[AArch64][SVE] Improve code generation for vector_splice for Imm == -1

[CarolineConcatto]

Summary

This patch implements vector_splice in tablegen for:

a) when the immediate is equal to -1 (Imm==1) and uses:
     INSR  +  LASTB

For instance :
@llvm.experimental.vector.splice(Vector_1, Vector_2, -1)
@llvm.experimental.vector.splice(<A,B,C,D>, <E,F,G,H>, 1) ==> <D, E, F, G>

LAST   RegLast, Vector_1                // RegLast = D
INSR   Res, (Vec2Rev >> 1), RegLast     // Res = D + E, F, G

[sdesmalen]

Hi @CarolineConcatto,

• For a splice with a positive index, we can use the EXT instruction (if the index fits the immediate range). Because EXT operates on i8 element types, you'll need to take into account the scaling. You can use a ComplexPattern to test if the immediate is in range and then have it scaled, see for example sve_cnt_mul_imm which uses SelectCntImm to check whether the immediate is in range, and then scales the value.
• For a splice with a negative index, rather than using a REV you could generate a predicate that is: 0, 1, 1, ... 1 using (not(whilelt XZR, -index)).

The most important cases to optimize are: -1 and +1, as those will occur most in practice. In our downstream compiler, we use WHILE+LASTB+INSR instead of SPLICE for the -1 case (which I now realise may save the one not instruction I mention above).

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

7401

If all we do is return Op, then you can just as well say it's Legal. However, in my other comment I'm suggesting to do custom-lowering to bitcast the types, so that we don't need as many patterns.

llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td

1242

The type isn't actually relevant for splice, although the container width is. If we can lower the splice operation to always work on integer types (converting the input fp vector, and a casting the result back to the original fp type afterwards), then we don't need patterns for all these cases.

[CarolineConcatto]

• Address Sander's suggestion and use EXT when lane/Imm is bigger than 0
• and use INSR + LAST for lane/Imm equals to -1

[CarolineConcatto]

• Implement SelectImm to use with EXT in attemp to scale index

[CarolineConcatto]

• Split the patch to handles only Immediate == -1
• Move cast Float from/to Integer into DAGCombiner.cpp

[CarolineConcatto]

• Move bitcast for AArch64ISelLowering

[sdesmalen]

Can [WIP] be removed from the title?

[sdesmalen]

nit on the commit message:

INSR   Res, (Vec2Rev >> 1), RegLast     // Res = D + G, F, E

should be:

INSR   Res, (Vec2Rev >> 1), RegLast     // Res = D + E, F, G

[CarolineConcatto]

• update commit message with --verbatim

[sdesmalen]

LGTM!

nit: The commit message still refers to Vec2Rev (I didn't spot that in my previous suggestion)

llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp

bool SelectCntImm(SDValue N, SDValue &Imm) {
     if ((MulImm >= Min) && (MulImm <= Max)) {
       Imm = CurDAG->getTargetConstant(MulImm, SDLoc(N), MVT::i32);
       return true;
     }
 
     return false;
   }
 
+  template <signed Min, signed Max, signed Scale>
+  bool SelectImm(SDValue N, SDValue &Imm) {
+    if (!isa<ConstantSDNode>(N))
+      return false;
+
+    int64_t MulImm = cast<ConstantSDNode>(N)->getSExtValue();
+
+    MulImm *= Scale;
+    if ((MulImm >= Min * Scale) && (MulImm <= Max * Scale)) {
+      Imm = CurDAG->getTargetConstant(MulImm, SDLoc(N), MVT::i32);
+      return true;
+    }
+
+    return false;
+  }
+
   /// Form sequences of consecutive 64/128-bit registers for use in NEON
   /// instructions making use of a vector-list (e.g. ldN, tbl). Vecs must have
   /// between 1 and 4 elements. If it contains a single element that is returned
   /// unchanged; otherwise a REG_SEQUENCE value is returned.
   SDValue createDTuple(ArrayRef<SDValue> Vecs);
   SDValue createQTuple(ArrayRef<SDValue> Vecs);
   // Form a sequence of SVE registers for instructions using list of vectors,
   // e.g. structured loads and stores (ldN, stN).

llvm/lib/Target/AArch64/AArch64ISelLowering.h

private:
   SDValue LowerSCALAR_TO_VECTOR(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerSPLAT_VECTOR(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerDUPQLane(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerToPredicatedOp(SDValue Op, SelectionDAG &DAG, unsigned NewOp,
                               bool OverrideNEON = false) const;
   SDValue LowerToScalableOp(SDValue Op, SelectionDAG &DAG) const;
+  SDValue LowerVECTOR_SPLICE(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerEXTRACT_SUBVECTOR(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerINSERT_SUBVECTOR(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerDIV(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerMUL(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerVectorSRA_SRL_SHL(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerShiftParts(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerVSETCC(SDValue Op, SelectionDAG &DAG) const;
   SDValue LowerCTPOP(SDValue Op, SelectionDAG &DAG) const;

