diff --git a/llvm/lib/Target/AMDGPU/AMDGPUCombine.td b/llvm/lib/Target/AMDGPU/AMDGPUCombine.td
--- a/llvm/lib/Target/AMDGPU/AMDGPUCombine.td
+++ b/llvm/lib/Target/AMDGPU/AMDGPUCombine.td
@@ -88,6 +88,12 @@
          [{ return RegBankHelper.matchFPMed3ToClamp(*${fmed3}, ${matchinfo}); }]),
   (apply [{ RegBankHelper.applyClamp(*${fmed3}, ${matchinfo}); }])>;
 
+def shift_shufflevector_fold : GICombineRule<
+  (defs root:$shift, register_matchinfo:$matchinfo),
+  (match (wip_match_opcode G_LSHR):$shift,
+         [{ return RegBankHelper.matchShiftShuffleVectorFold(*${shift}, ${matchinfo}); }]),
+  (apply [{ RegBankHelper.applyShiftShuffleVectorFold(*${shift}, ${matchinfo}); }])>;
+
 def remove_fcanonicalize_matchinfo : GIDefMatchData<"Register">;
 
 def remove_fcanonicalize : GICombineRule<
@@ -128,7 +134,8 @@
 def AMDGPURegBankCombinerHelper : GICombinerHelper<
   "AMDGPUGenRegBankCombinerHelper",
   [zext_trunc_fold, int_minmax_to_med3, ptr_add_immed_chain,
-   fp_minmax_to_clamp, fp_minmax_to_med3, fmed3_intrinsic_to_clamp]> {
+   fp_minmax_to_clamp, fp_minmax_to_med3, fmed3_intrinsic_to_clamp,
+   shift_shufflevector_fold]> {
   let DisableRuleOption = "amdgpuregbankcombiner-disable-rule";
   let StateClass = "AMDGPURegBankCombinerHelperState";
   let AdditionalArguments = [];
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankCombiner.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegBankCombiner.cpp
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankCombiner.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankCombiner.cpp
@@ -74,6 +74,9 @@
   void applyMed3(MachineInstr &MI, Med3MatchInfo &MatchInfo);
   void applyClamp(MachineInstr &MI, Register &Reg);
 
+  bool matchShiftShuffleVectorFold(MachineInstr &MI, Register &Reg);
+  void applyShiftShuffleVectorFold(MachineInstr &MI, Register &Reg);
+
 private:
   AMDGPU::SIModeRegisterDefaults getMode();
   bool getIEEE();
@@ -328,6 +331,57 @@
   MI.eraseFromParent();
 }
 
+bool AMDGPURegBankCombinerHelper::matchShiftShuffleVectorFold(MachineInstr &MI,
+                                                              Register &Reg) {
+  // Fold (G_LSHR (G_BITCAST (G_SHUFFLE_VECTOR %a, %b, shufflemask(1, ?))), K)
+  // into a simple G_BITCAST(%a) if the shift amount (K) is 1/2 of the
+  // destination type and the vector types have 2 elements.
+  //
+  // TODO: This could be made more generic (e.g. to support more vector sizes,
+  // or even LSHL)
+  //  if need be.
+
+  assert(MI.getOpcode() == AMDGPU::G_LSHR);
+
+  LLT DstTy = MRI.getType(MI.getOperand(0).getReg());
+  const auto ShiftAmount =
+      getIConstantVRegValWithLookThrough(MI.getOperand(2).getReg(), MRI);
+
+  // The shift amount is 1/2 of the scalar type.
+  if (!ShiftAmount || ShiftAmount->Value != (DstTy.getSizeInBits() / 2))
+    return false;
+
+  // The shift operand is a bitcast.
+  MachineInstr *ShiftOp = getDefIgnoringCopies(MI.getOperand(1).getReg(), MRI);
+  if (ShiftOp->getOpcode() != AMDGPU::G_BITCAST)
+    return false;
+
+  // The bitcast src is a SHUFFLE_VECTOR.
+  ShiftOp = getDefIgnoringCopies(ShiftOp->getOperand(1).getReg(), MRI);
+  if (ShiftOp->getOpcode() != AMDGPU::G_SHUFFLE_VECTOR)
+    return false;
+
+  // The SHUFFLE_VECTOR's dst type is a 2-element vector.
