXT: fsub h1, h1, h222; CHECK-SD-NEXT: fadd h0, h3, h023; CHECK-SD-NEXT: mov v1.h[1], v0.h[0]24; CHECK-SD-NEXT: fmov d0, d125; CHECK-SD-NEXT: ret26;27; CHECK-GI-LABEL: complex_add_v2f16:28; CHECK-GI: // %bb.0: // %entry29; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q030; CHECK-GI-NEXT: // kill: def $d1 killed $d1 def $q131; CHECK-GI-NEXT: mov h2, v0.h[1]32; CHECK-GI-NEXT: mov h3, v1.h[1]33; CHECK-GI-NEXT: fsub h1, h1, h234; CHECK-GI-NEXT: fadd h0, h3, h035; CHECK-GI-NEXT: mov v1.h[1], v0.h[0]36; CHECK-GI-NEXT: fmov d0, d137; CHECK-GI-NEXT: ret38entry:39 %a.real = shufflevector <2 x half> %a, <2 x half> zeroinitializer, <1 x i32> <i32 0>40 %a.imag = shufflevector <2 x half> %a, <2 x half> zeroinitializer, <1 x i32> <i32 1>41 %b.real = shufflevector <2 x half> %b, <2 x half> zeroinitializer, <1 x i32> <i32 0>42 %b.imag = shufflevector <2 x half> %b, <2 x half> zeroinitializer, <1 x i32> <i32 1>43 %0 = fsub fast <1 x half> %b.real, %a.imag44 %1 = fadd fast <1 x half> %b.imag, %a.real45 %interleaved.vec = shufflevector <1 x half> %0, <1 x half> %1, <2 x i32> <i32 0, i32 1>46 ret <2 x half> %interleaved.vec47}48 49; Expected to transform50define <4 x half> @complex_add_v4f16(<4 x half> %a, <4 x half> %b) {51; CHECK-LABEL: complex_add_v4f16:52; CHECK: // %bb.0: // %entry53; CHECK-NEXT: fcadd v0.4h, v1.4h, v0.4h, #9054; CHECK-NEXT: ret55entry:56 %a.real = shufflevector <4 x half> %a, <4 x half> zeroinitializer, <2 x i32> <i32 0, i32 2>57 %a.imag = shufflevector <4 x half> %a, <4 x half> zeroinitializer, <2 x i32> <i32 1, i32 3>58 %b.real = shufflevector <4 x half> %b, <4 x half> zeroinitializer, <2 x i32> <i32 0, i32 2>59 %b.imag = shufflevector <4 x half> %b, <4 x half> zeroinitializer, <2 x i32> <i32 1, i32 3>60 %0 = fsub fast <2 x half> %b.real, %a.imag61 %1 = fadd fast <2 x half> %b.imag, %a.real62 %interleaved.vec = shufflevector <2 x half> %0, <2 x half> %1, <4 x i32> <i32 0, i32 2, i32 1, i32 3>63 ret <4 x half> %interleaved.vec64}65 66; Expected to transform67define <8 x half> @complex_add_v8f16(<8 x half> %a, <8 x half> %b) {68; CHECK-LABEL: complex_add_v8f16:69; CHECK: // %bb.0: // %entry70; CHECK-NEXT: fcadd v0.8h, v1.8h, v0.8h, #9071; CHECK-NEXT: ret72entry:73 %a.real = shufflevector <8 x half> %a, <8 x half> zeroinitializer, <4 x i32> <i32 0, i32 2, i32 4, i32 6>74 %a.imag = shufflevector <8 x half> %a, <8 x half> zeroinitializer, <4 x i32> <i32 1, i32 3, i32 5, i32 7>75 %b.real = shufflevector <8 x half> %b, <8 x half> zeroinitializer, <4 x i32> <i32 0, i32 2, i32 4, i32 6>76 %b.imag = shufflevector <8 x half> %b, <8 x half> zeroinitializer, <4 x i32> <i32 1, i32 3, i32 5, i32 7>77 %0 = fsub fast <4 x half> %b.real, %a.imag78 %1 = fadd fast <4 x half> %b.imag, %a.real79 %interleaved.vec = shufflevector <4 x half> %0, <4 x half> %1, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>80 ret <8 x half> %interleaved.vec81}82 83; Expected to transform84define <16 x half> @complex_add_v16f16(<16 x half> %a, <16 x half> %b) {85; CHECK-LABEL: complex_add_v16f16:86; CHECK: // %bb.0: // %entry87; CHECK-NEXT: fcadd v1.8h, v3.8h, v1.8h, #9088; CHECK-NEXT: fcadd v0.8h, v2.8h, v0.8h, #9089; CHECK-NEXT: ret90entry:91 %a.real = shufflevector <16 x half> %a, <16 x half> zeroinitializer, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>92 %a.imag = shufflevector <16 x half> %a, <16 x half> zeroinitializer, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>93 %b.real = shufflevector <16 x half> %b, <16 x half> zeroinitializer, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>94 %b.imag = shufflevector <16 x half> %b, <16 x half> zeroinitializer, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>95 %0 = fsub fast <8 x half> %b.real, %a.imag96 %1 = fadd fast <8 x half> %b.imag, %a.real97 %interleaved.vec = shufflevector <8 x half> %0, <8 x half> %1, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>98 ret <16 x half> %interleaved.vec99}100 101; Expected to transform102define <32 x half> @complex_add_v32f16(<32 x half> %a, <32 x half> %b) {103; CHECK-LABEL: complex_add_v32f16:104; CHECK: // %bb.0: // %entry105; CHECK-NEXT: fcadd v2.8h, v6.8h, v2.8h, #90106; CHECK-NEXT: fcadd v0.8h, v4.8h, v0.8h, #90107; CHECK-NEXT: fcadd v1.8h, v5.8h, v1.8h, #90108; CHECK-NEXT: fcadd v3.8h, v7.8h, v3.8h, #90109; CHECK-NEXT: ret110entry:111 %a.real = shufflevector <32 x half> %a, <32 x half> zeroinitializer, <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>112 %a.imag = shufflevector <32 x half> %a, <32 x half> zeroinitializer, <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>113 %b.real = shufflevector <32 x half> %b, <32 x half> zeroinitializer, <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>114 %b.imag = shufflevector <32 x half> %b, <32 x half> zeroinitializer, <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>115 %0 = fsub fast <16 x half> %b.real, %a.imag116 %1 = fadd fast <16 x half> %b.imag, %a.real117 %interleaved.vec = shufflevector <16 x half> %0, <16 x half> %1, <32 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23, i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27, i32 12, i32 28, i32 13, i32 29, i32 14, i32 30, i32 15, i32 31>118 ret <32 x half> %interleaved.vec119}120 121; Expected to transform122define <4 x half> @complex_add_v4f16_with_intrinsic(<4 x half> %a, <4 x half> %b) {123; CHECK-LABEL: complex_add_v4f16_with_intrinsic:124; CHECK: // %bb.0: // %entry125; CHECK-NEXT: fcadd v0.4h, v1.4h, v0.4h, #90126; CHECK-NEXT: ret127entry:128 %a.deinterleaved = tail call { <2 x half>, <2 x half> } @llvm.vector.deinterleave2.v4f16(<4 x half> %a)129 %a.real = extractvalue { <2 x half>, <2 x half> } %a.deinterleaved, 0130 %a.imag = extractvalue { <2 x half>, <2 x half> } %a.deinterleaved, 1131 %b.deinterleaved = tail call { <2 x half>, <2 x half> } @llvm.vector.deinterleave2.v4f16(<4 x half> %b)132 %b.real = extractvalue { <2 x half>, <2 x half> } %b.deinterleaved, 0133 %b.imag = extractvalue { <2 x half>, <2 x half> } %b.deinterleaved, 1134 %0 = fsub fast <2 x half> %b.real, %a.imag135 %1 = fadd fast <2 x half> %b.imag, %a.real136 %interleaved.vec = tail call <4 x half> @llvm.vector.interleave2.v4f16(<2 x half> %0, <2 x half> %1)137 ret <4 x half> %interleaved.vec138}139 140; Expected to transform141define <8 x half> @complex_add_v8f16_with_intrinsic(<8 x half> %a, <8 x half> %b) {142; CHECK-LABEL: complex_add_v8f16_with_intrinsic:143; CHECK: // %bb.0: // %entry144; CHECK-NEXT: fcadd v0.8h, v1.8h, v0.8h, #90145; CHECK-NEXT: ret146entry:147 %a.deinterleaved = tail call { <4 x half>, <4 x half> } @llvm.vector.deinterleave2.v8f16(<8 x half> %a)148 %a.real = extractvalue { <4 x half>, <4 x half> } %a.deinterleaved, 0149 %a.imag = extractvalue { <4 x half>, <4 x half> } %a.deinterleaved, 1150 %b.deinterleaved = tail call { <4 x half>, <4 x half> } @llvm.vector.deinterleave2.v8f16(<8 x half> %b)151 %b.real = extractvalue { <4 x half>, <4 x half> } %b.deinterleaved, 0152 %b.imag = extractvalue { <4 x half>, <4 x half> } %b.deinterleaved, 1153 %0 = fsub fast <4 x half> %b.real, %a.imag154 %1 = fadd fast <4 x half> %b.imag, %a.real155 %interleaved.vec = tail call <8 x half> @llvm.vector.interleave2.v8f16(<4 x half> %0, <4 x half> %1)156 ret <8 x half> %interleaved.vec157}158 159; Expected to transform160define <16 x half> @complex_add_v16f16_with_intrinsic(<16 x half> %a, <16 x half> %b) {161; CHECK-LABEL: complex_add_v16f16_with_intrinsic:162; CHECK: // %bb.0: // %entry163; CHECK-NEXT: fcadd v1.8h, v3.8h, v1.8h, #90164; CHECK-NEXT: fcadd v0.8h, v2.8h, v0.8h, #90165; CHECK-NEXT: ret166entry:167 %a.deinterleaved = tail call { <8 x half>, <8 x half> } @llvm.vector.deinterleave2.v16f16(<16 x half> %a)168 %a.real = extractvalue { <8 x half>, <8 x half> } %a.deinterleaved, 0169 %a.imag = extractvalue { <8 x half>, <8 x half> } %a.deinterleaved, 1170 %b.deinterleaved = tail call { <8 x half>, <8 x half> } @llvm.vector.deinterleave2.v16f16(<16 x half> %b)171 %b.real = extractvalue { <8 x half>, <8 x half> } %b.deinterleaved, 0172 %b.imag = extractvalue { <8 x half>, <8 x half> } %b.deinterleaved, 1173 %0 = fsub fast <8 x half> %b.real, %a.imag174 %1 = fadd fast <8 x half> %b.imag, %a.real175 %interleaved.vec = tail call <16 x half> @llvm.vector.interleave2.v16f16(<8 x half> %0, <8 x half> %1)176 ret <16 x half> %interleaved.vec177}178 179 180; Expected not to transform as it is integer181define <16 x i16> @complex_add_v16i16(<16 x i16> %a, <16 x i16> %b) {182; CHECK-SD-LABEL: complex_add_v16i16:183; CHECK-SD: // %bb.0: // %entry184; CHECK-SD-NEXT: uzp1 v4.8h, v2.8h, v3.8h185; CHECK-SD-NEXT: uzp1 v5.8h, v0.8h, v1.8h186; CHECK-SD-NEXT: uzp2 v0.8h, v0.8h, v1.8h187; CHECK-SD-NEXT: uzp2 v1.8h, v2.8h, v3.8h188; CHECK-SD-NEXT: sub v2.8h, v4.8h, v0.8h189; CHECK-SD-NEXT: add v1.8h, v1.8h, v5.8h190; CHECK-SD-NEXT: zip1 v0.8h, v2.8h, v1.8h191; CHECK-SD-NEXT: zip2 v1.8h, v2.8h, v1.8h192; CHECK-SD-NEXT: ret193;194; CHECK-GI-LABEL: complex_add_v16i16:195; CHECK-GI: // %bb.0: // %entry196; CHECK-GI-NEXT: uzp1 v4.8h, v0.8h, v1.8h197; CHECK-GI-NEXT: uzp2 v0.8h, v0.8h, v1.8h198; CHECK-GI-NEXT: uzp1 v1.8h, v2.8h, v3.8h199; CHECK-GI-NEXT: uzp2 v2.8h, v2.8h, v3.8h200; CHECK-GI-NEXT: sub v1.8h, v1.8h, v0.8h201; CHECK-GI-NEXT: add v2.8h, v2.8h, v4.8h202; CHECK-GI-NEXT: zip1 v0.8h, v1.8h, v2.8h203; CHECK-GI-NEXT: zip2 v1.8h, v1.8h, v2.8h204; CHECK-GI-NEXT: ret205entry:206 %a.real = shufflevector <16 x i16> %a, <16 x i16> zeroinitializer, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>207 %a.imag = shufflevector <16 x i16> %a, <16 x i16> zeroinitializer, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>208 %b.real = shufflevector <16 x i16> %b, <16 x i16> zeroinitializer, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>209 %b.imag = shufflevector <16 x i16> %b, <16 x i16> zeroinitializer, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>210 %0 = sub <8 x i16> %b.real, %a.imag211 %1 = add <8 x i16> %b.imag, %a.real212 %interleaved.vec = shufflevector <8 x i16> %0, <8 x i16> %1, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>213 ret <16 x i16> %interleaved.vec214}215 216 217declare { <2 x half>, <2 x half> } @llvm.vector.deinterleave2.v4f16(<4 x half>)218declare <4 x half> @llvm.vector.interleave2.v4f16(<2 x half>, <2 x half>)219 220declare { <4 x half>, <4 x half> } @llvm.vector.deinterleave2.v8f16(<8 x half>)221declare <8 x half> @llvm.vector.interleave2.v8f16(<4 x half>, <4 x half>)222 223declare { <8 x half>, <8 x half> } @llvm.vector.deinterleave2.v16f16(<16 x half>)224declare <16 x half> @llvm.vector.interleave2.v16f16(<8 x half>, <8 x half>)225