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

for (auto VT : {MVT::nxv16i8, MVT::nxv8i16, MVT::nxv4i32, MVT::nxv2i64}) {
       setOperationAction(ISD::FP_TO_UINT, VT, Custom);
       setOperationAction(ISD::FP_TO_SINT, VT, Custom);
       setOperationAction(ISD::MGATHER, VT, Custom);
       setOperationAction(ISD::MSCATTER, VT, Custom);
       setOperationAction(ISD::MUL, VT, Custom);
       setOperationAction(ISD::MULHS, VT, Custom);
       setOperationAction(ISD::MULHU, VT, Custom);
       setOperationAction(ISD::SPLAT_VECTOR, VT, Custom);
+      setOperationAction(ISD::VECTOR_SPLICE, VT, Custom);
       setOperationAction(ISD::SELECT, VT, Custom);
       setOperationAction(ISD::SETCC, VT, Custom);
       setOperationAction(ISD::SDIV, VT, Custom);
       setOperationAction(ISD::UDIV, VT, Custom);
       setOperationAction(ISD::SMIN, VT, Custom);
       setOperationAction(ISD::UMIN, VT, Custom);
       setOperationAction(ISD::SMAX, VT, Custom);
       setOperationAction(ISD::UMAX, VT, Custom);
for (auto VT : {MVT::nxv2f16, MVT::nxv4f16, MVT::nxv8f16, MVT::nxv2f32,
       setOperationAction(ISD::FSQRT, VT, Custom);
       setOperationAction(ISD::FABS, VT, Custom);
       setOperationAction(ISD::FP_EXTEND, VT, Custom);
       setOperationAction(ISD::FP_ROUND, VT, Custom);
       setOperationAction(ISD::VECREDUCE_FADD, VT, Custom);
       setOperationAction(ISD::VECREDUCE_FMAX, VT, Custom);
       setOperationAction(ISD::VECREDUCE_FMIN, VT, Custom);
       setOperationAction(ISD::VECREDUCE_SEQ_FADD, VT, Custom);
+      setOperationAction(ISD::VECTOR_SPLICE, VT, Custom);
 
       setOperationAction(ISD::SELECT_CC, VT, Expand);
     }
 
     for (auto VT : {MVT::nxv2bf16, MVT::nxv4bf16, MVT::nxv8bf16}) {
       setOperationAction(ISD::CONCAT_VECTORS, VT, Custom);
       setOperationAction(ISD::MGATHER, VT, Custom);
       setOperationAction(ISD::MSCATTER, VT, Custom);
void AArch64TargetLowering::addTypeForFixedLengthSVE(MVT VT) {
   setOperationAction(ISD::SETCC, VT, Custom);
   setOperationAction(ISD::SHL, VT, Custom);
   setOperationAction(ISD::SIGN_EXTEND, VT, Custom);
   setOperationAction(ISD::SIGN_EXTEND_INREG, VT, Custom);
   setOperationAction(ISD::SINT_TO_FP, VT, Custom);
   setOperationAction(ISD::SMAX, VT, Custom);
   setOperationAction(ISD::SMIN, VT, Custom);
   setOperationAction(ISD::SPLAT_VECTOR, VT, Custom);
+  setOperationAction(ISD::VECTOR_SPLICE, VT, Custom);
   setOperationAction(ISD::SRA, VT, Custom);
   setOperationAction(ISD::SRL, VT, Custom);
   setOperationAction(ISD::STORE, VT, Custom);
   setOperationAction(ISD::SUB, VT, Custom);
   setOperationAction(ISD::TRUNCATE, VT, Custom);
   setOperationAction(ISD::UDIV, VT, Custom);
   setOperationAction(ISD::UINT_TO_FP, VT, Custom);
   setOperationAction(ISD::UMAX, VT, Custom);
case ISD::BITREVERSE:
     return LowerBitreverse(Op, DAG);
   case ISD::BSWAP:
     return LowerToPredicatedOp(Op, DAG, AArch64ISD::BSWAP_MERGE_PASSTHRU);
   case ISD::CTLZ:
     return LowerToPredicatedOp(Op, DAG, AArch64ISD::CTLZ_MERGE_PASSTHRU,
                                /*OverrideNEON=*/true);
   case ISD::CTTZ:
     return LowerCTTZ(Op, DAG);
+  case ISD::VECTOR_SPLICE:
+    return LowerVECTOR_SPLICE(Op, DAG);
   }
 }
 
 bool AArch64TargetLowering::mergeStoresAfterLegalization(EVT VT) const {
   return !Subtarget->useSVEForFixedLengthVectors();
 }
 
 bool AArch64TargetLowering::useSVEForFixedLengthVectorVT(
if (CC2 != AArch64CC::AL) {
     SDValue CC2Val = DAG.getConstant(CC2, dl, MVT::i32);
     return DAG.getNode(AArch64ISD::CSEL, dl, VT, TVal, CS1, CC2Val, Cmp);
   }
 
