[AArch64][GISel] Add handling for G_VECREDUCE_FMAXIMUM and G_VECREDUC…

…E_FMINIMUM This is a lot of copy-pasting for the existing handling of G_VECREDUCE_FMAX/G_VECREDUCE_FMIN to add handling for G_VECREDUCE_FMAXIMUM/G_VECREDUCE_FMINIMUM in the same way. Differential Revision: https://reviews.llvm.org/D156615
llvm · Aug 14, 2023 · a3f2751 · a3f2751
1 parent 660fded
commit a3f2751
Show file tree

Hide file tree

Showing 18 changed files with 666 additions and 299 deletions.
diff --git a/llvm/docs/GlobalISel/GenericOpcode.rst b/llvm/docs/GlobalISel/GenericOpcode.rst
@@ -655,10 +655,10 @@ G_VECREDUCE_FADD, G_VECREDUCE_FMUL
 
 These reductions are relaxed variants which may reduce the elements in any order.
 
-G_VECREDUCE_FMAX, G_VECREDUCE_FMIN
+G_VECREDUCE_FMAX, G_VECREDUCE_FMIN, G_VECREDUCE_FMAXIMUM, G_VECREDUCE_FMINIMUM
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-FMIN/FMAX nodes can have flags, for NaN/NoNaN variants.
+FMIN/FMAX/FMINIMUM/FMAXIMUM nodes can have flags, for NaN/NoNaN variants.
 
 
 Integer/bitwise reductions

diff --git a/llvm/include/llvm/CodeGen/GlobalISel/GenericMachineInstrs.h b/llvm/include/llvm/CodeGen/GlobalISel/GenericMachineInstrs.h
@@ -409,6 +409,8 @@ class GVecReduce : public GenericMachineInstr {
     case TargetOpcode::G_VECREDUCE_FMUL:
     case TargetOpcode::G_VECREDUCE_FMAX:
     case TargetOpcode::G_VECREDUCE_FMIN:
+    case TargetOpcode::G_VECREDUCE_FMAXIMUM:
+    case TargetOpcode::G_VECREDUCE_FMINIMUM:
     case TargetOpcode::G_VECREDUCE_ADD:
     case TargetOpcode::G_VECREDUCE_MUL:
     case TargetOpcode::G_VECREDUCE_AND:
@@ -441,6 +443,12 @@ class GVecReduce : public GenericMachineInstr {
     case TargetOpcode::G_VECREDUCE_FMIN:
       ScalarOpc = TargetOpcode::G_FMINNUM;
       break;
+    case TargetOpcode::G_VECREDUCE_FMAXIMUM:
+      ScalarOpc = TargetOpcode::G_FMAXIMUM;
+      break;
+    case TargetOpcode::G_VECREDUCE_FMINIMUM:
+      ScalarOpc = TargetOpcode::G_FMINIMUM;
+      break;
     case TargetOpcode::G_VECREDUCE_ADD:
       ScalarOpc = TargetOpcode::G_ADD;
       break;

diff --git a/llvm/include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h b/llvm/include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h
@@ -1973,6 +1973,19 @@ class MachineIRBuilder {
   MachineInstrBuilder buildVecReduceFMin(const DstOp &Dst, const SrcOp &Src) {
     return buildInstr(TargetOpcode::G_VECREDUCE_FMIN, {Dst}, {Src});
   }
+
+  /// Build and insert \p Res = G_VECREDUCE_FMAXIMUM \p Src
+  MachineInstrBuilder buildVecReduceFMaximum(const DstOp &Dst,
+                                             const SrcOp &Src) {
+    return buildInstr(TargetOpcode::G_VECREDUCE_FMAXIMUM, {Dst}, {Src});
+  }
+
+  /// Build and insert \p Res = G_VECREDUCE_FMINIMUM \p Src
+  MachineInstrBuilder buildVecReduceFMinimum(const DstOp &Dst,
+                                             const SrcOp &Src) {
+    return buildInstr(TargetOpcode::G_VECREDUCE_FMINIMUM, {Dst}, {Src});
+  }
+
   /// Build and insert \p Res = G_VECREDUCE_ADD \p Src
   MachineInstrBuilder buildVecReduceAdd(const DstOp &Dst, const SrcOp &Src) {
     return buildInstr(TargetOpcode::G_VECREDUCE_ADD, {Dst}, {Src});

