Index: lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
+++ lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
@@ -1696,6 +1696,75 @@
                            OpsToWaterfall, MRI);
     return;
   }
+  case AMDGPU::G_INSERT_VECTOR_ELT: {
+    SmallVector<Register, 2> InsRegs(OpdMapper.getVRegs(2));
+
+    assert(empty(OpdMapper.getVRegs(0)));
+    assert(empty(OpdMapper.getVRegs(1)));
+    assert(empty(OpdMapper.getVRegs(3)));
+
+    if (InsRegs.empty()) {
+      applyDefaultMapping(OpdMapper);
+      executeInWaterfallLoop(MI, MRI, { 3 });
+      return;
+    }
+
+    Register DstReg = MI.getOperand(0).getReg();
+    Register SrcReg = MI.getOperand(1).getReg();
+    Register InsReg = MI.getOperand(2).getReg();
+    Register IdxReg = MI.getOperand(3).getReg();
+    LLT SrcTy = MRI.getType(SrcReg);
+    LLT InsTy = MRI.getType(InsReg);
+
+    assert(InsTy.getSizeInBits() == 64);
+
+    const LLT S32 = LLT::scalar(32);
+    LLT Vec32 = LLT::vector(2 * SrcTy.getNumElements(), 32);
+
+    MachineIRBuilder B(MI);
+    auto CastSrc = B.buildBitcast(Vec32, SrcReg);
+    auto One = B.buildConstant(S32, 1);
+
+    // Split the vector index into 32-bit pieces. Prepare to move all of the
+    // new instructions into a waterfall loop if necessary.
+    //
+    // Don't put the bitcast or constant in the loop.
+    MachineInstrSpan Span(MachineBasicBlock::iterator(&MI), &B.getMBB());
+
+    // Compute 32-bit element indices, (2 * OrigIdx, 2 * OrigIdx + 1).
+    auto IdxLo = B.buildShl(S32, IdxReg, One);
+    auto IdxHi = B.buildAdd(S32, IdxLo, One);
+
+    auto InsLo = B.buildInsertVectorElement(Vec32, CastSrc, InsRegs[0], IdxLo);
+    auto InsHi = B.buildInsertVectorElement(Vec32, InsLo, InsRegs[1], IdxHi);
+    B.buildBitcast(DstReg, InsHi);
+
+    const RegisterBank *DstBank = getRegBank(DstReg, MRI, *TRI);
+    const RegisterBank *SrcBank = getRegBank(SrcReg, MRI, *TRI);
+    const RegisterBank *InsSrcBank = getRegBank(InsReg, MRI, *TRI);
+
+    MRI.setRegBank(InsReg, *InsSrcBank);
+    MRI.setRegBank(CastSrc.getReg(0), *SrcBank);
+    MRI.setRegBank(InsLo.getReg(0), *DstBank);
+    MRI.setRegBank(InsHi.getReg(0), *DstBank);
+    MRI.setRegBank(One.getReg(0), AMDGPU::SGPRRegBank);
+    MRI.setRegBank(IdxLo.getReg(0), AMDGPU::SGPRRegBank);
+    MRI.setRegBank(IdxHi.getReg(0), AMDGPU::SGPRRegBank);
+
+
+    SmallSet<Register, 4> OpsToWaterfall;
+    if (!collectWaterfallOperands(OpsToWaterfall, MI, MRI, { 3 })) {
+      MI.eraseFromParent();
+      return;
+    }
+
+    B.setInstr(*Span.begin());
+    MI.eraseFromParent();
+
+    executeInWaterfallLoop(B, make_range(Span.begin(), Span.end()),
+                           OpsToWaterfall, MRI);
+    return;
+  }
   case AMDGPU::G_INTRINSIC: {
     switch (MI.getIntrinsicID()) {
     case Intrinsic::amdgcn_s_buffer_load: {
@@ -2421,15 +2490,18 @@
     unsigned VecSize = MRI.getType(MI.getOperand(0).getReg()).getSizeInBits();
     unsigned InsertSize = MRI.getType(MI.getOperand(2).getReg()).getSizeInBits();
     unsigned IdxSize = MRI.getType(MI.getOperand(3).getReg()).getSizeInBits();
-    unsigned InsertEltBank = getRegBankID(MI.getOperand(2).getReg(), MRI, *TRI);
-    unsigned IdxBank = getRegBankID(MI.getOperand(3).getReg(), MRI, *TRI);
+    unsigned SrcBankID = getRegBankID(MI.getOperand(1).getReg(), MRI, *TRI);
+    unsigned InsertEltBankID = getRegBankID(MI.getOperand(2).getReg(),
+                                            MRI, *TRI);
+    unsigned IdxBankID = getRegBankID(MI.getOperand(3).getReg(), MRI, *TRI);
 
     OpdsMapping[0] = AMDGPU::getValueMapping(OutputBankID, VecSize);
-    OpdsMapping[1] = AMDGPU::getValueMapping(OutputBankID, VecSize);
-    OpdsMapping[2] = AMDGPU::getValueMapping(InsertEltBank, InsertSize);
+    OpdsMapping[1] = AMDGPU::getValueMapping(SrcBankID, VecSize);
+    OpdsMapping[2] = AMDGPU::getValueMappingSGPR64Only(InsertEltBankID,
+                                                       InsertSize);
 