+  LLT VectorTy = MRI.getType(ShiftOp->getOperand(0).getReg());
+  if (VectorTy.getNumElements() != 2)
+    return false;
+
+  // The shufflemask starts with 1.
+  if (ShiftOp->getOperand(3).getShuffleMask()[0] != 1)
+    return false;
+
+  // We can just replace this with a bitcast of the LHS of the ShiftOp.
+  Reg = ShiftOp->getOperand(1).getReg();
+  return true;
+}
+
+void AMDGPURegBankCombinerHelper::applyShiftShuffleVectorFold(MachineInstr &MI,
+                                                              Register &Reg) {
+  B.setInstrAndDebugLoc(MI);
+  B.buildBitcast(MI.getOperand(0).getReg(), Reg);
+  MI.eraseFromParent();
+}
+
 AMDGPU::SIModeRegisterDefaults AMDGPURegBankCombinerHelper::getMode() {
   return MF.getInfo<SIMachineFunctionInfo>()->getMode();
 }
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-shift-shufflevector.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-shift-shufflevector.mir
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-shift-shufflevector.mir
@@ -0,0 +1,186 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn-amd-amdhsa -run-pass=amdgpu-regbank-combiner -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name:            lshr16_v2s16_mask10
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0
+    ; CHECK-LABEL: name: lshr16_v2s16_mask10
+    ; CHECK: liveins: $vgpr0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
+    ; CHECK-NEXT: $vgpr0 = COPY [[BITCAST]](s32)
+    %0:_(<2 x s16>) = COPY $vgpr0
+    %1:_(<2 x s16>) = G_IMPLICIT_DEF
+    %2:_(<2 x s16>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 0)
+    %3:_(s32) = G_BITCAST %2
+    %4:_(s32) = G_CONSTANT i32 16
+    %5:_(s32) = G_LSHR %3, %4
+    $vgpr0 = COPY %5
+...
+
+---
+name:            lshr8_v2s16_mask10_nofold
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0
+    ; CHECK-LABEL: name: lshr8_v2s16_mask10_nofold
+    ; CHECK: liveins: $vgpr0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+    ; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
+    ; CHECK-NEXT: [[SHUF:%[0-9]+]]:_(<2 x s16>) = G_SHUFFLE_VECTOR [[COPY]](<2 x s16>), [[DEF]], shufflemask(1, 0)
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[SHUF]](<2 x s16>)
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 8
+    ; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
+    ; CHECK-NEXT: $vgpr0 = COPY [[LSHR]](s32)
+    %0:_(<2 x s16>) = COPY $vgpr0
+    %1:_(<2 x s16>) = G_IMPLICIT_DEF
+    %2:_(<2 x s16>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 0)
+    %3:_(s32) = G_BITCAST %2
+    %4:_(s32) = G_CONSTANT i32 8
+    %5:_(s32) = G_LSHR %3, %4
+    $vgpr0 = COPY %5
+...
+
+---
+name:            lshr16_v2s16_mask11
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0
+    ; CHECK-LABEL: name: lshr16_v2s16_mask11
+    ; CHECK: liveins: $vgpr0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
+    ; CHECK-NEXT: $vgpr0 = COPY [[BITCAST]](s32)
+    %0:_(<2 x s16>) = COPY $vgpr0
+    %1:_(<2 x s16>) = G_IMPLICIT_DEF
+    %2:_(<2 x s16>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 1)
+    %3:_(s32) = G_BITCAST %2
+    %4:_(s32) = G_CONSTANT i32 16
+    %5:_(s32) = G_LSHR %3, %4
+    $vgpr0 = COPY %5
+...
+
+---
+name:            lshr16_v2s16_mask01_nofold
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0
+    ; CHECK-LABEL: name: lshr16_v2s16_mask01_nofold
+    ; CHECK: liveins: $vgpr0
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+    ; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
+    ; CHECK-NEXT: [[SHUF:%[0-9]+]]:_(<2 x s16>) = G_SHUFFLE_VECTOR [[COPY]](<2 x s16>), [[DEF]], shufflemask(0, 1)
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[SHUF]](<2 x s16>)
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
+    ; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
+    ; CHECK-NEXT: $vgpr0 = COPY [[LSHR]](s32)
+    %0:_(<2 x s16>) = COPY $vgpr0
+    %1:_(<2 x s16>) = G_IMPLICIT_DEF
+    %2:_(<2 x s16>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(0, 1)
+    %3:_(s32) = G_BITCAST %2
+    %4:_(s32) = G_CONSTANT i32 16
+    %5:_(s32) = G_LSHR %3, %4
+    $vgpr0 = COPY %5
+...