   // Otherwise, return the output of the first CSEL.
   return CS1;
 }
 
+SDValue AArch64TargetLowering::LowerVECTOR_SPLICE(SDValue Op,
+                                                  SelectionDAG &DAG) const {
+
+  EVT Ty = Op.getValueType();
+  assert(Ty.isScalableVector() && "Expected Scalable Type");
+  EVT VTy = Ty.isFloatingPoint()
+                ? getPackedSVEVectorVT(Ty.getVectorElementCount())
+                : Ty;
+
+  SDValue LHS = Ty.isFloatingPoint() ? DAG.getBitcast(VTy, Op.getOperand(0))
+                                     : Op.getOperand(0);
+  SDValue RHS = Ty.isFloatingPoint() ? DAG.getBitcast(VTy, Op.getOperand(1))
+                                     : Op.getOperand(1);
+  SDValue Idx = Op.getOperand(2);
+  SDLoc DL(Op);
+  if ((!Op.getConstantOperandAPInt(2).isNegative() &&
+       Ty.getVectorMinNumElements() > Op.getConstantOperandVal(2)) ||
+      Op.getConstantOperandAPInt(2) == -1) {
+    SDValue Splice = DAG.getNode(ISD::VECTOR_SPLICE, DL, VTy, LHS, RHS, Idx);
+    if (Ty.isFloatingPoint())
+      return DAG.getNode(ISD::BITCAST, DL, Ty, Splice);
+    return Splice;
+  }
+  return SDValue();
+}
+
 SDValue AArch64TargetLowering::LowerSELECT_CC(SDValue Op,
                                               SelectionDAG &DAG) const {
   ISD::CondCode CC = cast<CondCodeSDNode>(Op.getOperand(4))->get();
   SDValue LHS = Op.getOperand(0);
   SDValue RHS = Op.getOperand(1);
   SDValue TVal = Op.getOperand(2);
   SDValue FVal = Op.getOperand(3);
   SDLoc DL(Op);

llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td

def : Pat<(nxv8f16 (concat_vectors nxv4f16:$v1, nxv4f16:$v2)),
             (UZP1_ZZZ_H $v1, $v2)>;
   def : Pat<(nxv4f32 (concat_vectors nxv2f32:$v1, nxv2f32:$v2)),
             (UZP1_ZZZ_S $v1, $v2)>;
   def : Pat<(nxv4bf16 (concat_vectors nxv2bf16:$v1, nxv2bf16:$v2)),
             (UZP1_ZZZ_S $v1, $v2)>;
   def : Pat<(nxv8bf16 (concat_vectors nxv4bf16:$v1, nxv4bf16:$v2)),
             (UZP1_ZZZ_H $v1, $v2)>;
 
+  // Splice with lane equal to -1
+  def : Pat<(nxv16i8 (vector_splice (nxv16i8 ZPR:$Z1), (nxv16i8 ZPR:$Z2), (i64 -1))),
+            (INSR_ZV_B (REV_ZZ_B ZPR:$Z2), (INSERT_SUBREG (IMPLICIT_DEF),
+            (LASTB_VPZ_B (PTRUE_B 31), ZPR:$Z1), bsub))>;
+  def : Pat<(nxv8i16 (vector_splice (nxv8i16 ZPR:$Z1), (nxv8i16 ZPR:$Z2), (i64 -1))),
+            (INSR_ZV_H (REV_ZZ_H ZPR:$Z2), (INSERT_SUBREG (IMPLICIT_DEF),
+            (LASTB_VPZ_H (PTRUE_H 31), ZPR:$Z1), hsub))>;
+  def : Pat<(nxv4i32 (vector_splice (nxv4i32 ZPR:$Z1), (nxv4i32 ZPR:$Z2), (i64 -1))),
+            (INSR_ZV_S (REV_ZZ_S ZPR:$Z2), (INSERT_SUBREG (IMPLICIT_DEF),
+            (LASTB_VPZ_S (PTRUE_S 31), ZPR:$Z1), ssub))>;
+  def : Pat<(nxv2i64 (vector_splice (nxv2i64 ZPR:$Z1), (nxv2i64 ZPR:$Z2), (i64 -1))),
+            (INSR_ZV_D (REV_ZZ_D ZPR:$Z2), (INSERT_SUBREG (IMPLICIT_DEF),
+            (LASTB_VPZ_D (PTRUE_D 31), ZPR:$Z1), dsub))>;
+
   defm CMPHS_PPzZZ : sve_int_cmp_0<0b000, "cmphs", SETUGE, SETULE>;
   defm CMPHI_PPzZZ : sve_int_cmp_0<0b001, "cmphi", SETUGT, SETULT>;
   defm CMPGE_PPzZZ : sve_int_cmp_0<0b100, "cmpge", SETGE, SETLE>;
   defm CMPGT_PPzZZ : sve_int_cmp_0<0b101, "cmpgt", SETGT, SETLT>;
   defm CMPEQ_PPzZZ : sve_int_cmp_0<0b110, "cmpeq", SETEQ, SETEQ>;
   defm CMPNE_PPzZZ : sve_int_cmp_0<0b111, "cmpne", SETNE, SETNE>;
 