     // The index can be either if the source vector is VGPR.
-    OpdsMapping[3] = AMDGPU::getValueMapping(IdxBank, IdxSize);
+    OpdsMapping[3] = AMDGPU::getValueMapping(IdxBankID, IdxSize);
     break;
   }
   case AMDGPU::G_UNMERGE_VALUES: {
Index: test/CodeGen/AMDGPU/GlobalISel/regbankselect-insert-vector-elt.mir
===================================================================
--- test/CodeGen/AMDGPU/GlobalISel/regbankselect-insert-vector-elt.mir
+++ test/CodeGen/AMDGPU/GlobalISel/regbankselect-insert-vector-elt.mir
@@ -56,8 +56,7 @@
     ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
     ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[COPY2]](s32)
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
     ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $vgpr0
@@ -69,17 +68,35 @@
 ---
 name: insert_vector_elt_v4i32_s_s_v
 legalized: true
+tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
 
     ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_s_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
     ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[COPY2]](s32)
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
     ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $sgpr4
@@ -91,17 +108,35 @@
 ---
 name: insert_vector_elt_v4i32_s_v_v
 legalized: true
+tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
 
     ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_v_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
     ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[COPY2]](s32)
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr1
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
     ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $vgpr0
@@ -113,16 +148,35 @@
 ---
 name: insert_vector_elt_var_v4i32_v_s_v
 legalized: true
+tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
 
     ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_s_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
     ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
     ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $sgpr4
@@ -134,12 +188,14 @@
 ---
 name: insert_vector_elt_var_v4i32_v_v_s
 legalized: true
+tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
 
     ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_v_s
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
     ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
     ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
@@ -155,16 +211,35 @@
 ---
 name: insert_vector_elt_var_v4i32_v_v_v
 legalized: true
+tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
 
     ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_v_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
     ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr5
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr5
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
     ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $vgpr4
@@ -172,3 +247,299 @@
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
+
+---
+name: insert_vector_elt_v8s64_s_s_s
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_s_s
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
+    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; CHECK: [[IVEC:%[0-9]+]]:sgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[COPY2]](s32)
+    ; CHECK: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY [[IVEC]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    %1:_(s64) = COPY $sgpr16_sgpr17
+    %2:_(s32) = COPY $sgpr18
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_v_s_s
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_s_s
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
+    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[COPY2]](s32)
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[IVEC]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    %1:_(s64) = COPY $sgpr16_sgpr17
+    %2:_(s32) = COPY $sgpr18
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_s_v_s
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_v_s
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
+    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr16
+    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; CHECK: [[BITCAST:%[0-9]+]]:sgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    %1:_(s64) = COPY $vgpr0_vgpr1
+    %2:_(s32) = COPY $sgpr16
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_s_s_v
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_s_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
+    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<8 x s64>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<8 x s64>) = G_PHI [[DEF]](<8 x s64>), %bb.0, %3(<8 x s64>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[IVEC]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    %1:_(s64) = COPY $sgpr16_sgpr17
+    %2:_(s32) = COPY $vgpr0
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_s_v_v
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_v_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
+    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr2
+    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; CHECK: [[BITCAST:%[0-9]+]]:sgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF4:%[0-9]+]]:vgpr(<8 x s64>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF5:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF5]], %bb.0, %25, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %8(s32), %bb.1
+    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %9(s32), %bb.1
+    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %10(<16 x s32>), %bb.1
+    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
+    ; CHECK: [[PHI5:%[0-9]+]]:vgpr(<8 x s64>) = G_PHI [[DEF4]](<8 x s64>), %bb.0, %3(<8 x s64>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
+    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    %1:_(s64) = COPY $vgpr0_vgpr1
+    %2:_(s32) = COPY $vgpr2
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_v_v_s
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_v_s
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
+    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    %1:_(s64) = COPY $vgpr16_vgpr17
+    %2:_(s32) = COPY $sgpr18
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_v_s_v
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_s_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
+    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr16
+    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<8 x s64>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF1]], %bb.0, %9, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<8 x s64>) = G_PHI [[DEF]](<8 x s64>), %bb.0, %3(<8 x s64>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[V_READFIRSTLANE_B32_]](s32)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[IVEC]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    %1:_(s64) = COPY $sgpr0_sgpr1
+    %2:_(s32) = COPY $vgpr16
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...
+
+---
+name: insert_vector_elt_v8s64_v_v_v
+legalized: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+
+    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_v_v
+    ; CHECK: successors: %bb.1(0x80000000)
+    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr18
+    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF4:%[0-9]+]]:vgpr(<8 x s64>) = G_IMPLICIT_DEF
+    ; CHECK: [[DEF5:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
+    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
+    ; CHECK: .1:
+    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+    ; CHECK: [[PHI:%[0-9]+]]:sreg_64 = PHI [[DEF5]], %bb.0, %25, %bb.1
+    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %8(s32), %bb.1
+    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %9(s32), %bb.1
+    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %10(<16 x s32>), %bb.1
+    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
+    ; CHECK: [[PHI5:%[0-9]+]]:vgpr(<8 x s64>) = G_PHI [[DEF4]](<8 x s64>), %bb.0, %3(<8 x s64>), %bb.1
+    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
+    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
+    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
+    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64 = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
+    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
+    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
+    ; CHECK: .2:
+    ; CHECK: successors: %bb.3(0x80000000)
+    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
+    ; CHECK: .3:
+    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    %1:_(s64) = COPY $vgpr16_vgpr17
+    %2:_(s32) = COPY $vgpr18
+    %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
+    $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
+...