+
+---
+name:            lshr32_v2s32_mask10
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+    ; CHECK-LABEL: name: lshr32_v2s32_mask10
+    ; CHECK: liveins: $vgpr0_vgpr1
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s64) = G_BITCAST [[COPY]](<2 x s32>)
+    ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[BITCAST]](s64)
+    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    %1:_(<2 x s32>) = G_IMPLICIT_DEF
+    %2:_(<2 x s32>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 0)
+    %3:_(s64) = G_BITCAST %2
+    %4:_(s64) = G_CONSTANT i64 32
+    %5:_(s64) = G_LSHR %3, %4
+    $vgpr0_vgpr1 = COPY %5
+...
+
+---
+name:            lshr16_v2s32_mask10_nofold
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+    ; CHECK-LABEL: name: lshr16_v2s32_mask10_nofold
+    ; CHECK: liveins: $vgpr0_vgpr1
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    ; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<2 x s32>) = G_IMPLICIT_DEF
+    ; CHECK-NEXT: [[SHUF:%[0-9]+]]:_(<2 x s32>) = G_SHUFFLE_VECTOR [[COPY]](<2 x s32>), [[DEF]], shufflemask(1, 0)
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s64) = G_BITCAST [[SHUF]](<2 x s32>)
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 16
+    ; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(s64) = G_LSHR [[BITCAST]], [[C]](s64)
+    ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[LSHR]](s64)
+    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    %1:_(<2 x s32>) = G_IMPLICIT_DEF
+    %2:_(<2 x s32>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 0)
+    %3:_(s64) = G_BITCAST %2
+    %4:_(s64) = G_CONSTANT i64 16
+    %5:_(s64) = G_LSHR %3, %4
+    $vgpr0_vgpr1 = COPY %5
+...
+
+---
+name:            lshr32_v2s32_mask11
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+    ; CHECK-LABEL: name: lshr32_v2s32_mask11
+    ; CHECK: liveins: $vgpr0_vgpr1
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s64) = G_BITCAST [[COPY]](<2 x s32>)
+    ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[BITCAST]](s64)
+    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    %1:_(<2 x s32>) = G_IMPLICIT_DEF
+    %2:_(<2 x s32>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(1, 1)
+    %3:_(s64) = G_BITCAST %2
+    %4:_(s64) = G_CONSTANT i64 32
+    %5:_(s64) = G_LSHR %3, %4
+    $vgpr0_vgpr1 = COPY %5
+...
+
+---
+name:            lshr32_v2s32_mask01_nofold
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1
+    ; CHECK-LABEL: name: lshr32_v2s32_mask01_nofold
+    ; CHECK: liveins: $vgpr0_vgpr1
+    ; CHECK-NEXT: {{  $}}
+    ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    ; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<2 x s32>) = G_IMPLICIT_DEF
+    ; CHECK-NEXT: [[SHUF:%[0-9]+]]:_(<2 x s32>) = G_SHUFFLE_VECTOR [[COPY]](<2 x s32>), [[DEF]], shufflemask(0, 1)
+    ; CHECK-NEXT: [[BITCAST:%[0-9]+]]:_(s64) = G_BITCAST [[SHUF]](<2 x s32>)
+    ; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 32
+    ; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(s64) = G_LSHR [[BITCAST]], [[C]](s64)
+    ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[LSHR]](s64)
+    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
+    %1:_(<2 x s32>) = G_IMPLICIT_DEF
+    %2:_(<2 x s32>) = G_SHUFFLE_VECTOR %0:_, %1:_, shufflemask(0, 1)
+    %3:_(s64) = G_BITCAST %2
+    %4:_(s64) = G_CONSTANT i64 32
+    %5:_(s64) = G_LSHR %3, %4
+    $vgpr0_vgpr1 = COPY %5
+...