   defm CMPEQ_WIDE_PPzZZ : sve_int_cmp_0_wide<0b010, "cmpeq", int_aarch64_sve_cmpeq_wide>;
def : Pat<(vector_extract (nxv2f16 ZPR:$Zs), (i64 0)),
             (f16 (EXTRACT_SUBREG ZPR:$Zs, hsub))>;
   def : Pat<(vector_extract (nxv4f32 ZPR:$Zs), (i64 0)),
             (f32 (EXTRACT_SUBREG ZPR:$Zs, ssub))>;
   def : Pat<(vector_extract (nxv2f32 ZPR:$Zs), (i64 0)),
             (f32 (EXTRACT_SUBREG ZPR:$Zs, ssub))>;
   def : Pat<(vector_extract (nxv2f64 ZPR:$Zs), (i64 0)),
             (f64 (EXTRACT_SUBREG ZPR:$Zs, dsub))>;
   }
+
+  // Splice with lane bigger or equalt to 0
+  def : Pat<(nxv16i8 (vector_splice (nxv16i8 ZPR:$Z1), (nxv16i8 ZPR:$Z2), (i64 (sve_imm_0_16 i32:$index)))),
+            (EXT_ZZI  ZPR:$Z1, ZPR:$Z2, sve_imm_0_16:$index)>;
+  def : Pat<(nxv8i16 (vector_splice (nxv8i16 ZPR:$Z1), (nxv8i16 ZPR:$Z2), (i64 (sve_imm_0_8 i32:$index)))),
+            (EXT_ZZI  ZPR:$Z1, ZPR:$Z2, sve_imm_0_8:$index)>;
+  def : Pat<(nxv4i32 (vector_splice (nxv4i32 ZPR:$Z1), (nxv4i32 ZPR:$Z2), (i64 (sve_imm_0_4 i32:$index)))),
+           (EXT_ZZI  ZPR:$Z1, ZPR:$Z2, sve_imm_0_4:$index)>;
+  def : Pat<(nxv2i64 (vector_splice (nxv2i64 ZPR:$Z1), (nxv2i64 ZPR:$Z2), (i64 (sve_imm_0_2 i32:$index)))),
+          (EXT_ZZI  ZPR:$Z1, ZPR:$Z2, sve_imm_0_2:$index)>;
+
 }
 
 let Predicates = [HasSVE, HasMatMulInt8] in {
   defm  SMMLA_ZZZ : sve_int_matmul<0b00, "smmla", int_aarch64_sve_smmla>;
   defm  UMMLA_ZZZ : sve_int_matmul<0b11, "ummla", int_aarch64_sve_ummla>;
   defm USMMLA_ZZZ : sve_int_matmul<0b10, "usmmla", int_aarch64_sve_usmmla>;
   defm USDOT_ZZZ  : sve_int_dot_mixed<"usdot", int_aarch64_sve_usdot>;
   defm USDOT_ZZZI : sve_int_dot_mixed_indexed<0, "usdot", int_aarch64_sve_usdot_lane>;

llvm/lib/Target/AArch64/SVEInstrFormats.td

def sve_incdec_imm : Operand<i32>, TImmLeaf<i32, [{
   let EncoderMethod = "getSVEIncDecImm";
   let DecoderMethod = "DecodeSVEIncDecImm";
 }
 
 // This allows i32 immediate extraction from i64 based arithmetic.
 def sve_cnt_mul_imm : ComplexPattern<i32, 1, "SelectCntImm<1, 16, 1, false>">;
 def sve_cnt_shl_imm : ComplexPattern<i32, 1, "SelectCntImm<1, 16, 1, true>">;
 
+def sve_imm_0_2  : ComplexPattern<i32, 1, "SelectImm<0, 2, 8>">;
+def sve_imm_0_4  : ComplexPattern<i32, 1, "SelectImm<0, 4, 4>">;
+def sve_imm_0_8  : ComplexPattern<i32, 1, "SelectImm<0, 8, 2>">;
+def sve_imm_0_16 : ComplexPattern<i32, 1, "SelectImm<0, 16, 1>">;
+
 def int_aarch64_sve_cntp_oneuse : PatFrag<(ops node:$pred, node:$src2),
                                           (int_aarch64_sve_cntp node:$pred, node:$src2), [{
   return N->hasOneUse();
 }]>;
 
 //===----------------------------------------------------------------------===//
 // SVE PTrue - These are used extensively throughout the pattern matching so
 //             it's important we define them first.

llvm/test/CodeGen/AArch64/named-vector-shuffles-sve.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -verify-machineinstrs < %s | FileCheck %s
 
 target triple = "aarch64-unknown-linux-gnu"
 