diff --git a/llvm/include/llvm/CodeGen/GlobalISel/Utils.h b/llvm/include/llvm/CodeGen/GlobalISel/Utils.h
@@ -57,6 +57,8 @@ class APFloat;
   case TargetOpcode::G_VECREDUCE_FMUL:                                         \
   case TargetOpcode::G_VECREDUCE_FMAX:                                         \
   case TargetOpcode::G_VECREDUCE_FMIN:                                         \
+  case TargetOpcode::G_VECREDUCE_FMAXIMUM:                                     \
+  case TargetOpcode::G_VECREDUCE_FMINIMUM:                                     \
   case TargetOpcode::G_VECREDUCE_ADD:                                          \
   case TargetOpcode::G_VECREDUCE_MUL:                                          \
   case TargetOpcode::G_VECREDUCE_AND:                                          \
@@ -72,6 +74,8 @@ class APFloat;
   case TargetOpcode::G_VECREDUCE_FMUL:                                         \
   case TargetOpcode::G_VECREDUCE_FMAX:                                         \
   case TargetOpcode::G_VECREDUCE_FMIN:                                         \
+  case TargetOpcode::G_VECREDUCE_FMAXIMUM:                                     \
+  case TargetOpcode::G_VECREDUCE_FMINIMUM:                                     \
   case TargetOpcode::G_VECREDUCE_ADD:                                          \
   case TargetOpcode::G_VECREDUCE_MUL:                                          \
   case TargetOpcode::G_VECREDUCE_AND:                                          \

diff --git a/llvm/include/llvm/Support/TargetOpcodes.def b/llvm/include/llvm/Support/TargetOpcodes.def
@@ -811,6 +811,8 @@ HANDLE_TARGET_OPCODE(G_VECREDUCE_FADD)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_FMUL)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_FMAX)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_FMIN)
+HANDLE_TARGET_OPCODE(G_VECREDUCE_FMAXIMUM)
+HANDLE_TARGET_OPCODE(G_VECREDUCE_FMINIMUM)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_ADD)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_MUL)
 HANDLE_TARGET_OPCODE(G_VECREDUCE_AND)

diff --git a/llvm/include/llvm/Target/GenericOpcodes.td b/llvm/include/llvm/Target/GenericOpcodes.td
@@ -1398,6 +1398,8 @@ def G_VECREDUCE_FMUL : VectorReduction;
 
 def G_VECREDUCE_FMAX : VectorReduction;
 def G_VECREDUCE_FMIN : VectorReduction;
+def G_VECREDUCE_FMAXIMUM : VectorReduction;
+def G_VECREDUCE_FMINIMUM : VectorReduction;
 
 def G_VECREDUCE_ADD : VectorReduction;
 def G_VECREDUCE_MUL : VectorReduction;

diff --git a/llvm/include/llvm/Target/GlobalISel/SelectionDAGCompat.td b/llvm/include/llvm/Target/GlobalISel/SelectionDAGCompat.td
@@ -165,6 +165,8 @@ def : GINodeEquiv<G_LLROUND, llround>;
 def : GINodeEquiv<G_VECREDUCE_FADD, vecreduce_fadd>;
 def : GINodeEquiv<G_VECREDUCE_FMAX, vecreduce_fmax>;
 def : GINodeEquiv<G_VECREDUCE_FMIN, vecreduce_fmin>;
+def : GINodeEquiv<G_VECREDUCE_FMAXIMUM, vecreduce_fmaximum>;
+def : GINodeEquiv<G_VECREDUCE_FMINIMUM, vecreduce_fminimum>;
 
 def : GINodeEquiv<G_STRICT_FADD, strict_fadd>;
 def : GINodeEquiv<G_STRICT_FSUB, strict_fsub>;

diff --git a/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp b/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp
@@ -1797,6 +1797,10 @@ unsigned IRTranslator::getSimpleIntrinsicOpcode(Intrinsic::ID ID) {
       return TargetOpcode::G_VECREDUCE_FMIN;
     case Intrinsic::vector_reduce_fmax:
       return TargetOpcode::G_VECREDUCE_FMAX;
+    case Intrinsic::vector_reduce_fminimum:
+      return TargetOpcode::G_VECREDUCE_FMINIMUM;
+    case Intrinsic::vector_reduce_fmaximum:
+      return TargetOpcode::G_VECREDUCE_FMAXIMUM;
     case Intrinsic::vector_reduce_add:
       return TargetOpcode::G_VECREDUCE_ADD;
     case Intrinsic::vector_reduce_mul:

diff --git a/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp b/llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp
@@ -2649,6 +2649,8 @@ LegalizerHelper::widenScalar(MachineInstr &MI, unsigned TypeIdx, LLT WideTy) {
   }
   case TargetOpcode::G_VECREDUCE_FMIN:
   case TargetOpcode::G_VECREDUCE_FMAX:
+  case TargetOpcode::G_VECREDUCE_FMINIMUM:
+  case TargetOpcode::G_VECREDUCE_FMAXIMUM:
     if (TypeIdx != 0)
       return UnableToLegalize;
     Observer.changingInstr(MI);

diff --git a/llvm/lib/CodeGen/MachineVerifier.cpp b/llvm/lib/CodeGen/MachineVerifier.cpp
@@ -1720,6 +1720,8 @@ void MachineVerifier::verifyPreISelGenericInstruction(const MachineInstr *MI) {
   case TargetOpcode::G_VECREDUCE_FMUL:
   case TargetOpcode::G_VECREDUCE_FMAX:
   case TargetOpcode::G_VECREDUCE_FMIN:
+  case TargetOpcode::G_VECREDUCE_FMAXIMUM:
+  case TargetOpcode::G_VECREDUCE_FMINIMUM:
   case TargetOpcode::G_VECREDUCE_ADD:
   case TargetOpcode::G_VECREDUCE_MUL:
   case TargetOpcode::G_VECREDUCE_AND:

diff --git a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp
@@ -850,7 +850,8 @@ AArch64LegalizerInfo::AArch64LegalizerInfo(const AArch64Subtarget &ST)
       .clampMaxNumElements(1, s32, 4)
       .lower();
 
-  getActionDefinitionsBuilder({G_VECREDUCE_FMIN, G_VECREDUCE_FMAX})
+  getActionDefinitionsBuilder({G_VECREDUCE_FMIN, G_VECREDUCE_FMAX,
+                               G_VECREDUCE_FMINIMUM, G_VECREDUCE_FMAXIMUM})
       .legalFor({{s32, v4s32}, {s32, v2s32}, {s64, v2s64}})
       .legalIf([=](const LegalityQuery &Query) {
         const auto &Ty = Query.Types[1];

diff --git a/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp
@@ -996,6 +996,8 @@ AArch64RegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
   case TargetOpcode::G_VECREDUCE_FMUL:
   case TargetOpcode::G_VECREDUCE_FMAX:
   case TargetOpcode::G_VECREDUCE_FMIN:
+  case TargetOpcode::G_VECREDUCE_FMAXIMUM:
+  case TargetOpcode::G_VECREDUCE_FMINIMUM:
   case TargetOpcode::G_VECREDUCE_ADD:
   case TargetOpcode::G_VECREDUCE_MUL:
   case TargetOpcode::G_VECREDUCE_AND:

diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/irtranslator-reductions.ll b/llvm/test/CodeGen/AArch64/GlobalISel/irtranslator-reductions.ll
@@ -66,6 +66,8 @@ define double @fmul_fast(double %start, <4 x double> %vec) {
 
 declare float @llvm.vector.reduce.fmax.v4f32(<4 x float>)
 declare float @llvm.vector.reduce.fmin.v4f32(<4 x float>)
+declare float @llvm.vector.reduce.fmaximum.v4f32(<4 x float>)
+declare float @llvm.vector.reduce.fminimum.v4f32(<4 x float>)
 
 define float @fmax(<4 x float> %vec) {
   ; CHECK-LABEL: name: fmax
@@ -106,6 +108,45 @@ define float @fmin_nnan(<4 x float> %vec) {
   ret float %res
 }
 
+define float @fmaximum(<4 x float> %vec) {
+  ; CHECK-LABEL: name: fmaximum
+  ; CHECK: bb.1 (%ir-block.0):
+  ; CHECK:   liveins: $q0
+  ; CHECK:   [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $q0
+  ; CHECK:   [[BITCAST:%[0-9]+]]:_(<4 x s32>) = G_BITCAST [[COPY]](<2 x s64>)
+  ; CHECK:   [[VECREDUCE_FMAX:%[0-9]+]]:_(s32) = G_VECREDUCE_FMAXIMUM [[BITCAST]](<4 x s32>)
+  ; CHECK:   $s0 = COPY [[VECREDUCE_FMAX]](s32)
+  ; CHECK:   RET_ReallyLR implicit $s0
+  %res = call float @llvm.vector.reduce.fmaximum.v4f32(<4 x float> %vec)
+  ret float %res
+}
+
+define float @fminimum(<4 x float> %vec) {
+  ; CHECK-LABEL: name: fminimum
+  ; CHECK: bb.1 (%ir-block.0):
+  ; CHECK:   liveins: $q0
+  ; CHECK:   [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $q0
+  ; CHECK:   [[BITCAST:%[0-9]+]]:_(<4 x s32>) = G_BITCAST [[COPY]](<2 x s64>)
+  ; CHECK:   [[VECREDUCE_FMIN:%[0-9]+]]:_(s32) = G_VECREDUCE_FMINIMUM [[BITCAST]](<4 x s32>)
+  ; CHECK:   $s0 = COPY [[VECREDUCE_FMIN]](s32)
+  ; CHECK:   RET_ReallyLR implicit $s0
+  %res = call float @llvm.vector.reduce.fminimum.v4f32(<4 x float> %vec)
+  ret float %res
+}
+
+define float @fminimum_nnan(<4 x float> %vec) {
+  ; CHECK-LABEL: name: fminimum_nnan
+  ; CHECK: bb.1 (%ir-block.0):
+  ; CHECK:   liveins: $q0
+  ; CHECK:   [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $q0
+  ; CHECK:   [[BITCAST:%[0-9]+]]:_(<4 x s32>) = G_BITCAST [[COPY]](<2 x s64>)
+  ; CHECK:   [[VECREDUCE_FMIN:%[0-9]+]]:_(s32) = nnan G_VECREDUCE_FMINIMUM [[BITCAST]](<4 x s32>)
+  ; CHECK:   $s0 = COPY [[VECREDUCE_FMIN]](s32)
+  ; CHECK:   RET_ReallyLR implicit $s0
+  %res = call nnan float @llvm.vector.reduce.fminimum.v4f32(<4 x float> %vec)
+  ret float %res
+}
+
 declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)
 
 define i32 @add(<4 x i32> %vec) {

diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/legalize-reduce-fminmax.mir b/llvm/test/CodeGen/AArch64/GlobalISel/legalize-reduce-fminmax.mir
@@ -0,0 +1,93 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -O0 -mtriple=aarch64 -run-pass=legalizer -global-isel-abort=1 %s -o - | FileCheck %s
+
+---
+name:            fmin_v2s32
+tracksRegLiveness: true
+body:             |
+  bb.1:
+    liveins: $d0
+
+    ; CHECK-LABEL: name: fmin_v2s32
+    ; CHECK: liveins: $d0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $d0
+    ; CHECK-NEXT: [[VECREDUCE_FMIN:%[0-9]+]]:_(s32) = G_VECREDUCE_FMIN [[COPY]](<2 x s32>)
+    ; CHECK-NEXT: $s0 = COPY [[VECREDUCE_FMIN]](s32)
+    ; CHECK-NEXT: RET_ReallyLR implicit $s0
+    %0:_(<2 x s32>) = COPY $d0
+    %1:_(s32) = G_VECREDUCE_FMIN %0(<2 x s32>)
+    $s0 = COPY %1(s32)
+    RET_ReallyLR implicit $s0
+
+...
+---
+name:            fmax_v8s16
+tracksRegLiveness: true
+body:             |
+  bb.1:
+    liveins: $q0
+
+    ; CHECK-LABEL: name: fmax_v8s16
+    ; CHECK: liveins: $q0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<8 x s16>) = COPY $q0
+    ; CHECK-NEXT: [[UV:%[0-9]+]]:_(<4 x s16>), [[UV1:%[0-9]+]]:_(<4 x s16>) = G_UNMERGE_VALUES [[COPY]](<8 x s16>)
+    ; CHECK-NEXT: [[FPEXT:%[0-9]+]]:_(<4 x s32>) = G_FPEXT [[UV]](<4 x s16>)
+    ; CHECK-NEXT: [[FPEXT1:%[0-9]+]]:_(<4 x s32>) = G_FPEXT [[UV1]](<4 x s16>)
+    ; CHECK-NEXT: [[FMAXNUM:%[0-9]+]]:_(<4 x s32>) = G_FMAXNUM [[FPEXT]], [[FPEXT1]]
+    ; CHECK-NEXT: [[VECREDUCE_FMAX:%[0-9]+]]:_(s32) = G_VECREDUCE_FMAX [[FMAXNUM]](<4 x s32>)
+    ; CHECK-NEXT: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[VECREDUCE_FMAX]](s32)
+    ; CHECK-NEXT: $h0 = COPY [[FPTRUNC]](s16)
+    ; CHECK-NEXT: RET_ReallyLR implicit $h0
+    %0:_(<8 x s16>) = COPY $q0
+    %1:_(s16) = G_VECREDUCE_FMAX %0(<8 x s16>)
+    $h0 = COPY %1(s16)
+    RET_ReallyLR implicit $h0
+
+...
+---
+name:            fminimum_v2s32
+tracksRegLiveness: true
+body:             |
+  bb.1:
+    liveins: $d0
+
+    ; CHECK-LABEL: name: fminimum_v2s32
+    ; CHECK: liveins: $d0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $d0
+    ; CHECK-NEXT: [[VECREDUCE_FMINIMUM:%[0-9]+]]:_(s32) = G_VECREDUCE_FMINIMUM [[COPY]](<2 x s32>)
+    ; CHECK-NEXT: $s0 = COPY [[VECREDUCE_FMINIMUM]](s32)
+    ; CHECK-NEXT: RET_ReallyLR implicit $s0
+    %0:_(<2 x s32>) = COPY $d0
+    %1:_(s32) = G_VECREDUCE_FMINIMUM %0(<2 x s32>)
+    $s0 = COPY %1(s32)
+    RET_ReallyLR implicit $s0
+
+...
+---
+name:            fmaximum_v8s16
+tracksRegLiveness: true
+body:             |
+  bb.1:
+    liveins: $q0
+
+    ; CHECK-LABEL: name: fmaximum_v8s16
+    ; CHECK: liveins: $q0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<8 x s16>) = COPY $q0
+    ; CHECK-NEXT: [[UV:%[0-9]+]]:_(<4 x s16>), [[UV1:%[0-9]+]]:_(<4 x s16>) = G_UNMERGE_VALUES [[COPY]](<8 x s16>)
+    ; CHECK-NEXT: [[FPEXT:%[0-9]+]]:_(<4 x s32>) = G_FPEXT [[UV]](<4 x s16>)
+    ; CHECK-NEXT: [[FPEXT1:%[0-9]+]]:_(<4 x s32>) = G_FPEXT [[UV1]](<4 x s16>)
+    ; CHECK-NEXT: [[FMAXIMUM:%[0-9]+]]:_(<4 x s32>) = G_FMAXIMUM [[FPEXT]], [[FPEXT1]]
+    ; CHECK-NEXT: [[VECREDUCE_FMAXIMUM:%[0-9]+]]:_(s32) = G_VECREDUCE_FMAXIMUM [[FMAXIMUM]](<4 x s32>)
+    ; CHECK-NEXT: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[VECREDUCE_FMAXIMUM]](s32)
+    ; CHECK-NEXT: $h0 = COPY [[FPTRUNC]](s16)
+    ; CHECK-NEXT: RET_ReallyLR implicit $h0
+    %0:_(<8 x s16>) = COPY $q0
+    %1:_(s16) = G_VECREDUCE_FMAXIMUM %0(<8 x s16>)
+    $h0 = COPY %1(s16)
+    RET_ReallyLR implicit $h0
+
+...
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/legalizer-info-validation.mir b/llvm/test/CodeGen/AArch64/GlobalISel/legalizer-info-validation.mir
@@ -726,6 +726,14 @@
 # DEBUG-NEXT: G_VECREDUCE_FMIN (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: G_VECREDUCE_FMAXIMUM (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
+# DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
+# DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: G_VECREDUCE_FMINIMUM (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
+# DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
+# DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: G_VECREDUCE_ADD (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected

diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/regbankselect-reductions.mir b/llvm/test/CodeGen/AArch64/GlobalISel/regbankselect-reductions.mir
@@ -41,3 +41,24 @@ body:             |
     RET_ReallyLR implicit $w0
 
 ...
+---
+name:            fmaximum_v4s32
+legalized:       true
+tracksRegLiveness: true
+body:             |
+  bb.1:
+    liveins: $q0
+
+    ; CHECK-LABEL: name: fmaximum_v4s32
+    ; CHECK: liveins: $q0
+    ; CHECK: [[COPY:%[0-9]+]]:fpr(<4 x s32>) = COPY $q0
+    ; CHECK: [[VECREDUCE_FMAXIMUM:%[0-9]+]]:fpr(s32) = G_VECREDUCE_FMAXIMUM [[COPY]](<4 x s32>)
+    ; CHECK: $w0 = COPY [[VECREDUCE_FMAXIMUM]](s32)
+    ; CHECK: RET_ReallyLR implicit $w0
+    %0:_(<4 x s32>) = COPY $q0
+    %1:_(s32) = G_VECREDUCE_FMAXIMUM %0(<4 x s32>)
+    $w0 = COPY %1(s32)
+    RET_ReallyLR implicit $w0
+
+...
+