 ;
 ; VECTOR_SPLICE (index)
 ;
 
 define <vscale x 16 x i8> @splice_nxv16i8_first_idx(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i8_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1b { z0.b }, p0, [sp]
-; CHECK-NEXT:    st1b { z1.b }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1b { z0.b }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i8> @llvm.experimental.vector.splice.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, i32 0)
   ret <vscale x 16 x i8> %res
 }
 
 define <vscale x 16 x i8> @splice_nxv16i8_last_idx(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i8_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1b { z0.b }, p0, [sp]
-; CHECK-NEXT:    st1b { z1.b }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0xf
-; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #15
 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i8> @llvm.experimental.vector.splice.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, i32 15)
   ret <vscale x 16 x i8> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 16 x i8> @splice_nxv16i8_clamped_idx(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i8_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i8> @llvm.experimental.vector.splice.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, i32 16)
   ret <vscale x 16 x i8> %res
 }
 
 define <vscale x 8 x i16> @splice_nxv8i16_first_idx(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i16_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1h { z0.h }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i16> @llvm.experimental.vector.splice.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, i32 0)
   ret <vscale x 8 x i16> %res
 }
 
 define <vscale x 8 x i16> @splice_nxv8i16_last_idx(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i16_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0xe
-; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #14
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i16> @llvm.experimental.vector.splice.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, i32 7)
   ret <vscale x 8 x i16> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 8 x i16> @splice_nxv8i16_clamped_idx(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i16_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i16> @llvm.experimental.vector.splice.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, i32 8)
   ret <vscale x 8 x i16> %res
 }
 
 define <vscale x 4 x i32> @splice_nxv4i32_first_idx(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i32_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1w { z0.s }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, i32 0)
   ret <vscale x 4 x i32> %res
 }
 
 define <vscale x 4 x i32> @splice_nxv4i32_last_idx(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i32_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0xc
-; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #12
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, i32 3)
   ret <vscale x 4 x i32> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 4 x i32> @splice_nxv4i32_clamped_idx(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i32_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, i32 4)
   ret <vscale x 4 x i32> %res
 }
 
 define <vscale x 2 x i64> @splice_nxv2i64_first_idx(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i64_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i64> @llvm.experimental.vector.splice.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, i32 0)
   ret <vscale x 2 x i64> %res
 }
 
 define <vscale x 2 x i64> @splice_nxv2i64_last_idx(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i64_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #8
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i64> @llvm.experimental.vector.splice.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, i32 1)
   ret <vscale x 2 x i64> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 2 x i64> @splice_nxv2i64_clamped_idx(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i64_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i64> @llvm.experimental.vector.splice.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, i32 2)
   ret <vscale x 2 x i64> %res
 }
 
 define <vscale x 8 x half> @splice_nxv8f16_first_idx(<vscale x 8 x half> %a, <vscale x 8 x half> %b) #0 {
 ; CHECK-LABEL: splice_nxv8f16_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1h { z0.h }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x half> @llvm.experimental.vector.splice.nxv8f16(<vscale x 8 x half> %a, <vscale x 8 x half> %b, i32 0)
   ret <vscale x 8 x half> %res
 }
 
 define <vscale x 8 x half> @splice_nxv8f16_last_idx(<vscale x 8 x half> %a, <vscale x 8 x half> %b) #0 {
 ; CHECK-LABEL: splice_nxv8f16_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0xe
-; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #14
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x half> @llvm.experimental.vector.splice.nxv8f16(<vscale x 8 x half> %a, <vscale x 8 x half> %b, i32 7)
   ret <vscale x 8 x half> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 8 x half> @splice_nxv8f16_clamped_idx(<vscale x 8 x half> %a, <vscale x 8 x half> %b) #0 {
 ; CHECK-LABEL: splice_nxv8f16_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x half> @llvm.experimental.vector.splice.nxv8f16(<vscale x 8 x half> %a, <vscale x 8 x half> %b, i32 8)
   ret <vscale x 8 x half> %res
 }
 
 define <vscale x 4 x float> @splice_nxv4f32_first_idx(<vscale x 4 x float> %a, <vscale x 4 x float> %b) #0 {
 ; CHECK-LABEL: splice_nxv4f32_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1w { z0.s }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x float> @llvm.experimental.vector.splice.nxv4f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b, i32 0)
   ret <vscale x 4 x float> %res
 }
 
 define <vscale x 4 x float> @splice_nxv4f32_last_idx(<vscale x 4 x float> %a, <vscale x 4 x float> %b) #0 {
 ; CHECK-LABEL: splice_nxv4f32_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0xc
-; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #12
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x float> @llvm.experimental.vector.splice.nxv4f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b, i32 3)
   ret <vscale x 4 x float> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 4 x float> @splice_nxv4f32_clamped_idx(<vscale x 4 x float> %a, <vscale x 4 x float> %b) #0 {
 ; CHECK-LABEL: splice_nxv4f32_clamped_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x float> @llvm.experimental.vector.splice.nxv4f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b, i32 4)
   ret <vscale x 4 x float> %res
 }
 
 define <vscale x 2 x double> @splice_nxv2f64_first_idx(<vscale x 2 x double> %a, <vscale x 2 x double> %b) #0 {
 ; CHECK-LABEL: splice_nxv2f64_first_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [sp]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 0)
   ret <vscale x 2 x double> %res
 }
 
 define <vscale x 2 x double> @splice_nxv2f64_last_idx(<vscale x 2 x double> %a, <vscale x 2 x double> %b) #0 {
 ; CHECK-LABEL: splice_nxv2f64_last_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #8
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 1)
   ret <vscale x 2 x double> %res
 }
 
 ; Ensure index is clamped when we cannot prove it's less than VL-1.
 define <vscale x 2 x double> @splice_nxv2f64_clamped_idx(<vscale x 2 x double> %a, <vscale x 2 x double> %b) #0 {
 ; CHECK-LABEL: splice_nxv2f64_clamped_idx:
; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 2)
   ret <vscale x 2 x double> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 2 x i1> @splice_nxv2i1_idx(<vscale x 2 x i1> %a, <vscale x 2 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i1_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    mov z0.d, p0/z, #1 // =0x1
+; CHECK-NEXT:    mov z0.d, p1/z, #1 // =0x1
+; CHECK-NEXT:    mov z1.d, p0/z, #1 // =0x1
+; CHECK-NEXT:    ext z1.b, z1.b, z0.b, #8
+; CHECK-NEXT:    and z1.d, z1.d, #0x1
 ; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov z1.d, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x8]
-; CHECK-NEXT:    and z0.d, z0.d, #0x1
-; CHECK-NEXT:    cmpne p0.d, p0/z, z0.d, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    cmpne p0.d, p0/z, z1.d, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i1> @llvm.experimental.vector.splice.nxv2i1(<vscale x 2 x i1> %a, <vscale x 2 x i1> %b, i32 1)
   ret <vscale x 2 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 4 x i1> @splice_nxv4i1_idx(<vscale x 4 x i1> %a, <vscale x 4 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i1_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    mov z0.s, p0/z, #1 // =0x1
+; CHECK-NEXT:    mov z0.s, p1/z, #1 // =0x1
+; CHECK-NEXT:    mov z1.s, p0/z, #1 // =0x1
+; CHECK-NEXT:    ext z1.b, z1.b, z0.b, #8
+; CHECK-NEXT:    and z1.s, z1.s, #0x1
 ; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov z1.s, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x8]
-; CHECK-NEXT:    and z0.s, z0.s, #0x1
-; CHECK-NEXT:    cmpne p0.s, p0/z, z0.s, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    cmpne p0.s, p0/z, z1.s, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i1> @llvm.experimental.vector.splice.nxv4i1(<vscale x 4 x i1> %a, <vscale x 4 x i1> %b, i32 2)
   ret <vscale x 4 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 8 x i1> @splice_nxv8i1_idx(<vscale x 8 x i1> %a, <vscale x 8 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i1_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    mov z0.h, p0/z, #1 // =0x1
+; CHECK-NEXT:    mov z0.h, p1/z, #1 // =0x1
+; CHECK-NEXT:    mov z1.h, p0/z, #1 // =0x1
+; CHECK-NEXT:    ext z1.b, z1.b, z0.b, #8
+; CHECK-NEXT:    and z1.h, z1.h, #0x1
 ; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov z1.h, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x8]
-; CHECK-NEXT:    and z0.h, z0.h, #0x1
-; CHECK-NEXT:    cmpne p0.h, p0/z, z0.h, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    cmpne p0.h, p0/z, z1.h, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i1> @llvm.experimental.vector.splice.nxv8i1(<vscale x 8 x i1> %a, <vscale x 8 x i1> %b, i32 4)
   ret <vscale x 8 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 16 x i1> @splice_nxv16i1_idx(<vscale x 16 x i1> %a, <vscale x 16 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i1_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    mov z0.b, p0/z, #1 // =0x1
+; CHECK-NEXT:    mov z0.b, p1/z, #1 // =0x1
+; CHECK-NEXT:    mov z1.b, p0/z, #1 // =0x1
+; CHECK-NEXT:    ext z1.b, z1.b, z0.b, #8
+; CHECK-NEXT:    and z1.b, z1.b, #0x1
 ; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    mov z1.b, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1b { z0.b }, p0, [sp]
-; CHECK-NEXT:    st1b { z1.b }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x8]
-; CHECK-NEXT:    and z0.b, z0.b, #0x1
-; CHECK-NEXT:    cmpne p0.b, p0/z, z0.b, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    cmpne p0.b, p0/z, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i1> @llvm.experimental.vector.splice.nxv16i1(<vscale x 16 x i1> %a, <vscale x 16 x i1> %b, i32 8)
   ret <vscale x 16 x i1> %res
 }
 
 ; Verify promote type legalisation works as expected.
 define <vscale x 2 x i8> @splice_nxv2i8_idx(<vscale x 2 x i8> %a, <vscale x 2 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i8_idx:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
-; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    orr x8, x8, #0x8
-; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x8]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ext z0.b, z0.b, z1.b, #8
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i8> @llvm.experimental.vector.splice.nxv2i8(<vscale x 2 x i8> %a, <vscale x 2 x i8> %b, i32 1)
   ret <vscale x 2 x i8> %res
 }
 
 ; Verify splitvec type legalisation works as expected.
 define <vscale x 8 x i32> @splice_nxv8i32_idx(<vscale x 8 x i32> %a, <vscale x 8 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i32_idx:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i8> @llvm.experimental.vector.splice.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, i32 -16)
   ret <vscale x 16 x i8> %res
 }
 
 define <vscale x 16 x i8> @splice_nxv16i8_1(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i8_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.b
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1b { z0.b }, p0, [sp]
-; CHECK-NEXT:    st1b { z1.b }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-1
-; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb b2, p0, z0.b
+; CHECK-NEXT:    rev z0.b, z1.b
+; CHECK-NEXT:    insr z0.b, b2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i8> @llvm.experimental.vector.splice.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, i32 -1)
   ret <vscale x 16 x i8> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 16 x i8> @splice_nxv16i8_clamped(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i8_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i16> @llvm.experimental.vector.splice.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, i32 -8)
   ret <vscale x 8 x i16> %res
 }
 
 define <vscale x 8 x i16> @splice_nxv8i16_1(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i16_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-2
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb h2, p0, z0.h
+; CHECK-NEXT:    rev z0.h, z1.h
+; CHECK-NEXT:    insr z0.h, h2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i16> @llvm.experimental.vector.splice.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, i32 -1)
   ret <vscale x 8 x i16> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 8 x i16> @splice_nxv8i16_clamped(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i16_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, i32 -4)
   ret <vscale x 4 x i32> %res
 }
 
 define <vscale x 4 x i32> @splice_nxv4i32_1(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i32_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-4
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb s2, p0, z0.s
+; CHECK-NEXT:    rev z0.s, z1.s
+; CHECK-NEXT:    insr z0.s, s2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, i32 -1)
   ret <vscale x 4 x i32> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 4 x i32> @splice_nxv4i32_clamped(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i32_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i64> @llvm.experimental.vector.splice.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, i32 -2)
   ret <vscale x 2 x i64> %res
 }
 
 define <vscale x 2 x i64> @splice_nxv2i64_1(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i64_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-8
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb d2, p0, z0.d
+; CHECK-NEXT:    rev z0.d, z1.d
+; CHECK-NEXT:    insr z0.d, d2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i64> @llvm.experimental.vector.splice.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, i32 -1)
   ret <vscale x 2 x i64> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 2 x i64> @splice_nxv2i64_clamped(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i64_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x half> @llvm.experimental.vector.splice.nxv8f16(<vscale x 8 x half> %a, <vscale x 8 x half> %b, i32 -8)
   ret <vscale x 8 x half> %res
 }
 
 define <vscale x 8 x half> @splice_nxv8f16_1(<vscale x 8 x half> %a, <vscale x 8 x half> %b) #0 {
 ; CHECK-LABEL: splice_nxv8f16_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.h
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-2
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb h2, p0, z0.h
+; CHECK-NEXT:    rev z0.h, z1.h
+; CHECK-NEXT:    insr z0.h, h2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x half> @llvm.experimental.vector.splice.nxv8f16(<vscale x 8 x half> %a, <vscale x 8 x half> %b, i32 -1)
   ret <vscale x 8 x half> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 8 x half> @splice_nxv8f16_clamped(<vscale x 8 x half> %a, <vscale x 8 x half> %b) #0 {
 ; CHECK-LABEL: splice_nxv8f16_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x float> @llvm.experimental.vector.splice.nxv4f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b, i32 -4)
   ret <vscale x 4 x float> %res
 }
 
 define <vscale x 4 x float> @splice_nxv4f32_1(<vscale x 4 x float> %a, <vscale x 4 x float> %b) #0 {
 ; CHECK-LABEL: splice_nxv4f32_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.s
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-4
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb s2, p0, z0.s
+; CHECK-NEXT:    rev z0.s, z1.s
+; CHECK-NEXT:    insr z0.s, s2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x float> @llvm.experimental.vector.splice.nxv4f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b, i32 -1)
   ret <vscale x 4 x float> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 4 x float> @splice_nxv4f32_clamped(<vscale x 4 x float> %a, <vscale x 4 x float> %b) #0 {
 ; CHECK-LABEL: splice_nxv4f32_clamped:

 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 -2)
   ret <vscale x 2 x double> %res
 }
 
 define <vscale x 2 x double> @splice_nxv2f64_1(<vscale x 2 x double> %a, <vscale x 2 x double> %b) #0 {
 ; CHECK-LABEL: splice_nxv2f64_1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-8
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb d2, p0, z0.d
+; CHECK-NEXT:    rev z0.d, z1.d
+; CHECK-NEXT:    insr z0.d, d2
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 -1)
   ret <vscale x 2 x double> %res
 }
 
 ; Ensure number of trailing elements is clamped when we cannot prove it's less than VL.
 define <vscale x 2 x double> @splice_nxv2f64_clamped(<vscale x 2 x double> %a, <vscale x 2 x double> %b) #0 {
 ; CHECK-LABEL: splice_nxv2f64_clamped:
; CHECK-NEXT:    ret
   %res = call <vscale x 2 x double> @llvm.experimental.vector.splice.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b, i32 -3)
   ret <vscale x 2 x double> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 2 x i1> @splice_nxv2i1(<vscale x 2 x i1> %a, <vscale x 2 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    mov z0.d, p0/z, #1 // =0x1
 ; CHECK-NEXT:    ptrue p0.d
 ; CHECK-NEXT:    mov z1.d, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1d { z0.d }, p0, [sp]
-; CHECK-NEXT:    st1d { z1.d }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-8
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    and z0.d, z0.d, #0x1
-; CHECK-NEXT:    cmpne p0.d, p0/z, z0.d, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb d0, p0, z0.d
+; CHECK-NEXT:    rev z1.d, z1.d
+; CHECK-NEXT:    insr z1.d, d0
+; CHECK-NEXT:    and z1.d, z1.d, #0x1
+; CHECK-NEXT:    cmpne p0.d, p0/z, z1.d, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 2 x i1> @llvm.experimental.vector.splice.nxv2i1(<vscale x 2 x i1> %a, <vscale x 2 x i1> %b, i32 -1)
   ret <vscale x 2 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 4 x i1> @splice_nxv4i1(<vscale x 4 x i1> %a, <vscale x 4 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv4i1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    mov z0.s, p0/z, #1 // =0x1
 ; CHECK-NEXT:    ptrue p0.s
 ; CHECK-NEXT:    mov z1.s, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1w { z0.s }, p0, [sp]
-; CHECK-NEXT:    st1w { z1.s }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-4
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    and z0.s, z0.s, #0x1
-; CHECK-NEXT:    cmpne p0.s, p0/z, z0.s, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb s0, p0, z0.s
+; CHECK-NEXT:    rev z1.s, z1.s
+; CHECK-NEXT:    insr z1.s, s0
+; CHECK-NEXT:    and z1.s, z1.s, #0x1
+; CHECK-NEXT:    cmpne p0.s, p0/z, z1.s, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i1> @llvm.experimental.vector.splice.nxv4i1(<vscale x 4 x i1> %a, <vscale x 4 x i1> %b, i32 -1)
   ret <vscale x 4 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 8 x i1> @splice_nxv8i1(<vscale x 8 x i1> %a, <vscale x 8 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv8i1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    mov z0.h, p0/z, #1 // =0x1
 ; CHECK-NEXT:    ptrue p0.h
 ; CHECK-NEXT:    mov z1.h, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    ptrue p1.b
-; CHECK-NEXT:    st1h { z0.h }, p0, [sp]
-; CHECK-NEXT:    st1h { z1.h }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-2
-; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x8, x9]
-; CHECK-NEXT:    and z0.h, z0.h, #0x1
-; CHECK-NEXT:    cmpne p0.h, p0/z, z0.h, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb h0, p0, z0.h
+; CHECK-NEXT:    rev z1.h, z1.h
+; CHECK-NEXT:    insr z1.h, h0
+; CHECK-NEXT:    and z1.h, z1.h, #0x1
+; CHECK-NEXT:    cmpne p0.h, p0/z, z1.h, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i1> @llvm.experimental.vector.splice.nxv8i1(<vscale x 8 x i1> %a, <vscale x 8 x i1> %b, i32 -1)
   ret <vscale x 8 x i1> %res
 }
 
 ; Ensure predicate based splice is promoted to use ZPRs.
 define <vscale x 16 x i1> @splice_nxv16i1(<vscale x 16 x i1> %a, <vscale x 16 x i1> %b) #0 {
 ; CHECK-LABEL: splice_nxv16i1:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    addvl sp, sp, #-2
 ; CHECK-NEXT:    mov z0.b, p0/z, #1 // =0x1
 ; CHECK-NEXT:    ptrue p0.b
 ; CHECK-NEXT:    mov z1.b, p1/z, #1 // =0x1
-; CHECK-NEXT:    mov x8, sp
-; CHECK-NEXT:    st1b { z0.b }, p0, [sp]
-; CHECK-NEXT:    st1b { z1.b }, p0, [x8, #1, mul vl]
-; CHECK-NEXT:    addvl x8, x8, #1
-; CHECK-NEXT:    mov x9, #-1
-; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x8, x9]
-; CHECK-NEXT:    and z0.b, z0.b, #0x1
-; CHECK-NEXT:    cmpne p0.b, p0/z, z0.b, #0
-; CHECK-NEXT:    addvl sp, sp, #2
-; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    lastb b0, p0, z0.b
+; CHECK-NEXT:    rev z1.b, z1.b
+; CHECK-NEXT:    insr z1.b, b0
+; CHECK-NEXT:    and z1.b, z1.b, #0x1
+; CHECK-NEXT:    cmpne p0.b, p0/z, z1.b, #0
 ; CHECK-NEXT:    ret
   %res = call <vscale x 16 x i1> @llvm.experimental.vector.splice.nxv16i1(<vscale x 16 x i1> %a, <vscale x 16 x i1> %b, i32 -1)
   ret <vscale x 16 x i1> %res
 }
 
 ; Verify promote type legalisation works as expected.
 define <vscale x 2 x i8> @splice_nxv2i8(<vscale x 2 x i8> %a, <vscale x 2 x i8> %b) #0 {
 ; CHECK-LABEL: splice_nxv2i8: