[RISCV] Add special case to selectImm for constants that can be created with (ADD (SLLI C, 32), C).

[craig.topper]

Summary

Where C is a simm32.

This costs an extra temporary register, but avoids a constant pool.

[asb]

I'd not really explored such materialisations before as adding another dimension to cost modeling (register usage) felt difficult to reason about...but intuitively I can't see spending an extra register being problematic in this specific case where a constant pool would have been used, and given how relatively register rich RISC-V is.

I am however seeing some compile failures for this e.g. 20000523-1.c from the GCC torture suite (e.g. compiled rv64imafdc for lp64 ABI at O0)

/home/asb/llvm-project/llvm/lib/MC/MCInst.cpp:58: bool llvm::MCOperand::isBareSymbolRef() const: Assertion `isExpr() && "isBareSymbolRef expects only expressions"' failed.
...

[reames]

General approach seems entirely reasonable to me.

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

212

Very minor: This code can produce a sequence with minimum length of 3, so you can avoid running it when the original sequence length is 3.

225

If you swap these operands, does that help add immediate formation when the two halves are very small?

I'm thinking of the case 0x00020002. This can be done as LI + SHL + ADDI. With only a single register.

This might not matter, and we might need a second special case for this, if so, feel free to punt to future patch.

Thinking about, it's probably a subcase of having a general 32 bit constant hi, + a 12 bit low half which can always be done in 4 instructions, and 3 depending on the exact details of the high half.

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

225

This should be ADD not ADDI.

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

209

This should be ADD

225

I assume you meant 0x0000000200000002? That should be handled by the base algorithm. https://godbolt.org/z/1rPsbh5a7

The base algorithm is basically subtract sign extended version of lo12 bits, count trailing zeros, shift right by that amount, recurse. The subtract becomes an ADDI and the shift right becomes a SLLI.

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

4403

This should be ADD

[craig.topper]

Use ADD instead of ADDI.

[reames]

LGTM

[fmayer]

I think this broke sanitizer tests: https://lab.llvm.org/buildbot/#/builders/238/builds/3867/steps/13/logs/stdio

[fmayer]

/b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43: runtime error: signed integer overflow: 9223372034904144827 - -1950630981 cannot be represented in type 'int64_t' (aka 'long')
    #0 0xaaaae4525208 in selectImm(llvm::SelectionDAG*, llvm::SDLoc const&, llvm::MVT, long, llvm::RISCVSubtarget const&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43
    #1 0xaaaae451fff8 in llvm::RISCVDAGToDAGISel::Select(llvm::SDNode*) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:863:23
    #2 0xaaaae6204148 in llvm::SelectionDAGISel::DoInstructionSelection() /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1178:7
    #3 0xaaaae6202418 in llvm::SelectionDAGISel::CodeGenAndEmitDAG() /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:949:5
    #4 0xaaaae61feb54 in llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function const&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1700:7
    #5 0xaaaae61fa77c in llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:482:3
    #6 0xaaaae5469750 in llvm::MachineFunctionPass::runOnFunction(llvm::Function&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/MachineFunctionPass.cpp:91:13
    #7 0xaaaae5b2acf0 in llvm::FPPassManager::runOnFunction(llvm::Function&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1435:27
    #8 0xaaaae5b34fd0 in llvm::FPPassManager::runOnModule(llvm::Module&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1481:16
    #9 0xaaaae5b2b7f0 in runOnModule /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1550:27
    #10 0xaaaae5b2b7f0 in llvm::legacy::PassManagerImpl::run(llvm::Module&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:535:44
    #11 0xaaaae31ae314 in compileModule(char**, llvm::LLVMContext&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/tools/llc/llc.cpp:756:8
    #12 0xaaaae31ac0b8 in main /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/tools/llc/llc.cpp:420:22
    #13 0xffffbb03777c  (/lib/aarch64-linux-gnu/libc.so.6+0x2777c) (BuildId: fe1fc959438108d405a33383d7c3f00762a5bb93)
    #14 0xffffbb037854 in __libc_start_main (/lib/aarch64-linux-gnu/libc.so.6+0x27854) (BuildId: fe1fc959438108d405a33383d7c3f00762a5bb93)
    #15 0xaaaae3180fec in _start (/b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm_build_ubsan/bin/llc+0x5e90fec)
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43 in

[craig.topper]

In D152236#4401604, @fmayer wrote:

/b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43: runtime error: signed integer overflow: 9223372034904144827 - -1950630981 cannot be represented in type 'int64_t' (aka 'long')
    #0 0xaaaae4525208 in selectImm(llvm::SelectionDAG*, llvm::SDLoc const&, llvm::MVT, long, llvm::RISCVSubtarget const&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43
    #1 0xaaaae451fff8 in llvm::RISCVDAGToDAGISel::Select(llvm::SDNode*) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:863:23
    #2 0xaaaae6204148 in llvm::SelectionDAGISel::DoInstructionSelection() /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1178:7
    #3 0xaaaae6202418 in llvm::SelectionDAGISel::CodeGenAndEmitDAG() /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:949:5
    #4 0xaaaae61feb54 in llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function const&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1700:7
    #5 0xaaaae61fa77c in llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:482:3
    #6 0xaaaae5469750 in llvm::MachineFunctionPass::runOnFunction(llvm::Function&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/CodeGen/MachineFunctionPass.cpp:91:13
    #7 0xaaaae5b2acf0 in llvm::FPPassManager::runOnFunction(llvm::Function&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1435:27
    #8 0xaaaae5b34fd0 in llvm::FPPassManager::runOnModule(llvm::Module&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1481:16
    #9 0xaaaae5b2b7f0 in runOnModule /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1550:27
    #10 0xaaaae5b2b7f0 in llvm::legacy::PassManagerImpl::run(llvm::Module&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:535:44
    #11 0xaaaae31ae314 in compileModule(char**, llvm::LLVMContext&) /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/tools/llc/llc.cpp:756:8
    #12 0xaaaae31ac0b8 in main /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/tools/llc/llc.cpp:420:22
    #13 0xffffbb03777c  (/lib/aarch64-linux-gnu/libc.so.6+0x2777c) (BuildId: fe1fc959438108d405a33383d7c3f00762a5bb93)
    #14 0xffffbb037854 in __libc_start_main (/lib/aarch64-linux-gnu/libc.so.6+0x27854) (BuildId: fe1fc959438108d405a33383d7c3f00762a5bb93)
    #15 0xaaaae3180fec in _start (/b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm_build_ubsan/bin/llc+0x5e90fec)
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior /b/sanitizer-aarch64-linux-bootstrap-ubsan/build/llvm-project/llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp:214:43 in

Thanks, I'll take a look.

The revert needs a cleanup of a test that was added after this patch was commit. I'll take care of that too.

[fmayer]

Seems like I made matters worse. Un-reverting this. Please fix the UBSAan error.

[craig.topper]

This was recommitted

[dtcxzyw]

Our CI reported that this patch caused code bloat (average ~0.1%) in llvm-test-suite compiled with -Os. We can turn this off when optimizing for size (-Os/-Oz).

[craig.topper]

In D152236#4405188, @dtcxzyw wrote:

Our CI reported that this patch caused code bloat (average ~0.1%) in llvm-test-suite compiled with -Os. We can turn this off when optimizing for size (-Os/-Oz).

https://reviews.llvm.org/D152602

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

static SDValue selectImmSeq(SelectionDAG *CurDAG, const SDLoc &DL, const MVT VT,
   return SrcReg;
 }
 
 static SDValue selectImm(SelectionDAG *CurDAG, const SDLoc &DL, const MVT VT,
                          int64_t Imm, const RISCVSubtarget &Subtarget) {
   RISCVMatInt::InstSeq Seq =
       RISCVMatInt::generateInstSeq(Imm, Subtarget.getFeatureBits());
 
+  // See if we can create this constant as (ADD (SLLI X, 32), X) where X is at
+  // worst an LUI+ADDIW. This will require an extra register, but avoids a
+  // constant pool.
+  if (Seq.size() > 3) {
+    int64_t LoVal = SignExtend64<32>(Imm);
+    int64_t HiVal = SignExtend64<32>((Imm - LoVal) >> 32);
+    if (LoVal == HiVal) {
+      RISCVMatInt::InstSeq SeqLo =
+          RISCVMatInt::generateInstSeq(LoVal, Subtarget.getFeatureBits());
+      if ((SeqLo.size() + 2) < Seq.size()) {
+        SDValue Lo = selectImmSeq(CurDAG, DL, VT, SeqLo);
+
+        SDValue SLLI = SDValue(
+            CurDAG->getMachineNode(RISCV::SLLI, DL, VT, Lo,
+                                   CurDAG->getTargetConstant(32, DL, VT)),
+            0);
+        return SDValue(CurDAG->getMachineNode(RISCV::ADD, DL, VT, Lo, SLLI),
+                       0);
+      }
+    }
+  }
+
+  // Otherwise, use the original sequence.
   return selectImmSeq(CurDAG, DL, VT, Seq);
 }
 
 static SDValue createTuple(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
                            unsigned NF, RISCVII::VLMUL LMUL) {
   static const unsigned M1TupleRegClassIDs[] = {
       RISCV::VRN2M1RegClassID, RISCV::VRN3M1RegClassID, RISCV::VRN4M1RegClassID,
       RISCV::VRN5M1RegClassID, RISCV::VRN6M1RegClassID, RISCV::VRN7M1RegClassID,

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

static SDValue lowerConstant(SDValue Op, SelectionDAG &DAG,
   if (!Subtarget.useConstantPoolForLargeInts())
     return Op;
 
   RISCVMatInt::InstSeq Seq =
       RISCVMatInt::generateInstSeq(Imm, Subtarget.getFeatureBits());
   if (Seq.size() <= Subtarget.getMaxBuildIntsCost())
     return Op;
 
+  // Special case. See if we can build the constant as (ADD (SLLI X, 32), X) do
+  // that if it will avoid a constant pool.
+  // It will require an extra temporary register though.
+  int64_t LoVal = SignExtend64<32>(Imm);
+  int64_t HiVal = SignExtend64<32>((Imm - LoVal) >> 32);
+  if (LoVal == HiVal) {
+    RISCVMatInt::InstSeq SeqLo =
+        RISCVMatInt::generateInstSeq(LoVal, Subtarget.getFeatureBits());
+    if ((SeqLo.size() + 2) <= Subtarget.getMaxBuildIntsCost())
+      return Op;
+  }
+
   // Expand to a constant pool using the default expansion code.
   return SDValue();
 }
 
 static SDValue LowerATOMIC_FENCE(SDValue Op, SelectionDAG &DAG,
                                  const RISCVSubtarget &Subtarget) {
   SDLoc dl(Op);
   AtomicOrdering FenceOrdering =

llvm/test/CodeGen/RISCV/bswap-bitreverse.ll

 ; RV64I-NEXT:    slli a4, a4, 24
 ; RV64I-NEXT:    srliw a3, a0, 24
 ; RV64I-NEXT:    slli a3, a3, 32
 ; RV64I-NEXT:    or a3, a4, a3
 ; RV64I-NEXT:    and a2, a0, a2
 ; RV64I-NEXT:    slli a2, a2, 40
 ; RV64I-NEXT:    slli a0, a0, 56
 ; RV64I-NEXT:    or a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_0)
-; RV64I-NEXT:    ld a2, %lo(.LCPI6_0)(a2)
 ; RV64I-NEXT:    or a0, a0, a3
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 4
+; RV64I-NEXT:    lui a2, 61681
+; RV64I-NEXT:    addiw a2, a2, -241
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI6_1)(a2)
 ; RV64I-NEXT:    slli a0, a0, 4
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    srli a1, a0, 2
+; RV64I-NEXT:    lui a2, 209715
+; RV64I-NEXT:    addiw a2, a2, 819
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI6_2)(a2)
 ; RV64I-NEXT:    slli a0, a0, 2
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
 ; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
 ; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_bitreverse_i64:
 ; RV32ZBB:       # %bb.0:

 ; RV32ZBB-NEXT:    and a0, a0, a5
 ; RV32ZBB-NEXT:    slli a0, a0, 1
 ; RV32ZBB-NEXT:    or a1, a1, a0
 ; RV32ZBB-NEXT:    mv a0, a2
 ; RV32ZBB-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: test_bitreverse_i64:
 ; RV64ZBB:       # %bb.0:
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_0)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_0)(a1)
 ; RV64ZBB-NEXT:    rev8 a0, a0
-; RV64ZBB-NEXT:    srli a2, a0, 4
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_1)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_1)(a1)
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 4
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 2
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI6_2)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI6_2)(a1)
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 2
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 1
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 1
-; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    or a0, a1, a0
 ; RV64ZBB-NEXT:    ret
 ;
 ; RV32ZBKB-LABEL: test_bitreverse_i64:
 ; RV32ZBKB:       # %bb.0:
 ; RV32ZBKB-NEXT:    rev8 a1, a1
 ; RV32ZBKB-NEXT:    brev8 a2, a1
 ; RV32ZBKB-NEXT:    rev8 a0, a0
 ; RV32ZBKB-NEXT:    brev8 a1, a0

 ; RV32I-NEXT:    and a2, a2, a5
 ; RV32I-NEXT:    and a1, a1, a5
 ; RV32I-NEXT:    slli a1, a1, 1
 ; RV32I-NEXT:    or a1, a2, a1
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bswap_bitreverse_i64:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    lui a1, %hi(.LCPI9_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI9_0)(a1)
-; RV64I-NEXT:    srli a2, a0, 4
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
-; RV64I-NEXT:    lui a1, %hi(.LCPI9_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI9_1)(a1)
+; RV64I-NEXT:    srli a1, a0, 4
+; RV64I-NEXT:    lui a2, 61681
+; RV64I-NEXT:    addiw a2, a2, -241
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 4
-; RV64I-NEXT:    or a0, a2, a0
-; RV64I-NEXT:    srli a2, a0, 2
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
-; RV64I-NEXT:    lui a1, %hi(.LCPI9_2)
-; RV64I-NEXT:    ld a1, %lo(.LCPI9_2)(a1)
+; RV64I-NEXT:    or a0, a1, a0
+; RV64I-NEXT:    srli a1, a0, 2
+; RV64I-NEXT:    lui a2, 209715
+; RV64I-NEXT:    addiw a2, a2, 819
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 2
-; RV64I-NEXT:    or a0, a2, a0
-; RV64I-NEXT:    srli a2, a0, 1
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    or a0, a1, a0
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
-; RV64I-NEXT:    or a0, a2, a0
+; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_bswap_bitreverse_i64:
 ; RV32ZBB:       # %bb.0:
 ; RV32ZBB-NEXT:    srli a2, a0, 4
 ; RV32ZBB-NEXT:    lui a3, 61681
 ; RV32ZBB-NEXT:    addi a3, a3, -241
 ; RV32ZBB-NEXT:    and a2, a2, a3

 ; RV32ZBB-NEXT:    and a2, a2, a5
 ; RV32ZBB-NEXT:    and a1, a1, a5
 ; RV32ZBB-NEXT:    slli a1, a1, 1
 ; RV32ZBB-NEXT:    or a1, a2, a1
 ; RV32ZBB-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: test_bswap_bitreverse_i64:
 ; RV64ZBB:       # %bb.0:
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_0)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_0)(a1)
-; RV64ZBB-NEXT:    srli a2, a0, 4
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_1)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_1)(a1)
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 4
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 2
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI9_2)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI9_2)(a1)
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 2
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 1
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 1
-; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    or a0, a1, a0
 ; RV64ZBB-NEXT:    ret
 ;
 ; RV32ZBKB-LABEL: test_bswap_bitreverse_i64:
 ; RV32ZBKB:       # %bb.0:
 ; RV32ZBKB-NEXT:    brev8 a0, a0
 ; RV32ZBKB-NEXT:    brev8 a1, a1
 ; RV32ZBKB-NEXT:    ret
 ;

 ; RV32I-NEXT:    and a2, a2, a5
 ; RV32I-NEXT:    and a1, a1, a5
 ; RV32I-NEXT:    slli a1, a1, 1
 ; RV32I-NEXT:    or a1, a2, a1
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_bitreverse_bswap_i64:
 ; RV64I:       # %bb.0:
-; RV64I-NEXT:    lui a1, %hi(.LCPI12_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI12_0)(a1)
-; RV64I-NEXT:    srli a2, a0, 4
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
-; RV64I-NEXT:    lui a1, %hi(.LCPI12_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI12_1)(a1)
+; RV64I-NEXT:    srli a1, a0, 4
+; RV64I-NEXT:    lui a2, 61681
+; RV64I-NEXT:    addiw a2, a2, -241
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 4
-; RV64I-NEXT:    or a0, a2, a0
-; RV64I-NEXT:    srli a2, a0, 2
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
-; RV64I-NEXT:    lui a1, %hi(.LCPI12_2)
-; RV64I-NEXT:    ld a1, %lo(.LCPI12_2)(a1)
+; RV64I-NEXT:    or a0, a1, a0
+; RV64I-NEXT:    srli a1, a0, 2
+; RV64I-NEXT:    lui a2, 209715
+; RV64I-NEXT:    addiw a2, a2, 819
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 2
-; RV64I-NEXT:    or a0, a2, a0
-; RV64I-NEXT:    srli a2, a0, 1
-; RV64I-NEXT:    and a2, a2, a1
-; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    or a0, a1, a0
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a2
 ; RV64I-NEXT:    slli a0, a0, 1
-; RV64I-NEXT:    or a0, a2, a0
+; RV64I-NEXT:    or a0, a1, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_bitreverse_bswap_i64:
 ; RV32ZBB:       # %bb.0:
 ; RV32ZBB-NEXT:    srli a2, a0, 4
 ; RV32ZBB-NEXT:    lui a3, 61681
 ; RV32ZBB-NEXT:    addi a3, a3, -241
 ; RV32ZBB-NEXT:    and a2, a2, a3

 ; RV32ZBB-NEXT:    and a2, a2, a5
 ; RV32ZBB-NEXT:    and a1, a1, a5
 ; RV32ZBB-NEXT:    slli a1, a1, 1
 ; RV32ZBB-NEXT:    or a1, a2, a1
 ; RV32ZBB-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: test_bitreverse_bswap_i64:
 ; RV64ZBB:       # %bb.0:
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_0)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_0)(a1)
-; RV64ZBB-NEXT:    srli a2, a0, 4
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_1)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_1)(a1)
+; RV64ZBB-NEXT:    srli a1, a0, 4
+; RV64ZBB-NEXT:    lui a2, 61681
+; RV64ZBB-NEXT:    addiw a2, a2, -241
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 4
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 2
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
-; RV64ZBB-NEXT:    lui a1, %hi(.LCPI12_2)
-; RV64ZBB-NEXT:    ld a1, %lo(.LCPI12_2)(a1)
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 2
+; RV64ZBB-NEXT:    lui a2, 209715
+; RV64ZBB-NEXT:    addiw a2, a2, 819
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 2
-; RV64ZBB-NEXT:    or a0, a2, a0
-; RV64ZBB-NEXT:    srli a2, a0, 1
-; RV64ZBB-NEXT:    and a2, a2, a1
-; RV64ZBB-NEXT:    and a0, a0, a1
+; RV64ZBB-NEXT:    or a0, a1, a0
+; RV64ZBB-NEXT:    srli a1, a0, 1
+; RV64ZBB-NEXT:    lui a2, 349525
+; RV64ZBB-NEXT:    addiw a2, a2, 1365
+; RV64ZBB-NEXT:    slli a3, a2, 32
+; RV64ZBB-NEXT:    add a2, a2, a3
+; RV64ZBB-NEXT:    and a1, a1, a2
+; RV64ZBB-NEXT:    and a0, a0, a2
 ; RV64ZBB-NEXT:    slli a0, a0, 1
-; RV64ZBB-NEXT:    or a0, a2, a0
+; RV64ZBB-NEXT:    or a0, a1, a0
 ; RV64ZBB-NEXT:    ret
 ;
 ; RV32ZBKB-LABEL: test_bitreverse_bswap_i64:
 ; RV32ZBKB:       # %bb.0:
 ; RV32ZBKB-NEXT:    brev8 a0, a0
 ; RV32ZBKB-NEXT:    brev8 a1, a1
 ; RV32ZBKB-NEXT:    ret
 ;

llvm/test/CodeGen/RISCV/ctlz-cttz-ctpop.ll

 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 8
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 16
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 32
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    not a0, a0
-; RV64I-NEXT:    lui a1, %hi(.LCPI11_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI11_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI11_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI11_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI11_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI11_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI11_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI11_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ; RV64I-NEXT:  .LBB11_2:
 ; RV64I-NEXT:    li a0, 64
 ; RV64I-NEXT:    ret

 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 8
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 16
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 32
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    not a0, a0
-; RV64M-NEXT:    lui a1, %hi(.LCPI11_0)
-; RV64M-NEXT:    ld a1, %lo(.LCPI11_0)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI11_1)
-; RV64M-NEXT:    ld a2, %lo(.LCPI11_1)(a2)
-; RV64M-NEXT:    srli a3, a0, 1
-; RV64M-NEXT:    and a1, a3, a1
+; RV64M-NEXT:    srli a1, a0, 1
+; RV64M-NEXT:    lui a2, 349525
+; RV64M-NEXT:    addiw a2, a2, 1365
+; RV64M-NEXT:    slli a3, a2, 32
+; RV64M-NEXT:    add a2, a2, a3
+; RV64M-NEXT:    and a1, a1, a2
 ; RV64M-NEXT:    sub a0, a0, a1
-; RV64M-NEXT:    and a1, a0, a2
+; RV64M-NEXT:    lui a1, 209715
+; RV64M-NEXT:    addiw a1, a1, 819
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a2, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 2
-; RV64M-NEXT:    and a0, a0, a2
-; RV64M-NEXT:    add a0, a1, a0
-; RV64M-NEXT:    lui a1, %hi(.LCPI11_2)
-; RV64M-NEXT:    ld a1, %lo(.LCPI11_2)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI11_3)
-; RV64M-NEXT:    ld a2, %lo(.LCPI11_3)(a2)
-; RV64M-NEXT:    srli a3, a0, 4
-; RV64M-NEXT:    add a0, a0, a3
 ; RV64M-NEXT:    and a0, a0, a1
-; RV64M-NEXT:    mul a0, a0, a2
+; RV64M-NEXT:    add a0, a2, a0
+; RV64M-NEXT:    srli a1, a0, 4
+; RV64M-NEXT:    add a0, a0, a1
+; RV64M-NEXT:    lui a1, 61681
+; RV64M-NEXT:    addiw a1, a1, -241
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a0, a0, a1
+; RV64M-NEXT:    lui a1, 4112
+; RV64M-NEXT:    addiw a1, a1, 257
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    mul a0, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 56
 ; RV64M-NEXT:    ret
 ; RV64M-NEXT:  .LBB11_2:
 ; RV64M-NEXT:    li a0, 64
 ; RV64M-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_ctlz_i64:
 ; RV32ZBB:       # %bb.0:

 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 8
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 16
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 32
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    not a0, a0
-; RV64I-NEXT:    lui a1, %hi(.LCPI15_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI15_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI15_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI15_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI15_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI15_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI15_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI15_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ;
 ; RV32M-LABEL: test_ctlz_i64_zero_undef:
 ; RV32M:       # %bb.0:

 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 8
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 16
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    srli a1, a0, 32
 ; RV64M-NEXT:    or a0, a0, a1
 ; RV64M-NEXT:    not a0, a0
-; RV64M-NEXT:    lui a1, %hi(.LCPI15_0)
-; RV64M-NEXT:    ld a1, %lo(.LCPI15_0)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI15_1)
-; RV64M-NEXT:    ld a2, %lo(.LCPI15_1)(a2)
-; RV64M-NEXT:    srli a3, a0, 1
-; RV64M-NEXT:    and a1, a3, a1
+; RV64M-NEXT:    srli a1, a0, 1
+; RV64M-NEXT:    lui a2, 349525
+; RV64M-NEXT:    addiw a2, a2, 1365
+; RV64M-NEXT:    slli a3, a2, 32
+; RV64M-NEXT:    add a2, a2, a3
+; RV64M-NEXT:    and a1, a1, a2
 ; RV64M-NEXT:    sub a0, a0, a1
-; RV64M-NEXT:    and a1, a0, a2
+; RV64M-NEXT:    lui a1, 209715
+; RV64M-NEXT:    addiw a1, a1, 819
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a2, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 2
-; RV64M-NEXT:    and a0, a0, a2
-; RV64M-NEXT:    add a0, a1, a0
-; RV64M-NEXT:    lui a1, %hi(.LCPI15_2)
-; RV64M-NEXT:    ld a1, %lo(.LCPI15_2)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI15_3)
-; RV64M-NEXT:    ld a2, %lo(.LCPI15_3)(a2)
-; RV64M-NEXT:    srli a3, a0, 4
-; RV64M-NEXT:    add a0, a0, a3
 ; RV64M-NEXT:    and a0, a0, a1
-; RV64M-NEXT:    mul a0, a0, a2
+; RV64M-NEXT:    add a0, a2, a0
+; RV64M-NEXT:    srli a1, a0, 4
+; RV64M-NEXT:    add a0, a0, a1
+; RV64M-NEXT:    lui a1, 61681
+; RV64M-NEXT:    addiw a1, a1, -241
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a0, a0, a1
+; RV64M-NEXT:    lui a1, 4112
+; RV64M-NEXT:    addiw a1, a1, 257
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    mul a0, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 56
 ; RV64M-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_ctlz_i64_zero_undef:
 ; RV32ZBB:       # %bb.0:
 ; RV32ZBB-NEXT:    bnez a1, .LBB15_2
 ; RV32ZBB-NEXT:  # %bb.1:
 ; RV32ZBB-NEXT:    clz a0, a0

 ; RV32I-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32I-NEXT:    addi sp, sp, 32
 ; RV32I-NEXT:    ret
 ;
 ; RV64I-LABEL: test_ctpop_i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    addi sp, sp, -16
 ; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI19_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI19_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI19_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI19_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ;
 ; RV32M-LABEL: test_ctpop_i64:
 ; RV32M:       # %bb.0:

 ; RV32M-NEXT:    mul a0, a0, a5
 ; RV32M-NEXT:    srli a0, a0, 24
 ; RV32M-NEXT:    add a0, a0, a1
 ; RV32M-NEXT:    li a1, 0
 ; RV32M-NEXT:    ret
 ;
 ; RV64M-LABEL: test_ctpop_i64:
 ; RV64M:       # %bb.0:
-; RV64M-NEXT:    lui a1, %hi(.LCPI19_0)
-; RV64M-NEXT:    ld a1, %lo(.LCPI19_0)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI19_1)
-; RV64M-NEXT:    ld a2, %lo(.LCPI19_1)(a2)
-; RV64M-NEXT:    srli a3, a0, 1
-; RV64M-NEXT:    and a1, a3, a1
+; RV64M-NEXT:    srli a1, a0, 1
+; RV64M-NEXT:    lui a2, 349525
+; RV64M-NEXT:    addiw a2, a2, 1365
+; RV64M-NEXT:    slli a3, a2, 32
+; RV64M-NEXT:    add a2, a2, a3
+; RV64M-NEXT:    and a1, a1, a2
 ; RV64M-NEXT:    sub a0, a0, a1
-; RV64M-NEXT:    and a1, a0, a2
+; RV64M-NEXT:    lui a1, 209715
+; RV64M-NEXT:    addiw a1, a1, 819
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a2, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 2
-; RV64M-NEXT:    and a0, a0, a2
-; RV64M-NEXT:    add a0, a1, a0
-; RV64M-NEXT:    lui a1, %hi(.LCPI19_2)
-; RV64M-NEXT:    ld a1, %lo(.LCPI19_2)(a1)
-; RV64M-NEXT:    lui a2, %hi(.LCPI19_3)
-; RV64M-NEXT:    ld a2, %lo(.LCPI19_3)(a2)
-; RV64M-NEXT:    srli a3, a0, 4
-; RV64M-NEXT:    add a0, a0, a3
 ; RV64M-NEXT:    and a0, a0, a1
-; RV64M-NEXT:    mul a0, a0, a2
+; RV64M-NEXT:    add a0, a2, a0
+; RV64M-NEXT:    srli a1, a0, 4
+; RV64M-NEXT:    add a0, a0, a1
+; RV64M-NEXT:    lui a1, 61681
+; RV64M-NEXT:    addiw a1, a1, -241
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    and a0, a0, a1
+; RV64M-NEXT:    lui a1, 4112
+; RV64M-NEXT:    addiw a1, a1, 257
+; RV64M-NEXT:    slli a2, a1, 32
+; RV64M-NEXT:    add a1, a1, a2
+; RV64M-NEXT:    mul a0, a0, a1
 ; RV64M-NEXT:    srli a0, a0, 56
 ; RV64M-NEXT:    ret
 ;
 ; RV32ZBB-LABEL: test_ctpop_i64:
 ; RV32ZBB:       # %bb.0:
 ; RV32ZBB-NEXT:    cpop a1, a1
 ; RV32ZBB-NEXT:    cpop a0, a0
 ; RV32ZBB-NEXT:    add a0, a0, a1

 ; RV32XTHEADBB-NEXT:    lw s5, 4(sp) # 4-byte Folded Reload
 ; RV32XTHEADBB-NEXT:    addi sp, sp, 32
 ; RV32XTHEADBB-NEXT:    ret
 ;
 ; RV64XTHEADBB-LABEL: test_ctpop_i64:
 ; RV64XTHEADBB:       # %bb.0:
 ; RV64XTHEADBB-NEXT:    addi sp, sp, -16
 ; RV64XTHEADBB-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64XTHEADBB-NEXT:    lui a1, %hi(.LCPI19_0)
-; RV64XTHEADBB-NEXT:    ld a1, %lo(.LCPI19_0)(a1)
-; RV64XTHEADBB-NEXT:    lui a2, %hi(.LCPI19_1)
-; RV64XTHEADBB-NEXT:    ld a2, %lo(.LCPI19_1)(a2)
-; RV64XTHEADBB-NEXT:    srli a3, a0, 1
-; RV64XTHEADBB-NEXT:    and a1, a3, a1
+; RV64XTHEADBB-NEXT:    srli a1, a0, 1
+; RV64XTHEADBB-NEXT:    lui a2, 349525
+; RV64XTHEADBB-NEXT:    addiw a2, a2, 1365
+; RV64XTHEADBB-NEXT:    slli a3, a2, 32
+; RV64XTHEADBB-NEXT:    add a2, a2, a3
+; RV64XTHEADBB-NEXT:    and a1, a1, a2
 ; RV64XTHEADBB-NEXT:    sub a0, a0, a1
-; RV64XTHEADBB-NEXT:    and a1, a0, a2
+; RV64XTHEADBB-NEXT:    lui a1, 209715
+; RV64XTHEADBB-NEXT:    addiw a1, a1, 819
+; RV64XTHEADBB-NEXT:    slli a2, a1, 32
+; RV64XTHEADBB-NEXT:    add a1, a1, a2
+; RV64XTHEADBB-NEXT:    and a2, a0, a1
 ; RV64XTHEADBB-NEXT:    srli a0, a0, 2
-; RV64XTHEADBB-NEXT:    and a0, a0, a2
-; RV64XTHEADBB-NEXT:    lui a2, %hi(.LCPI19_2)
-; RV64XTHEADBB-NEXT:    ld a2, %lo(.LCPI19_2)(a2)
-; RV64XTHEADBB-NEXT:    add a0, a1, a0
+; RV64XTHEADBB-NEXT:    and a0, a0, a1
+; RV64XTHEADBB-NEXT:    add a0, a2, a0
 ; RV64XTHEADBB-NEXT:    srli a1, a0, 4
 ; RV64XTHEADBB-NEXT:    add a0, a0, a1
-; RV64XTHEADBB-NEXT:    and a0, a0, a2
-; RV64XTHEADBB-NEXT:    lui a1, %hi(.LCPI19_3)
-; RV64XTHEADBB-NEXT:    ld a1, %lo(.LCPI19_3)(a1)
+; RV64XTHEADBB-NEXT:    lui a1, 61681
+; RV64XTHEADBB-NEXT:    addiw a1, a1, -241
+; RV64XTHEADBB-NEXT:    slli a2, a1, 32
+; RV64XTHEADBB-NEXT:    add a1, a1, a2
+; RV64XTHEADBB-NEXT:    and a0, a0, a1
+; RV64XTHEADBB-NEXT:    lui a1, 4112
+; RV64XTHEADBB-NEXT:    addiw a1, a1, 257
+; RV64XTHEADBB-NEXT:    slli a2, a1, 32
+; RV64XTHEADBB-NEXT:    add a1, a1, a2
 ; RV64XTHEADBB-NEXT:    call __muldi3@plt
 ; RV64XTHEADBB-NEXT:    srli a0, a0, 56
 ; RV64XTHEADBB-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64XTHEADBB-NEXT:    addi sp, sp, 16
 ; RV64XTHEADBB-NEXT:    ret
   %1 = call i64 @llvm.ctpop.i64(i64 %a)
   ret i64 %1
 }

llvm/test/CodeGen/RISCV/div-by-constant.ll

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a3, a1
 ; RV32-NEXT:    mul a0, a5, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: udiv64_constant_no_add:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI2_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI2_0)(a1)
+; RV64-NEXT:    lui a1, 838861
+; RV64-NEXT:    addiw a1, a1, -819
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    mulhu a0, a0, a1
 ; RV64-NEXT:    srli a0, a0, 2
 ; RV64-NEXT:    ret
   %1 = udiv i64 %a, 5
   ret i64 %1
 }
 
 define i64 @udiv64_constant_add(i64 %a) nounwind {

 ; RV32-NEXT:    li a3, 0
 ; RV32-NEXT:    call __divdi3@plt
 ; RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: sdiv64_constant_add_srai:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI14_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI14_0)(a1)
+; RV64-NEXT:    lui a1, 559241
+; RV64-NEXT:    addiw a1, a1, -1911
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    mulh a1, a0, a1
 ; RV64-NEXT:    add a0, a1, a0
 ; RV64-NEXT:    srli a1, a0, 63
 ; RV64-NEXT:    srai a0, a0, 3
 ; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    ret
   %1 = sdiv i64 %a, 15
   ret i64 %1
 }
 
 define i64 @sdiv64_constant_sub_srai(i64 %a) nounwind {
 ; RV32-LABEL: sdiv64_constant_sub_srai:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
 ; RV32-NEXT:    li a2, -3
 ; RV32-NEXT:    li a3, -1
 ; RV32-NEXT:    call __divdi3@plt
 ; RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: sdiv64_constant_sub_srai:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI15_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI15_0)(a1)
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    mulh a1, a0, a1
 ; RV64-NEXT:    sub a1, a1, a0
 ; RV64-NEXT:    srli a0, a1, 63
 ; RV64-NEXT:    srai a1, a1, 1
 ; RV64-NEXT:    add a0, a1, a0
 ; RV64-NEXT:    ret
   %1 = sdiv i64 %a, -3
   ret i64 %1

llvm/test/CodeGen/RISCV/div.ll

 ;
 ; RV64I-LABEL: udiv64_constant:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a1, 5
 ; RV64I-NEXT:    tail __udivdi3@plt
 ;
 ; RV64IM-LABEL: udiv64_constant:
 ; RV64IM:       # %bb.0:
-; RV64IM-NEXT:    lui a1, %hi(.LCPI5_0)
-; RV64IM-NEXT:    ld a1, %lo(.LCPI5_0)(a1)
+; RV64IM-NEXT:    lui a1, 838861
+; RV64IM-NEXT:    addiw a1, a1, -819
+; RV64IM-NEXT:    slli a2, a1, 32
+; RV64IM-NEXT:    add a1, a1, a2
 ; RV64IM-NEXT:    mulhu a0, a0, a1
 ; RV64IM-NEXT:    srli a0, a0, 2
 ; RV64IM-NEXT:    ret
   %1 = udiv i64 %a, 5
   ret i64 %1
 }
 
 define i64 @udiv64_constant_lhs(i64 %a) nounwind {

llvm/test/CodeGen/RISCV/pr56457.ll

 ; CHECK-NEXT:    or a0, a0, a1
 ; CHECK-NEXT:    slli a1, a0, 49
 ; CHECK-NEXT:    srli a1, a1, 53
 ; CHECK-NEXT:    or a0, a0, a1
 ; CHECK-NEXT:    slli a1, a0, 49
 ; CHECK-NEXT:    srli a1, a1, 57
 ; CHECK-NEXT:    or a0, a0, a1
 ; CHECK-NEXT:    not a0, a0
+; CHECK-NEXT:    srli a1, a0, 1
+; CHECK-NEXT:    lui a2, 1
+; CHECK-NEXT:    addiw a2, a2, 1365
+; CHECK-NEXT:    and a1, a1, a2
 ; CHECK-NEXT:    slli a0, a0, 49
 ; CHECK-NEXT:    srli a0, a0, 49
-; CHECK-NEXT:    lui a1, %hi(.LCPI0_0)
-; CHECK-NEXT:    ld a1, %lo(.LCPI0_0)(a1)
-; CHECK-NEXT:    lui a2, %hi(.LCPI0_1)
-; CHECK-NEXT:    ld a2, %lo(.LCPI0_1)(a2)
-; CHECK-NEXT:    srli a3, a0, 1
-; CHECK-NEXT:    and a1, a3, a1
 ; CHECK-NEXT:    sub a0, a0, a1
-; CHECK-NEXT:    and a1, a0, a2
+; CHECK-NEXT:    lui a1, 209715
+; CHECK-NEXT:    addiw a1, a1, 819
+; CHECK-NEXT:    slli a2, a1, 32
+; CHECK-NEXT:    add a1, a1, a2
+; CHECK-NEXT:    and a2, a0, a1
 ; CHECK-NEXT:    srli a0, a0, 2
-; CHECK-NEXT:    and a0, a0, a2
-; CHECK-NEXT:    add a0, a1, a0
-; CHECK-NEXT:    lui a1, %hi(.LCPI0_2)
-; CHECK-NEXT:    ld a1, %lo(.LCPI0_2)(a1)
-; CHECK-NEXT:    lui a2, %hi(.LCPI0_3)
-; CHECK-NEXT:    ld a2, %lo(.LCPI0_3)(a2)
-; CHECK-NEXT:    srli a3, a0, 4
-; CHECK-NEXT:    add a0, a0, a3
 ; CHECK-NEXT:    and a0, a0, a1
-; CHECK-NEXT:    mul a0, a0, a2
+; CHECK-NEXT:    add a0, a2, a0
+; CHECK-NEXT:    srli a1, a0, 4
+; CHECK-NEXT:    add a0, a0, a1
+; CHECK-NEXT:    lui a1, 61681
+; CHECK-NEXT:    addiw a1, a1, -241
+; CHECK-NEXT:    slli a2, a1, 32
+; CHECK-NEXT:    add a1, a1, a2
+; CHECK-NEXT:    and a0, a0, a1
+; CHECK-NEXT:    lui a1, 4112
+; CHECK-NEXT:    addiw a1, a1, 257
+; CHECK-NEXT:    slli a2, a1, 32
+; CHECK-NEXT:    add a1, a1, a2
+; CHECK-NEXT:    mul a0, a0, a1
 ; CHECK-NEXT:    srli a0, a0, 56
 ; CHECK-NEXT:    ret
 ; CHECK-NEXT:  .LBB0_2:
 ; CHECK-NEXT:    li a0, 15
 ; CHECK-NEXT:    ret
   %a = call i15 @llvm.ctlz.i15(i15 %x, i1 false)
   ret i15 %a
 }

llvm/test/CodeGen/RISCV/rv64xtheadbb.ll

 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 8
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 16
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 32
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    not a0, a0
-; RV64I-NEXT:    lui a1, %hi(.LCPI5_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI5_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI5_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI5_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI5_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI5_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI5_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI5_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ; RV64I-NEXT:  .LBB5_2:
 ; RV64I-NEXT:    li a0, 64
 ; RV64I-NEXT:    ret

llvm/test/CodeGen/RISCV/rv64zbb.ll

 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 8
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 16
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    srli a1, a0, 32
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    not a0, a0
-; RV64I-NEXT:    lui a1, %hi(.LCPI5_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI5_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI5_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI5_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI5_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI5_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI5_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI5_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ; RV64I-NEXT:  .LBB5_2:
 ; RV64I-NEXT:    li a0, 64
 ; RV64I-NEXT:    ret

 
 declare i64 @llvm.ctpop.i64(i64)
 
 define i64 @ctpop_i64(i64 %a) nounwind {
 ; RV64I-LABEL: ctpop_i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    addi sp, sp, -16
 ; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lui a1, %hi(.LCPI13_0)
-; RV64I-NEXT:    ld a1, %lo(.LCPI13_0)(a1)
-; RV64I-NEXT:    lui a2, %hi(.LCPI13_1)
-; RV64I-NEXT:    ld a2, %lo(.LCPI13_1)(a2)
-; RV64I-NEXT:    srli a3, a0, 1
-; RV64I-NEXT:    and a1, a3, a1
+; RV64I-NEXT:    srli a1, a0, 1
+; RV64I-NEXT:    lui a2, 349525
+; RV64I-NEXT:    addiw a2, a2, 1365
+; RV64I-NEXT:    slli a3, a2, 32
+; RV64I-NEXT:    add a2, a2, a3
+; RV64I-NEXT:    and a1, a1, a2
 ; RV64I-NEXT:    sub a0, a0, a1
-; RV64I-NEXT:    and a1, a0, a2
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw a1, a1, 819
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a2, a0, a1
 ; RV64I-NEXT:    srli a0, a0, 2
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a2, %hi(.LCPI13_2)
-; RV64I-NEXT:    ld a2, %lo(.LCPI13_2)(a2)
-; RV64I-NEXT:    add a0, a1, a0
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    add a0, a2, a0
 ; RV64I-NEXT:    srli a1, a0, 4
 ; RV64I-NEXT:    add a0, a0, a1
-; RV64I-NEXT:    and a0, a0, a2
-; RV64I-NEXT:    lui a1, %hi(.LCPI13_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI13_3)(a1)
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw a1, a1, -241
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
+; RV64I-NEXT:    and a0, a0, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw a1, a1, 257
+; RV64I-NEXT:    slli a2, a1, 32
+; RV64I-NEXT:    add a1, a1, a2
 ; RV64I-NEXT:    call __muldi3@plt
 ; RV64I-NEXT:    srli a0, a0, 56
 ; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
 ; RV64I-NEXT:    addi sp, sp, 16
 ; RV64I-NEXT:    ret
 ;
 ; RV64ZBB-LABEL: ctpop_i64:
 ; RV64ZBB:       # %bb.0:

llvm/test/CodeGen/RISCV/rvv/bitreverse-sdnode.ll

 ; RV64-NEXT:    vsll.vi v10, v10, 24
 ; RV64-NEXT:    vand.vx v11, v8, a4
 ; RV64-NEXT:    vsll.vi v11, v11, 8
 ; RV64-NEXT:    vor.vv v10, v10, v11
 ; RV64-NEXT:    vsll.vx v11, v8, a0
 ; RV64-NEXT:    vand.vx v8, v8, a2
 ; RV64-NEXT:    vsll.vx v8, v8, a1
 ; RV64-NEXT:    vor.vv v8, v11, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI18_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI18_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI18_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI18_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI18_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI18_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x i64> @llvm.bitreverse.nxv1i64(<vscale x 1 x i64> %va)
   ret <vscale x 1 x i64> %a
 }

 ; RV64-NEXT:    vsll.vi v12, v12, 24
 ; RV64-NEXT:    vand.vx v14, v8, a4
 ; RV64-NEXT:    vsll.vi v14, v14, 8
 ; RV64-NEXT:    vor.vv v12, v12, v14
 ; RV64-NEXT:    vsll.vx v14, v8, a0
 ; RV64-NEXT:    vand.vx v8, v8, a2
 ; RV64-NEXT:    vsll.vx v8, v8, a1
 ; RV64-NEXT:    vor.vv v8, v14, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI19_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI19_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI19_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI19_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI19_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI19_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 2 x i64> @llvm.bitreverse.nxv2i64(<vscale x 2 x i64> %va)
   ret <vscale x 2 x i64> %a
 }

 ; RV64-NEXT:    vsll.vi v16, v16, 24
 ; RV64-NEXT:    vand.vx v20, v8, a4
 ; RV64-NEXT:    vsll.vi v20, v20, 8
 ; RV64-NEXT:    vor.vv v16, v16, v20
 ; RV64-NEXT:    vsll.vx v20, v8, a0
 ; RV64-NEXT:    vand.vx v8, v8, a2
 ; RV64-NEXT:    vsll.vx v8, v8, a1
 ; RV64-NEXT:    vor.vv v8, v20, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI20_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI20_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI20_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI20_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI20_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI20_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x i64> @llvm.bitreverse.nxv4i64(<vscale x 4 x i64> %va)
   ret <vscale x 4 x i64> %a
 }

 ; RV64-NEXT:    vsll.vi v24, v24, 24
 ; RV64-NEXT:    vand.vx v0, v8, a4
 ; RV64-NEXT:    vsll.vi v0, v0, 8
 ; RV64-NEXT:    vor.vv v24, v24, v0
 ; RV64-NEXT:    vsll.vx v0, v8, a0
 ; RV64-NEXT:    vand.vx v8, v8, a2
 ; RV64-NEXT:    vsll.vx v8, v8, a1
 ; RV64-NEXT:    vor.vv v8, v0, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI21_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI21_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI21_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI21_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI21_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI21_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 8 x i64> @llvm.bitreverse.nxv8i64(<vscale x 8 x i64> %va)
   ret <vscale x 8 x i64> %a
 }
 declare <vscale x 8 x i64> @llvm.bitreverse.nxv8i64(<vscale x 8 x i64>)

llvm/test/CodeGen/RISCV/rvv/bitreverse-vp.ll

 ; RV64-NEXT:    vsrl.vx v11, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v11, v11, a3, v0.t
 ; RV64-NEXT:    vor.vv v10, v11, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v11, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v11, v11, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v11, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.bitreverse.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v11, v8, a4
 ; RV64-NEXT:    vand.vx v11, v11, a3
 ; RV64-NEXT:    vor.vv v10, v11, v10
 ; RV64-NEXT:    vsrl.vi v11, v8, 24
 ; RV64-NEXT:    vand.vx v11, v11, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v11
-; RV64-NEXT:    lui a0, %hi(.LCPI37_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI37_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI37_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.bitreverse.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v14, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v14, v14, a3, v0.t
 ; RV64-NEXT:    vor.vv v12, v14, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v14, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v14, v14, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v14, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.bitreverse.nxv2i64(<vscale x 2 x i64> %va, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v14, v8, a4
 ; RV64-NEXT:    vand.vx v14, v14, a3
 ; RV64-NEXT:    vor.vv v12, v14, v12
 ; RV64-NEXT:    vsrl.vi v14, v8, 24
 ; RV64-NEXT:    vand.vx v14, v14, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v14
-; RV64-NEXT:    lui a0, %hi(.LCPI39_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI39_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI39_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.bitreverse.nxv2i64(<vscale x 2 x i64> %va, <vscale x 2 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v20, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v20, v20, a3, v0.t
 ; RV64-NEXT:    vor.vv v16, v20, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v20, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v20, v20, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v20, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.bitreverse.nxv4i64(<vscale x 4 x i64> %va, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v20, v8, a4
 ; RV64-NEXT:    vand.vx v20, v20, a3
 ; RV64-NEXT:    vor.vv v16, v20, v16
 ; RV64-NEXT:    vsrl.vi v20, v8, 24
 ; RV64-NEXT:    vand.vx v20, v20, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v20
-; RV64-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI41_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.bitreverse.nxv4i64(<vscale x 4 x i64> %va, <vscale x 4 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vor.vv v24, v16, v24, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24, v0.t
-; RV64-NEXT:    addi a1, sp, 16
-; RV64-NEXT:    vl8r.v v16, (a1) # Unknown-size Folded Reload
-; RV64-NEXT:    vor.vv v16, v16, v8, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    addi a0, sp, 16
+; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_1)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 4, v0.t
-; RV64-NEXT:    vor.vv v16, v8, v16, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 2, v0.t
-; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16

 ; RV64-NEXT:    vsrl.vx v0, v8, a4
 ; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
 ; RV64-NEXT:    vsrl.vi v0, v8, 24
 ; RV64-NEXT:    vand.vx v0, v0, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v0
-; RV64-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI43_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.bitreverse.nxv7i64(<vscale x 7 x i64> %va, <vscale x 7 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vor.vv v24, v16, v24, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24, v0.t
-; RV64-NEXT:    addi a1, sp, 16
-; RV64-NEXT:    vl8r.v v16, (a1) # Unknown-size Folded Reload
-; RV64-NEXT:    vor.vv v16, v16, v8, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    addi a0, sp, 16
+; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_1)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 4, v0.t
-; RV64-NEXT:    vor.vv v16, v8, v16, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_2)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 2, v0.t
-; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16

 ; RV64-NEXT:    vsrl.vx v0, v8, a4
 ; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
 ; RV64-NEXT:    vsrl.vi v0, v8, 24
 ; RV64-NEXT:    vand.vx v0, v0, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v0
-; RV64-NEXT:    lui a0, %hi(.LCPI45_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI45_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI45_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.bitreverse.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 %evl)

llvm/test/CodeGen/RISCV/rvv/ctlz-sdnode.ll

 ; RV64I-NEXT:    vsrl.vi v9, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    vsrl.vi v9, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v9, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI18_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI18_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI18_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI18_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v9, v9, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v9
-; RV64I-NEXT:    vand.vx v9, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v9, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v9, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI18_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI18_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI18_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI18_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v9
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_nxv1i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v10, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    vsrl.vi v10, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v10, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI19_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI19_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v10, v10, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v10
-; RV64I-NEXT:    vand.vx v10, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v10, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v10, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI19_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI19_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v10
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_nxv2i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v12, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    vsrl.vi v12, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v12, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI20_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI20_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI20_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI20_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v12, v12, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v12
-; RV64I-NEXT:    vand.vx v12, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v12, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v12, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI20_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI20_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI20_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI20_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v12
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_nxv4i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v16, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    vsrl.vi v16, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v16, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI21_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI21_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI21_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI21_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v16, v16, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v16
-; RV64I-NEXT:    vand.vx v16, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v16, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v16, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI21_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI21_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI21_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI21_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v16
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_nxv8i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v9, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    vsrl.vi v9, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v9, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v9
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI40_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI40_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v9, v9, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v9
-; RV64I-NEXT:    vand.vx v9, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v9, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v9, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI40_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI40_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v9
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv1i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v10, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    vsrl.vi v10, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v10, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v10
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI41_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI41_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v10, v10, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v10
-; RV64I-NEXT:    vand.vx v10, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v10, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v10, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI41_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI41_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v10
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv2i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v12, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    vsrl.vi v12, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v12, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v12
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI42_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI42_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v12, v12, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v12
-; RV64I-NEXT:    vand.vx v12, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v12, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v12, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI42_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI42_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v12
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv4i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

 ; RV64I-NEXT:    vsrl.vi v16, v8, 8
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    vsrl.vi v16, v8, 16
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    li a0, 32
 ; RV64I-NEXT:    vsrl.vx v16, v8, a0
 ; RV64I-NEXT:    vor.vv v8, v8, v16
 ; RV64I-NEXT:    vnot.v v8, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI43_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI43_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v16, v16, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v16
-; RV64I-NEXT:    vand.vx v16, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v16, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v16, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI43_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI43_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v16
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: ctlz_zero_undef_nxv8i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
 ; CHECK-F-NEXT:    vmset.m v0

llvm/test/CodeGen/RISCV/rvv/ctlz-vp.ll

 ; RV64-NEXT:    vsrl.vi v9, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.ctlz.nxv1i64(<vscale x 1 x i64> %va, i1 false, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }
 
 define <vscale x 1 x i64> @vp_ctlz_nxv1i64_unmasked(<vscale x 1 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v9, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI37_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI37_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.ctlz.nxv1i64(<vscale x 1 x i64> %va, i1 false, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v10, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.ctlz.nxv2i64(<vscale x 2 x i64> %va, i1 false, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }
 
 define <vscale x 2 x i64> @vp_ctlz_nxv2i64_unmasked(<vscale x 2 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v10, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vsrl.vi v10, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI39_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI39_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.ctlz.nxv2i64(<vscale x 2 x i64> %va, i1 false, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v12, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.ctlz.nxv4i64(<vscale x 4 x i64> %va, i1 false, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }
 
 define <vscale x 4 x i64> @vp_ctlz_nxv4i64_unmasked(<vscale x 4 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v12, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vsrl.vi v12, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.ctlz.nxv4i64(<vscale x 4 x i64> %va, i1 false, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 7 x i64> @llvm.vp.ctlz.nxv7i64(<vscale x 7 x i64> %va, i1 false, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }
 
 define <vscale x 7 x i64> @vp_ctlz_nxv7i64_unmasked(<vscale x 7 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.ctlz.nxv7i64(<vscale x 7 x i64> %va, i1 false, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 8 x i64> @llvm.vp.ctlz.nxv8i64(<vscale x 8 x i64> %va, i1 false, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }
 
 define <vscale x 8 x i64> @vp_ctlz_nxv8i64_unmasked(<vscale x 8 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI45_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI45_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.ctlz.nxv8i64(<vscale x 8 x i64> %va, i1 false, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a2, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v16, v8, v0.t
-; RV64-NEXT:    lui a3, %hi(.LCPI46_0)
-; RV64-NEXT:    ld a4, %lo(.LCPI46_0)(a3)
-; RV64-NEXT:    lui a3, %hi(.LCPI46_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI46_1)(a3)
 ; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
-; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vadd.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a5, %hi(.LCPI46_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI46_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI46_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI46_3)(a6)
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsub.vv v16, v16, v8, v0.t
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
+; RV64-NEXT:    vand.vx v8, v16, a4, v0.t
+; RV64-NEXT:    vsrl.vi v16, v16, 2, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi t0, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (t0) # Unknown-size Folded Spill
 ; RV64-NEXT:    bltu a0, a1, .LBB46_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    vsrl.vi v24, v16, 8
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    vsrl.vi v24, v16, 16
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    li a2, 32
 ; RV64-NEXT:    vsrl.vx v24, v16, a2
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    vnot.v v16, v16
-; RV64-NEXT:    lui a3, %hi(.LCPI47_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI47_0)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI47_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI47_1)(a4)
 ; RV64-NEXT:    vsrl.vi v24, v16, 1
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v24, a3
 ; RV64-NEXT:    vsub.vv v16, v16, v24
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v24, v16, a4
 ; RV64-NEXT:    vsrl.vi v16, v16, 2
 ; RV64-NEXT:    vand.vx v16, v16, a4
 ; RV64-NEXT:    vadd.vv v16, v24, v16
-; RV64-NEXT:    lui a5, %hi(.LCPI47_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI47_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI47_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI47_3)(a6)
 ; RV64-NEXT:    vsrl.vi v24, v16, 4
 ; RV64-NEXT:    vadd.vv v16, v16, v24
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v16, v16, a5
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v16, v16, a6
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v16, v16, a7
 ; RV64-NEXT:    bltu a0, a1, .LBB47_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB47_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

 ; RV64-NEXT:    vsrl.vi v9, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI84_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI84_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI84_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI84_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI84_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI84_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI84_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI84_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.ctlz.nxv1i64(<vscale x 1 x i64> %va, i1 true, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }
 
 define <vscale x 1 x i64> @vp_ctlz_zero_undef_nxv1i64_unmasked(<vscale x 1 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v9, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI85_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI85_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI85_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI85_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI85_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI85_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI85_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI85_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.ctlz.nxv1i64(<vscale x 1 x i64> %va, i1 true, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v10, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI86_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI86_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI86_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI86_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI86_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI86_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI86_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI86_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.ctlz.nxv2i64(<vscale x 2 x i64> %va, i1 true, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }
 
 define <vscale x 2 x i64> @vp_ctlz_zero_undef_nxv2i64_unmasked(<vscale x 2 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v10, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vsrl.vi v10, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI87_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI87_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI87_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI87_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI87_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI87_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI87_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI87_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.ctlz.nxv2i64(<vscale x 2 x i64> %va, i1 true, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v12, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI88_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI88_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI88_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI88_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI88_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI88_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI88_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI88_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.ctlz.nxv4i64(<vscale x 4 x i64> %va, i1 true, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }
 
 define <vscale x 4 x i64> @vp_ctlz_zero_undef_nxv4i64_unmasked(<vscale x 4 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v12, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vsrl.vi v12, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI89_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI89_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI89_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI89_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI89_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI89_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI89_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI89_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.ctlz.nxv4i64(<vscale x 4 x i64> %va, i1 true, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI90_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI90_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI90_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI90_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI90_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI90_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI90_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI90_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 7 x i64> @llvm.vp.ctlz.nxv7i64(<vscale x 7 x i64> %va, i1 true, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }
 
 define <vscale x 7 x i64> @vp_ctlz_zero_undef_nxv7i64_unmasked(<vscale x 7 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI91_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI91_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI91_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI91_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI91_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI91_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI91_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI91_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.ctlz.nxv7i64(<vscale x 7 x i64> %va, i1 true, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI92_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI92_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI92_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI92_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI92_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI92_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI92_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI92_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 8 x i64> @llvm.vp.ctlz.nxv8i64(<vscale x 8 x i64> %va, i1 true, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }
 
 define <vscale x 8 x i64> @vp_ctlz_zero_undef_nxv8i64_unmasked(<vscale x 8 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI93_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI93_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI93_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI93_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI93_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI93_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI93_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI93_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.ctlz.nxv8i64(<vscale x 8 x i64> %va, i1 true, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a2, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v16, v8, v0.t
-; RV64-NEXT:    lui a3, %hi(.LCPI94_0)
-; RV64-NEXT:    ld a4, %lo(.LCPI94_0)(a3)
-; RV64-NEXT:    lui a3, %hi(.LCPI94_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI94_1)(a3)
 ; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
-; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vadd.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a5, %hi(.LCPI94_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI94_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI94_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI94_3)(a6)
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsub.vv v16, v16, v8, v0.t
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
+; RV64-NEXT:    vand.vx v8, v16, a4, v0.t
+; RV64-NEXT:    vsrl.vi v16, v16, 2, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi t0, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (t0) # Unknown-size Folded Spill
 ; RV64-NEXT:    bltu a0, a1, .LBB94_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    vsrl.vi v24, v16, 8
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    vsrl.vi v24, v16, 16
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    li a2, 32
 ; RV64-NEXT:    vsrl.vx v24, v16, a2
 ; RV64-NEXT:    vor.vv v16, v16, v24
 ; RV64-NEXT:    vnot.v v16, v16
-; RV64-NEXT:    lui a3, %hi(.LCPI95_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI95_0)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI95_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI95_1)(a4)
 ; RV64-NEXT:    vsrl.vi v24, v16, 1
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v24, a3
 ; RV64-NEXT:    vsub.vv v16, v16, v24
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v24, v16, a4
 ; RV64-NEXT:    vsrl.vi v16, v16, 2
 ; RV64-NEXT:    vand.vx v16, v16, a4
 ; RV64-NEXT:    vadd.vv v16, v24, v16
-; RV64-NEXT:    lui a5, %hi(.LCPI95_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI95_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI95_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI95_3)(a6)
 ; RV64-NEXT:    vsrl.vi v24, v16, 4
 ; RV64-NEXT:    vadd.vv v16, v16, v24
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v16, v16, a5
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v16, v16, a6
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v16, v16, a7
 ; RV64-NEXT:    bltu a0, a1, .LBB95_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB95_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/ctpop-sdnode.ll

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ctpop_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI18_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI18_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI18_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI18_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI18_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI18_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI18_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI18_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x i64> @llvm.ctpop.nxv1i64(<vscale x 1 x i64> %va)
   ret <vscale x 1 x i64> %a
 }
 declare <vscale x 1 x i64> @llvm.ctpop.nxv1i64(<vscale x 1 x i64>)
 

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ctpop_nxv2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI19_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI19_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI19_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI19_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI19_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI19_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI19_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI19_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 2 x i64> @llvm.ctpop.nxv2i64(<vscale x 2 x i64> %va)
   ret <vscale x 2 x i64> %a
 }
 declare <vscale x 2 x i64> @llvm.ctpop.nxv2i64(<vscale x 2 x i64>)
 

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ctpop_nxv4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI20_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI20_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI20_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI20_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI20_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI20_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI20_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI20_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x i64> @llvm.ctpop.nxv4i64(<vscale x 4 x i64> %va)
   ret <vscale x 4 x i64> %a
 }
 declare <vscale x 4 x i64> @llvm.ctpop.nxv4i64(<vscale x 4 x i64>)
 

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: ctpop_nxv8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI21_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI21_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI21_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI21_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI21_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI21_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI21_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI21_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %a = call <vscale x 8 x i64> @llvm.ctpop.nxv8i64(<vscale x 8 x i64> %va)
   ret <vscale x 8 x i64> %a
 }
 declare <vscale x 8 x i64> @llvm.ctpop.nxv8i64(<vscale x 8 x i64>)

llvm/test/CodeGen/RISCV/rvv/ctpop-vp.ll

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI36_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.ctpop.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }
 
 define <vscale x 1 x i64> @vp_ctpop_nxv1i64_unmasked(<vscale x 1 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv1i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI37_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI37_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.ctpop.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI38_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.ctpop.nxv2i64(<vscale x 2 x i64> %va, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }
 
 define <vscale x 2 x i64> @vp_ctpop_nxv2i64_unmasked(<vscale x 2 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI39_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI39_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.ctpop.nxv2i64(<vscale x 2 x i64> %va, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.ctpop.nxv4i64(<vscale x 4 x i64> %va, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }
 
 define <vscale x 4 x i64> @vp_ctpop_nxv4i64_unmasked(<vscale x 4 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.ctpop.nxv4i64(<vscale x 4 x i64> %va, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv7i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 7 x i64> @llvm.vp.ctpop.nxv7i64(<vscale x 7 x i64> %va, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }
 
 define <vscale x 7 x i64> @vp_ctpop_nxv7i64_unmasked(<vscale x 7 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv7i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.ctpop.nxv7i64(<vscale x 7 x i64> %va, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI44_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 8 x i64> @llvm.vp.ctpop.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }
 
 define <vscale x 8 x i64> @vp_ctpop_nxv8i64_unmasked(<vscale x 8 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI45_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI45_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.ctpop.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }

 ; RV64-LABEL: vp_ctpop_nxv16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    addi sp, sp, -16
 ; RV64-NEXT:    .cfi_def_cfa_offset 16
 ; RV64-NEXT:    csrr a1, vlenb
 ; RV64-NEXT:    slli a1, a1, 4
 ; RV64-NEXT:    sub sp, sp, a1
 ; RV64-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
-; RV64-NEXT:    vmv1r.v v24, v0
 ; RV64-NEXT:    csrr a1, vlenb
 ; RV64-NEXT:    slli a1, a1, 3
 ; RV64-NEXT:    add a1, sp, a1
 ; RV64-NEXT:    addi a1, a1, 16
-; RV64-NEXT:    vs8r.v v8, (a1) # Unknown-size Folded Spill
+; RV64-NEXT:    vs8r.v v16, (a1) # Unknown-size Folded Spill
 ; RV64-NEXT:    csrr a1, vlenb
 ; RV64-NEXT:    srli a2, a1, 3
 ; RV64-NEXT:    vsetvli a3, zero, e8, mf4, ta, ma
-; RV64-NEXT:    vslidedown.vx v0, v0, a2
-; RV64-NEXT:    sub a2, a0, a1
-; RV64-NEXT:    sltu a3, a0, a2
-; RV64-NEXT:    addi a3, a3, -1
-; RV64-NEXT:    and a2, a3, a2
+; RV64-NEXT:    vslidedown.vx v24, v0, a2
+; RV64-NEXT:    mv a2, a0
+; RV64-NEXT:    bltu a0, a1, .LBB46_2
+; RV64-NEXT:  # %bb.1:
+; RV64-NEXT:    mv a2, a1
+; RV64-NEXT:  .LBB46_2:
 ; RV64-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
-; RV64-NEXT:    lui a2, %hi(.LCPI46_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI46_0)(a2)
-; RV64-NEXT:    lui a2, %hi(.LCPI46_1)
-; RV64-NEXT:    ld a2, %lo(.LCPI46_1)(a2)
-; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
+; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
+; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
-; RV64-NEXT:    vadd.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a4, %hi(.LCPI46_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI46_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI46_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI46_3)(a5)
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    addi a7, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a7) # Unknown-size Folded Spill
-; RV64-NEXT:    bltu a0, a1, .LBB46_2
-; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    mv a0, a1
-; RV64-NEXT:  .LBB46_2:
+; RV64-NEXT:    sub a1, a0, a1
+; RV64-NEXT:    sltu a0, a0, a1
+; RV64-NEXT:    addi a0, a0, -1
+; RV64-NEXT:    and a0, a0, a1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vmv1r.v v0, v24
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add a0, sp, a0
 ; RV64-NEXT:    addi a0, a0, 16
-; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
-; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
-; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
-; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
+; RV64-NEXT:    vsub.vv v16, v8, v16, v0.t
+; RV64-NEXT:    vand.vx v8, v16, a3, v0.t
+; RV64-NEXT:    vsrl.vi v16, v16, 2, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
-; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
+; RV64-NEXT:    vsrl.vx v16, v8, a6, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
-; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 4
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16
 ; RV64-NEXT:    ret
   %v = call <vscale x 16 x i64> @llvm.vp.ctpop.nxv16i64(<vscale x 16 x i64> %va, <vscale x 16 x i1> %m, i32 %evl)
   ret <vscale x 16 x i64> %v
 }

 ; RV32-NEXT:    slli a0, a0, 5
 ; RV32-NEXT:    add sp, sp, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_nxv16i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    csrr a1, vlenb
-; RV64-NEXT:    sub a2, a0, a1
-; RV64-NEXT:    sltu a3, a0, a2
-; RV64-NEXT:    addi a3, a3, -1
-; RV64-NEXT:    and a2, a3, a2
-; RV64-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
-; RV64-NEXT:    lui a2, %hi(.LCPI47_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI47_0)(a2)
-; RV64-NEXT:    lui a3, %hi(.LCPI47_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI47_1)(a3)
-; RV64-NEXT:    vsrl.vi v24, v16, 1
-; RV64-NEXT:    vand.vx v24, v24, a2
-; RV64-NEXT:    vsub.vv v16, v16, v24
-; RV64-NEXT:    vand.vx v24, v16, a3
-; RV64-NEXT:    vsrl.vi v16, v16, 2
-; RV64-NEXT:    vand.vx v16, v16, a3
-; RV64-NEXT:    vadd.vv v16, v24, v16
-; RV64-NEXT:    lui a4, %hi(.LCPI47_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI47_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI47_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI47_3)(a5)
-; RV64-NEXT:    vsrl.vi v24, v16, 4
-; RV64-NEXT:    vadd.vv v16, v16, v24
-; RV64-NEXT:    vand.vx v16, v16, a4
-; RV64-NEXT:    vmul.vx v16, v16, a5
-; RV64-NEXT:    li a6, 56
-; RV64-NEXT:    vsrl.vx v16, v16, a6
+; RV64-NEXT:    mv a2, a0
 ; RV64-NEXT:    bltu a0, a1, .LBB47_2
 ; RV64-NEXT:  # %bb.1:
-; RV64-NEXT:    mv a0, a1
+; RV64-NEXT:    mv a2, a1
 ; RV64-NEXT:  .LBB47_2:
-; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; RV64-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    vand.vx v24, v24, a2
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v8, a3
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vadd.vv v8, v24, v8
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5
+; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6
+; RV64-NEXT:    sub a1, a0, a1
+; RV64-NEXT:    sltu a0, a0, a1
+; RV64-NEXT:    addi a0, a0, -1
+; RV64-NEXT:    and a0, a0, a1
+; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; RV64-NEXT:    vsrl.vi v24, v16, 1
+; RV64-NEXT:    vand.vx v24, v24, a2
+; RV64-NEXT:    vsub.vv v16, v16, v24
+; RV64-NEXT:    vand.vx v24, v16, a3
+; RV64-NEXT:    vsrl.vi v16, v16, 2
+; RV64-NEXT:    vand.vx v16, v16, a3
+; RV64-NEXT:    vadd.vv v16, v24, v16
+; RV64-NEXT:    vsrl.vi v24, v16, 4
+; RV64-NEXT:    vadd.vv v16, v16, v24
+; RV64-NEXT:    vand.vx v16, v16, a4
+; RV64-NEXT:    vmul.vx v16, v16, a5
+; RV64-NEXT:    vsrl.vx v16, v16, a6
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 16 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
   %v = call <vscale x 16 x i64> @llvm.vp.ctpop.nxv16i64(<vscale x 16 x i64> %va, <vscale x 16 x i1> %m, i32 %evl)
   ret <vscale x 16 x i64> %v
 }

llvm/test/CodeGen/RISCV/rvv/cttz-sdnode.ll

 ;
 ; RV64I-LABEL: cttz_nxv1i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
 ; RV64I-NEXT:    vsub.vx v9, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v9
-; RV64I-NEXT:    lui a0, %hi(.LCPI18_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI18_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI18_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI18_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v9, v9, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v9
-; RV64I-NEXT:    vand.vx v9, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v9, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v9, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI18_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI18_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI18_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI18_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v9
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32F-LABEL: cttz_nxv1i64:
 ; RV32F:       # %bb.0:
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV32F-NEXT:    vmseq.vx v9, v8, zero

 ;
 ; RV64I-LABEL: cttz_nxv2i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
 ; RV64I-NEXT:    vsub.vx v10, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v10
-; RV64I-NEXT:    lui a0, %hi(.LCPI19_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI19_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v10, v10, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v10
-; RV64I-NEXT:    vand.vx v10, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v10, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v10, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI19_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI19_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI19_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI19_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v10
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32F-LABEL: cttz_nxv2i64:
 ; RV32F:       # %bb.0:
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV32F-NEXT:    vmseq.vx v10, v8, zero

 ;
 ; RV64I-LABEL: cttz_nxv4i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
 ; RV64I-NEXT:    vsub.vx v12, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v12
-; RV64I-NEXT:    lui a0, %hi(.LCPI20_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI20_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI20_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI20_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v12, v12, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v12
-; RV64I-NEXT:    vand.vx v12, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v12, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v12, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI20_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI20_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI20_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI20_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v12
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32F-LABEL: cttz_nxv4i64:
 ; RV32F:       # %bb.0:
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV32F-NEXT:    vmseq.vx v12, v8, zero

 ;
 ; RV64I-LABEL: cttz_nxv8i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
 ; RV64I-NEXT:    vsub.vx v16, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v16
-; RV64I-NEXT:    lui a0, %hi(.LCPI21_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI21_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI21_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI21_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v16, v16, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v16
-; RV64I-NEXT:    vand.vx v16, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v16, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v16, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI21_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI21_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI21_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI21_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v16
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; RV32F-LABEL: cttz_nxv8i64:
 ; RV32F:       # %bb.0:
 ; RV32F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV32F-NEXT:    vmseq.vx v16, v8, zero

 ;
 ; RV64I-LABEL: cttz_zero_undef_nxv1i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
 ; RV64I-NEXT:    vsub.vx v9, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v9
-; RV64I-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI40_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI40_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v9, v9, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v9
-; RV64I-NEXT:    vand.vx v9, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v9, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v9, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI40_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI40_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v9
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: cttz_zero_undef_nxv1i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v9, v8, 0

 ;
 ; RV64I-LABEL: cttz_zero_undef_nxv2i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
 ; RV64I-NEXT:    vsub.vx v10, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v10
-; RV64I-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI41_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI41_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v10, v10, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v10
-; RV64I-NEXT:    vand.vx v10, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v10, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v10, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI41_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI41_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v10
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: cttz_zero_undef_nxv2i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v10, v8, 0

 ;
 ; RV64I-LABEL: cttz_zero_undef_nxv4i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
 ; RV64I-NEXT:    vsub.vx v12, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v12
-; RV64I-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI42_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI42_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v12, v12, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v12
-; RV64I-NEXT:    vand.vx v12, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v12, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v12, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI42_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI42_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v12, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v12
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: cttz_zero_undef_nxv4i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v12, v8, 0

 ;
 ; RV64I-LABEL: cttz_zero_undef_nxv8i64:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    li a0, 1
 ; RV64I-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
 ; RV64I-NEXT:    vsub.vx v16, v8, a0
 ; RV64I-NEXT:    vnot.v v8, v8
 ; RV64I-NEXT:    vand.vv v8, v8, v16
-; RV64I-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64I-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI43_1)
-; RV64I-NEXT:    ld a1, %lo(.LCPI43_1)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 1
+; RV64I-NEXT:    lui a0, 349525
+; RV64I-NEXT:    addiw a0, a0, 1365
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v16, v16, a0
 ; RV64I-NEXT:    vsub.vv v8, v8, v16
-; RV64I-NEXT:    vand.vx v16, v8, a1
+; RV64I-NEXT:    lui a0, 209715
+; RV64I-NEXT:    addiw a0, a0, 819
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vand.vx v16, v8, a0
 ; RV64I-NEXT:    vsrl.vi v8, v8, 2
-; RV64I-NEXT:    vand.vx v8, v8, a1
+; RV64I-NEXT:    vand.vx v8, v8, a0
 ; RV64I-NEXT:    vadd.vv v8, v16, v8
-; RV64I-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64I-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
-; RV64I-NEXT:    lui a1, %hi(.LCPI43_3)
-; RV64I-NEXT:    ld a1, %lo(.LCPI43_3)(a1)
 ; RV64I-NEXT:    vsrl.vi v16, v8, 4
 ; RV64I-NEXT:    vadd.vv v8, v8, v16
+; RV64I-NEXT:    lui a0, 61681
+; RV64I-NEXT:    addiw a0, a0, -241
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
 ; RV64I-NEXT:    vand.vx v8, v8, a0
-; RV64I-NEXT:    vmul.vx v8, v8, a1
+; RV64I-NEXT:    lui a0, 4112
+; RV64I-NEXT:    addiw a0, a0, 257
+; RV64I-NEXT:    slli a1, a0, 32
+; RV64I-NEXT:    add a0, a0, a1
+; RV64I-NEXT:    vmul.vx v8, v8, a0
 ; RV64I-NEXT:    li a0, 56
 ; RV64I-NEXT:    vsrl.vx v8, v8, a0
 ; RV64I-NEXT:    ret
 ;
 ; CHECK-F-LABEL: cttz_zero_undef_nxv8i64:
 ; CHECK-F:       # %bb.0:
 ; CHECK-F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; CHECK-F-NEXT:    vrsub.vi v16, v8, 0

llvm/test/CodeGen/RISCV/rvv/cttz-vp.ll

 ;
 ; RV64-LABEL: vp_cttz_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v9, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI36_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI36_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI36_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI36_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.cttz.nxv1i64(<vscale x 1 x i64> %va, i1 false, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }
 
 define <vscale x 1 x i64> @vp_cttz_nxv1i64_unmasked(<vscale x 1 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_nxv1i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v9
-; RV64-NEXT:    lui a0, %hi(.LCPI37_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI37_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI37_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI37_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI37_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.cttz.nxv1i64(<vscale x 1 x i64> %va, i1 false, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_nxv2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v10, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI38_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI38_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI38_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI38_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.cttz.nxv2i64(<vscale x 2 x i64> %va, i1 false, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }
 
 define <vscale x 2 x i64> @vp_cttz_nxv2i64_unmasked(<vscale x 2 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_nxv2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v10
-; RV64-NEXT:    lui a0, %hi(.LCPI39_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI39_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI39_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI39_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI39_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.cttz.nxv2i64(<vscale x 2 x i64> %va, i1 false, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_nxv4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v12, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI40_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI40_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI40_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI40_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.cttz.nxv4i64(<vscale x 4 x i64> %va, i1 false, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }
 
 define <vscale x 4 x i64> @vp_cttz_nxv4i64_unmasked(<vscale x 4 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_nxv4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v12
-; RV64-NEXT:    lui a0, %hi(.LCPI41_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI41_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI41_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI41_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI41_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.cttz.nxv4i64(<vscale x 4 x i64> %va, i1 false, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_nxv7i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI42_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI42_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI42_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI42_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 7 x i64> @llvm.vp.cttz.nxv7i64(<vscale x 7 x i64> %va, i1 false, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }
 
 define <vscale x 7 x i64> @vp_cttz_nxv7i64_unmasked(<vscale x 7 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_nxv7i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI43_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI43_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI43_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI43_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI43_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.cttz.nxv7i64(<vscale x 7 x i64> %va, i1 false, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_nxv8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI44_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI44_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI44_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI44_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 8 x i64> @llvm.vp.cttz.nxv8i64(<vscale x 8 x i64> %va, i1 false, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }
 
 define <vscale x 8 x i64> @vp_cttz_nxv8i64_unmasked(<vscale x 8 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_nxv8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI45_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI45_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI45_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI45_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI45_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.cttz.nxv8i64(<vscale x 8 x i64> %va, i1 false, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }

 ; RV64-NEXT:    sltu a3, a0, a2
 ; RV64-NEXT:    addi a3, a3, -1
 ; RV64-NEXT:    and a3, a3, a2
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a3, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v8, v16, a2, v0.t
 ; RV64-NEXT:    vnot.v v16, v16, v0.t
 ; RV64-NEXT:    vand.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a3, %hi(.LCPI46_0)
-; RV64-NEXT:    ld a4, %lo(.LCPI46_0)(a3)
-; RV64-NEXT:    lui a3, %hi(.LCPI46_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI46_1)(a3)
 ; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
-; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vadd.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a5, %hi(.LCPI46_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI46_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI46_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI46_3)(a6)
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsub.vv v16, v16, v8, v0.t
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
+; RV64-NEXT:    vand.vx v8, v16, a4, v0.t
+; RV64-NEXT:    vsrl.vi v16, v16, 2, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi t0, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (t0) # Unknown-size Folded Spill
 ; RV64-NEXT:    bltu a0, a1, .LBB46_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB46_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vmv1r.v v0, v24
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add a0, sp, a0
 ; RV64-NEXT:    addi a0, a0, 16
 ; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload
 ; RV64-NEXT:    vsub.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    sltu a3, a0, a2
 ; RV64-NEXT:    addi a3, a3, -1
 ; RV64-NEXT:    and a3, a3, a2
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a3, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v24, v16, a2
 ; RV64-NEXT:    vnot.v v16, v16
 ; RV64-NEXT:    vand.vv v16, v16, v24
-; RV64-NEXT:    lui a3, %hi(.LCPI47_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI47_0)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI47_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI47_1)(a4)
 ; RV64-NEXT:    vsrl.vi v24, v16, 1
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v24, a3
 ; RV64-NEXT:    vsub.vv v16, v16, v24
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v24, v16, a4
 ; RV64-NEXT:    vsrl.vi v16, v16, 2
 ; RV64-NEXT:    vand.vx v16, v16, a4
 ; RV64-NEXT:    vadd.vv v16, v24, v16
-; RV64-NEXT:    lui a5, %hi(.LCPI47_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI47_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI47_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI47_3)(a6)
 ; RV64-NEXT:    vsrl.vi v24, v16, 4
 ; RV64-NEXT:    vadd.vv v16, v16, v24
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v16, v16, a5
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v16, v16, a6
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v16, v16, a7
 ; RV64-NEXT:    bltu a0, a1, .LBB47_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB47_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v9, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI84_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI84_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI84_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI84_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI84_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI84_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI84_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI84_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 1 x i64> @llvm.vp.cttz.nxv1i64(<vscale x 1 x i64> %va, i1 true, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }
 
 define <vscale x 1 x i64> @vp_cttz_zero_undef_nxv1i64_unmasked(<vscale x 1 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv1i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v9
-; RV64-NEXT:    lui a0, %hi(.LCPI85_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI85_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI85_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI85_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI85_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI85_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI85_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI85_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 1 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> poison, <vscale x 1 x i32> zeroinitializer
   %v = call <vscale x 1 x i64> @llvm.vp.cttz.nxv1i64(<vscale x 1 x i64> %va, i1 true, <vscale x 1 x i1> %m, i32 %evl)
   ret <vscale x 1 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v10, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI86_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI86_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI86_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI86_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI86_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI86_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI86_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI86_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 2 x i64> @llvm.vp.cttz.nxv2i64(<vscale x 2 x i64> %va, i1 true, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }
 
 define <vscale x 2 x i64> @vp_cttz_zero_undef_nxv2i64_unmasked(<vscale x 2 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v10
-; RV64-NEXT:    lui a0, %hi(.LCPI87_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI87_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI87_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI87_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI87_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI87_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI87_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI87_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 2 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
   %v = call <vscale x 2 x i64> @llvm.vp.cttz.nxv2i64(<vscale x 2 x i64> %va, i1 true, <vscale x 2 x i1> %m, i32 %evl)
   ret <vscale x 2 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v12, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI88_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI88_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI88_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI88_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI88_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI88_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI88_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI88_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 4 x i64> @llvm.vp.cttz.nxv4i64(<vscale x 4 x i64> %va, i1 true, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }
 
 define <vscale x 4 x i64> @vp_cttz_zero_undef_nxv4i64_unmasked(<vscale x 4 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v12
-; RV64-NEXT:    lui a0, %hi(.LCPI89_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI89_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI89_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI89_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI89_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI89_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI89_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI89_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 4 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
   %v = call <vscale x 4 x i64> @llvm.vp.cttz.nxv4i64(<vscale x 4 x i64> %va, i1 true, <vscale x 4 x i1> %m, i32 %evl)
   ret <vscale x 4 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv7i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI90_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI90_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI90_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI90_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI90_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI90_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI90_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI90_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 7 x i64> @llvm.vp.cttz.nxv7i64(<vscale x 7 x i64> %va, i1 true, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }
 
 define <vscale x 7 x i64> @vp_cttz_zero_undef_nxv7i64_unmasked(<vscale x 7 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv7i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI91_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI91_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI91_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI91_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI91_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI91_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI91_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI91_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 7 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 7 x i1> %head, <vscale x 7 x i1> poison, <vscale x 7 x i32> zeroinitializer
   %v = call <vscale x 7 x i64> @llvm.vp.cttz.nxv7i64(<vscale x 7 x i64> %va, i1 true, <vscale x 7 x i1> %m, i32 %evl)
   ret <vscale x 7 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI92_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI92_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI92_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI92_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI92_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI92_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI92_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI92_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <vscale x 8 x i64> @llvm.vp.cttz.nxv8i64(<vscale x 8 x i64> %va, i1 true, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }
 
 define <vscale x 8 x i64> @vp_cttz_zero_undef_nxv8i64_unmasked(<vscale x 8 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_nxv8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI93_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI93_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI93_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI93_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI93_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI93_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI93_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI93_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <vscale x 8 x i1> poison, i1 true, i32 0
   %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
   %v = call <vscale x 8 x i64> @llvm.vp.cttz.nxv8i64(<vscale x 8 x i64> %va, i1 true, <vscale x 8 x i1> %m, i32 %evl)
   ret <vscale x 8 x i64> %v
 }

 ; RV64-NEXT:    sltu a3, a0, a2
 ; RV64-NEXT:    addi a3, a3, -1
 ; RV64-NEXT:    and a3, a3, a2
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a3, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v8, v16, a2, v0.t
 ; RV64-NEXT:    vnot.v v16, v16, v0.t
 ; RV64-NEXT:    vand.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a3, %hi(.LCPI94_0)
-; RV64-NEXT:    ld a4, %lo(.LCPI94_0)(a3)
-; RV64-NEXT:    lui a3, %hi(.LCPI94_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI94_1)(a3)
 ; RV64-NEXT:    vsrl.vi v8, v16, 1, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
-; RV64-NEXT:    vsub.vv v8, v16, v8, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
-; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
-; RV64-NEXT:    vadd.vv v16, v16, v8, v0.t
-; RV64-NEXT:    lui a5, %hi(.LCPI94_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI94_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI94_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI94_3)(a6)
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsub.vv v16, v16, v8, v0.t
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
+; RV64-NEXT:    vand.vx v8, v16, a4, v0.t
+; RV64-NEXT:    vsrl.vi v16, v16, 2, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi t0, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (t0) # Unknown-size Folded Spill
 ; RV64-NEXT:    bltu a0, a1, .LBB94_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB94_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vmv1r.v v0, v24
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add a0, sp, a0
 ; RV64-NEXT:    addi a0, a0, 16
 ; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload
 ; RV64-NEXT:    vsub.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a4, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    vsrl.vx v8, v8, a7, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    sltu a3, a0, a2
 ; RV64-NEXT:    addi a3, a3, -1
 ; RV64-NEXT:    and a3, a3, a2
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a3, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v24, v16, a2
 ; RV64-NEXT:    vnot.v v16, v16
 ; RV64-NEXT:    vand.vv v16, v16, v24
-; RV64-NEXT:    lui a3, %hi(.LCPI95_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI95_0)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI95_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI95_1)(a4)
 ; RV64-NEXT:    vsrl.vi v24, v16, 1
+; RV64-NEXT:    lui a3, 349525
+; RV64-NEXT:    addiw a3, a3, 1365
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v24, a3
 ; RV64-NEXT:    vsub.vv v16, v16, v24
+; RV64-NEXT:    lui a4, 209715
+; RV64-NEXT:    addiw a4, a4, 819
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v24, v16, a4
 ; RV64-NEXT:    vsrl.vi v16, v16, 2
 ; RV64-NEXT:    vand.vx v16, v16, a4
 ; RV64-NEXT:    vadd.vv v16, v24, v16
-; RV64-NEXT:    lui a5, %hi(.LCPI95_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI95_2)(a5)
-; RV64-NEXT:    lui a6, %hi(.LCPI95_3)
-; RV64-NEXT:    ld a6, %lo(.LCPI95_3)(a6)
 ; RV64-NEXT:    vsrl.vi v24, v16, 4
 ; RV64-NEXT:    vadd.vv v16, v16, v24
+; RV64-NEXT:    lui a5, 61681
+; RV64-NEXT:    addiw a5, a5, -241
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vand.vx v16, v16, a5
+; RV64-NEXT:    lui a6, 4112
+; RV64-NEXT:    addiw a6, a6, 257
+; RV64-NEXT:    slli a7, a6, 32
+; RV64-NEXT:    add a6, a6, a7
 ; RV64-NEXT:    vmul.vx v16, v16, a6
 ; RV64-NEXT:    li a7, 56
 ; RV64-NEXT:    vsrl.vx v16, v16, a7
 ; RV64-NEXT:    bltu a0, a1, .LBB95_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB95_2:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse-vp.ll

 ; RV64-NEXT:    vsrl.vx v11, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v11, v11, a3, v0.t
 ; RV64-NEXT:    vor.vv v10, v11, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v11, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v11, v11, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v11, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v9, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.bitreverse.v2i64(<2 x i64> %va, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v11, v8, a4
 ; RV64-NEXT:    vand.vx v11, v11, a3
 ; RV64-NEXT:    vor.vv v10, v11, v10
 ; RV64-NEXT:    vsrl.vi v11, v8, 24
 ; RV64-NEXT:    vand.vx v11, v11, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v11
-; RV64-NEXT:    lui a0, %hi(.LCPI25_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI25_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI25_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.bitreverse.v2i64(<2 x i64> %va, <2 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v14, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v14, v14, a3, v0.t
 ; RV64-NEXT:    vor.vv v12, v14, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v14, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v14, v14, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v14, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v10, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.bitreverse.v4i64(<4 x i64> %va, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v14, v8, a4
 ; RV64-NEXT:    vand.vx v14, v14, a3
 ; RV64-NEXT:    vor.vv v12, v14, v12
 ; RV64-NEXT:    vsrl.vi v14, v8, 24
 ; RV64-NEXT:    vand.vx v14, v14, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v14
-; RV64-NEXT:    lui a0, %hi(.LCPI27_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI27_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI27_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.bitreverse.v4i64(<4 x i64> %va, <4 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v20, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v20, v20, a3, v0.t
 ; RV64-NEXT:    vor.vv v16, v20, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v20, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v20, v20, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v20, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v12, v8, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.bitreverse.v8i64(<8 x i64> %va, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vx v20, v8, a4
 ; RV64-NEXT:    vand.vx v20, v20, a3
 ; RV64-NEXT:    vor.vv v16, v20, v16
 ; RV64-NEXT:    vsrl.vi v20, v8, 24
 ; RV64-NEXT:    vand.vx v20, v20, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v20
-; RV64-NEXT:    lui a0, %hi(.LCPI29_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI29_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI29_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.bitreverse.v8i64(<8 x i64> %va, <8 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vor.vv v24, v16, v24, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24, v0.t
-; RV64-NEXT:    addi a1, sp, 16
-; RV64-NEXT:    vl8r.v v16, (a1) # Unknown-size Folded Reload
-; RV64-NEXT:    vor.vv v16, v16, v8, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    addi a0, sp, 16
+; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_1)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 4, v0.t
-; RV64-NEXT:    vor.vv v16, v8, v16, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_2)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 2, v0.t
-; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16

 ; RV64-NEXT:    vsrl.vx v0, v8, a4
 ; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
 ; RV64-NEXT:    vsrl.vi v0, v8, 24
 ; RV64-NEXT:    vand.vx v0, v0, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v0
-; RV64-NEXT:    lui a0, %hi(.LCPI31_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI31_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI31_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.bitreverse.v15i64(<15 x i64> %va, <15 x i1> %m, i32 %evl)

 ; RV64-NEXT:    vsrl.vx v16, v8, a4, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
 ; RV64-NEXT:    vor.vv v24, v16, v24, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 24, v0.t
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 8, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24, v0.t
-; RV64-NEXT:    addi a1, sp, 16
-; RV64-NEXT:    vl8r.v v16, (a1) # Unknown-size Folded Reload
-; RV64-NEXT:    vor.vv v16, v16, v8, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 4, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    addi a0, sp, 16
+; RV64-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_1)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 4, v0.t
-; RV64-NEXT:    vor.vv v16, v8, v16, v0.t
-; RV64-NEXT:    vsrl.vi v8, v16, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 4, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
+; RV64-NEXT:    vsrl.vi v16, v8, 2, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_2)(a0)
-; RV64-NEXT:    vsll.vi v16, v16, 2, v0.t
-; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
+; RV64-NEXT:    vsll.vi v8, v8, 2, v0.t
+; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vsll.vi v8, v8, 1, v0.t
 ; RV64-NEXT:    vor.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    csrr a0, vlenb
 ; RV64-NEXT:    slli a0, a0, 3
 ; RV64-NEXT:    add sp, sp, a0
 ; RV64-NEXT:    addi sp, sp, 16

 ; RV64-NEXT:    vsrl.vx v0, v8, a4
 ; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
 ; RV64-NEXT:    vsrl.vi v0, v8, 24
 ; RV64-NEXT:    vand.vx v0, v0, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 8
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v0
-; RV64-NEXT:    lui a0, %hi(.LCPI33_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_0)(a0)
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI33_1)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_1)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 2
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI33_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_2)(a0)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.bitreverse.v16i64(<16 x i64> %va, <16 x i1> %m, i32 %evl)

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll

 ; RV64-NEXT:    vsll.vi v10, v10, 8
 ; RV64-NEXT:    vand.vx v11, v8, a4
 ; RV64-NEXT:    vsll.vi v11, v11, 24
 ; RV64-NEXT:    vor.vv v10, v11, v10
 ; RV64-NEXT:    vsll.vx v11, v8, a1
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vsll.vx v8, v8, a2
 ; RV64-NEXT:    vor.vv v8, v11, v8
-; RV64-NEXT:    lui a1, %hi(.LCPI2_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI2_0)(a1)
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
+; RV64-NEXT:    lui a1, 61681
+; RV64-NEXT:    addiw a1, a1, -241
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vand.vx v9, v9, a1
 ; RV64-NEXT:    vand.vx v8, v8, a1
-; RV64-NEXT:    lui a1, %hi(.LCPI2_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI2_1)(a1)
 ; RV64-NEXT:    vsll.vi v8, v8, 4
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 2
+; RV64-NEXT:    lui a1, 209715
+; RV64-NEXT:    addiw a1, a1, 819
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vand.vx v9, v9, a1
 ; RV64-NEXT:    vand.vx v8, v8, a1
-; RV64-NEXT:    lui a1, %hi(.LCPI2_2)
-; RV64-NEXT:    ld a1, %lo(.LCPI2_2)(a1)
 ; RV64-NEXT:    vsll.vi v8, v8, 2
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vand.vx v9, v9, a1
 ; RV64-NEXT:    vand.vx v8, v8, a1
 ; RV64-NEXT:    vadd.vv v8, v8, v8
 ; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    vse64.v v8, (a0)
 ; RV64-NEXT:    ret
   %a = load <2 x i64>, ptr %x
   %b = load <2 x i64>, ptr %y

 ; LMULMAX2-RV64-NEXT:    vsll.vi v12, v12, 8
 ; LMULMAX2-RV64-NEXT:    vand.vx v14, v8, a4
 ; LMULMAX2-RV64-NEXT:    vsll.vi v14, v14, 24
 ; LMULMAX2-RV64-NEXT:    vor.vv v12, v14, v12
 ; LMULMAX2-RV64-NEXT:    vsll.vx v14, v8, a1
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a3
 ; LMULMAX2-RV64-NEXT:    vsll.vx v8, v8, a2
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v14, v8
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI5_0)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI5_0)(a1)
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v8, v12
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v8, v10
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v10, v8, 4
+; LMULMAX2-RV64-NEXT:    lui a1, 61681
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI5_1)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI5_1)(a1)
 ; LMULMAX2-RV64-NEXT:    vsll.vi v8, v8, 4
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v10, v8
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v10, v8, 2
+; LMULMAX2-RV64-NEXT:    lui a1, 209715
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI5_2)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI5_2)(a1)
 ; LMULMAX2-RV64-NEXT:    vsll.vi v8, v8, 2
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v10, v8
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v10, v8, 1
+; LMULMAX2-RV64-NEXT:    lui a1, 349525
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    vadd.vv v8, v8, v8
 ; LMULMAX2-RV64-NEXT:    vor.vv v8, v10, v8
 ; LMULMAX2-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64-NEXT:    ret
 ;
 ; LMULMAX1-RV32-LABEL: bitreverse_v4i64:

 ; LMULMAX1-RV64-NEXT:    vsll.vi v11, v11, 8
 ; LMULMAX1-RV64-NEXT:    vand.vx v12, v9, a5
 ; LMULMAX1-RV64-NEXT:    vsll.vi v12, v12, 24
 ; LMULMAX1-RV64-NEXT:    vor.vv v11, v12, v11
 ; LMULMAX1-RV64-NEXT:    vsll.vx v12, v9, a2
 ; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a4
 ; LMULMAX1-RV64-NEXT:    vsll.vx v9, v9, a3
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v12, v9
-; LMULMAX1-RV64-NEXT:    lui a7, %hi(.LCPI5_0)
-; LMULMAX1-RV64-NEXT:    ld a7, %lo(.LCPI5_0)(a7)
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v9, v11
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v9, v10
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 4
+; LMULMAX1-RV64-NEXT:    lui a7, 61681
+; LMULMAX1-RV64-NEXT:    addiw a7, a7, -241
+; LMULMAX1-RV64-NEXT:    slli t0, a7, 32
+; LMULMAX1-RV64-NEXT:    add a7, a7, t0
 ; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, a7
 ; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a7
-; LMULMAX1-RV64-NEXT:    lui t0, %hi(.LCPI5_1)
-; LMULMAX1-RV64-NEXT:    ld t0, %lo(.LCPI5_1)(t0)
 ; LMULMAX1-RV64-NEXT:    vsll.vi v9, v9, 4
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v10, v9
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 2
+; LMULMAX1-RV64-NEXT:    lui t0, 209715
+; LMULMAX1-RV64-NEXT:    addiw t0, t0, 819
+; LMULMAX1-RV64-NEXT:    slli t1, t0, 32
+; LMULMAX1-RV64-NEXT:    add t0, t0, t1
 ; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, t0
 ; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, t0
-; LMULMAX1-RV64-NEXT:    lui t1, %hi(.LCPI5_2)
-; LMULMAX1-RV64-NEXT:    ld t1, %lo(.LCPI5_2)(t1)
 ; LMULMAX1-RV64-NEXT:    vsll.vi v9, v9, 2
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v10, v9
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 1
+; LMULMAX1-RV64-NEXT:    lui t1, 349525
+; LMULMAX1-RV64-NEXT:    addiw t1, t1, 1365
+; LMULMAX1-RV64-NEXT:    slli t2, t1, 32
+; LMULMAX1-RV64-NEXT:    add t1, t1, t2
 ; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, t1
 ; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, t1
 ; LMULMAX1-RV64-NEXT:    vadd.vv v9, v9, v9
 ; LMULMAX1-RV64-NEXT:    vor.vv v9, v10, v9
 ; LMULMAX1-RV64-NEXT:    vsrl.vx v10, v8, a2
 ; LMULMAX1-RV64-NEXT:    vsrl.vx v11, v8, a3
 ; LMULMAX1-RV64-NEXT:    vand.vx v11, v11, a4
 ; LMULMAX1-RV64-NEXT:    vor.vv v10, v11, v10

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz-vp.ll

 ; RV64-NEXT:    vsrl.vi v9, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.ctlz.v2i64(<2 x i64> %va, i1 false, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }
 
 define <2 x i64> @vp_ctlz_v2i64_unmasked(<2 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v9, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI25_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI25_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.ctlz.v2i64(<2 x i64> %va, i1 false, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v10, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.ctlz.v4i64(<4 x i64> %va, i1 false, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }
 
 define <4 x i64> @vp_ctlz_v4i64_unmasked(<4 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v10, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vsrl.vi v10, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI27_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI27_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.ctlz.v4i64(<4 x i64> %va, i1 false, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v12, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.ctlz.v8i64(<8 x i64> %va, i1 false, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }
 
 define <8 x i64> @vp_ctlz_v8i64_unmasked(<8 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v12, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vsrl.vi v12, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI29_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI29_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.ctlz.v8i64(<8 x i64> %va, i1 false, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <15 x i64> @llvm.vp.ctlz.v15i64(<15 x i64> %va, i1 false, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }
 
 define <15 x i64> @vp_ctlz_v15i64_unmasked(<15 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI31_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI31_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.ctlz.v15i64(<15 x i64> %va, i1 false, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <16 x i64> @llvm.vp.ctlz.v16i64(<16 x i64> %va, i1 false, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }
 
 define <16 x i64> @vp_ctlz_v16i64_unmasked(<16 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI33_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI33_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.ctlz.v16i64(<16 x i64> %va, i1 false, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a1, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a2, %hi(.LCPI34_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI34_0)(a2)
-; RV64-NEXT:    lui a2, %hi(.LCPI34_1)
-; RV64-NEXT:    ld a2, %lo(.LCPI34_1)(a2)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
+; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a4, %hi(.LCPI34_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI34_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI34_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI34_3)(a5)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    addi a7, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a7) # Unknown-size Folded Spill
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a6, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    vsrl.vi v24, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vsrl.vi v24, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    li a1, 32
 ; RV64-NEXT:    vsrl.vx v24, v8, a1
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a2, %hi(.LCPI35_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI35_0)(a2)
-; RV64-NEXT:    lui a3, %hi(.LCPI35_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI35_1)(a3)
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    vand.vx v24, v24, a2
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v8, a3
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vadd.vv v8, v24, v8
-; RV64-NEXT:    lui a4, %hi(.LCPI35_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI35_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI35_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI35_3)(a5)
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1
 ; RV64-NEXT:    and a0, a0, a7
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

 ; RV64-NEXT:    vsrl.vi v9, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vsrl.vi v9, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v9, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI60_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI60_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI60_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI60_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI60_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI60_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI60_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI60_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.ctlz.v2i64(<2 x i64> %va, i1 true, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }
 
 define <2 x i64> @vp_ctlz_zero_undef_v2i64_unmasked(<2 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v9, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vsrl.vi v9, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v9, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v9
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI61_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI61_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI61_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI61_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI61_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI61_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI61_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI61_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.ctlz.v2i64(<2 x i64> %va, i1 true, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v10, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vsrl.vi v10, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v10, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI62_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI62_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI62_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI62_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI62_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI62_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI62_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI62_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.ctlz.v4i64(<4 x i64> %va, i1 true, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }
 
 define <4 x i64> @vp_ctlz_zero_undef_v4i64_unmasked(<4 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v10, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vsrl.vi v10, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v10, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v10
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI63_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI63_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI63_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI63_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI63_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI63_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI63_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI63_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.ctlz.v4i64(<4 x i64> %va, i1 true, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v12, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vsrl.vi v12, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v12, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI64_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI64_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI64_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI64_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI64_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI64_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI64_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI64_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.ctlz.v8i64(<8 x i64> %va, i1 true, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }
 
 define <8 x i64> @vp_ctlz_zero_undef_v8i64_unmasked(<8 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v12, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vsrl.vi v12, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v12, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v12
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI65_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI65_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI65_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI65_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI65_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI65_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI65_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI65_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.ctlz.v8i64(<8 x i64> %va, i1 true, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI66_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI66_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI66_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI66_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI66_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI66_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI66_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI66_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <15 x i64> @llvm.vp.ctlz.v15i64(<15 x i64> %va, i1 true, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }
 
 define <15 x i64> @vp_ctlz_zero_undef_v15i64_unmasked(<15 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI67_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI67_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI67_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI67_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI67_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI67_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI67_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI67_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.ctlz.v15i64(<15 x i64> %va, i1 true, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI68_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI68_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI68_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI68_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI68_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI68_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI68_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI68_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <16 x i64> @llvm.vp.ctlz.v16i64(<16 x i64> %va, i1 true, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }
 
 define <16 x i64> @vp_ctlz_zero_undef_v16i64_unmasked(<16 x i64> %va, i32 zeroext %evl) {

 ; RV64-NEXT:    vsrl.vi v16, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vsrl.vi v16, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    li a0, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a0
 ; RV64-NEXT:    vor.vv v8, v8, v16
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI69_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI69_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI69_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI69_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI69_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI69_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI69_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI69_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.ctlz.v16i64(<16 x i64> %va, i1 true, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    li a1, 32
 ; RV64-NEXT:    vsrl.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
-; RV64-NEXT:    lui a2, %hi(.LCPI70_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI70_0)(a2)
-; RV64-NEXT:    lui a2, %hi(.LCPI70_1)
-; RV64-NEXT:    ld a2, %lo(.LCPI70_1)(a2)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
+; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a4, %hi(.LCPI70_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI70_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI70_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI70_3)(a5)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    addi a7, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a7) # Unknown-size Folded Spill
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1

 ; RV64-NEXT:    vsrl.vi v16, v8, 8, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 16, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vor.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
-; RV64-NEXT:    vand.vx v16, v16, a3, v0.t
+; RV64-NEXT:    vand.vx v16, v16, a2, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    vsrl.vx v16, v8, a6, v0.t
 ; RV64-NEXT:    addi a0, sp, 16
 ; RV64-NEXT:    vl8r.v v8, (a0) # Unknown-size Folded Reload

 ; RV64-NEXT:    vsrl.vi v24, v8, 8
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vsrl.vi v24, v8, 16
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    li a1, 32
 ; RV64-NEXT:    vsrl.vx v24, v8, a1
 ; RV64-NEXT:    vor.vv v8, v8, v24
 ; RV64-NEXT:    vnot.v v8, v8
-; RV64-NEXT:    lui a2, %hi(.LCPI71_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI71_0)(a2)
-; RV64-NEXT:    lui a3, %hi(.LCPI71_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI71_1)(a3)
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a2, 349525
+; RV64-NEXT:    addiw a2, a2, 1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    vand.vx v24, v24, a2
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v8, a3
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vadd.vv v8, v24, v8
-; RV64-NEXT:    lui a4, %hi(.LCPI71_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI71_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI71_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI71_3)(a5)
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1
 ; RV64-NEXT:    and a0, a0, a7
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctlz.ll

 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 8
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v9
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 16
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v9
 ; LMULMAX2-RV64I-NEXT:    li a1, 32
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v9, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v9
 ; LMULMAX2-RV64I-NEXT:    vnot.v v8, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI3_0)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI3_0)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI3_1)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI3_1)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 1
+; LMULMAX2-RV64I-NEXT:    lui a1, 349525
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v9, v9, a1
 ; LMULMAX2-RV64I-NEXT:    vsub.vv v8, v8, v9
-; LMULMAX2-RV64I-NEXT:    vand.vx v9, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 209715
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v9, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v9, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI3_2)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI3_2)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI3_3)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI3_3)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v8, v9
+; LMULMAX2-RV64I-NEXT:    lui a1, 61681
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 4112
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    li a1, 56
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    ret
 ;
 ; LMULMAX2-RV32F-LABEL: ctlz_v2i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma

 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 8
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v10
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 16
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v10
 ; LMULMAX2-RV64I-NEXT:    li a1, 32
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v10, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vor.vv v8, v8, v10
 ; LMULMAX2-RV64I-NEXT:    vnot.v v8, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI7_0)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI7_0)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI7_1)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI7_1)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 1
+; LMULMAX2-RV64I-NEXT:    lui a1, 349525
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64I-NEXT:    vsub.vv v8, v8, v10
-; LMULMAX2-RV64I-NEXT:    vand.vx v10, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 209715
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v10, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v10, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI7_2)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI7_2)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI7_3)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI7_3)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v8, v10
+; LMULMAX2-RV64I-NEXT:    lui a1, 61681
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 4112
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    li a1, 56
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    ret
 ;
 ; LMULMAX2-RV32F-LABEL: ctlz_v4i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e32, m1, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctpop-vp.ll

 ; RV32-NEXT:    vmul.vv v8, v8, v9, v0.t
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI24_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.ctpop.v2i64(<2 x i64> %va, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }
 
 define <2 x i64> @vp_ctpop_v2i64_unmasked(<2 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    vmul.vv v8, v8, v9
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI25_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI25_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.ctpop.v2i64(<2 x i64> %va, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ; RV32-NEXT:    vmul.vv v8, v8, v10, v0.t
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI26_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.ctpop.v4i64(<4 x i64> %va, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }
 
 define <4 x i64> @vp_ctpop_v4i64_unmasked(<4 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    vmul.vv v8, v8, v10
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI27_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI27_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.ctpop.v4i64(<4 x i64> %va, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ; RV32-NEXT:    vmul.vv v8, v8, v12, v0.t
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI28_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.ctpop.v8i64(<8 x i64> %va, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }
 
 define <8 x i64> @vp_ctpop_v8i64_unmasked(<8 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    vmul.vv v8, v8, v12
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI29_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI29_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.ctpop.v8i64(<8 x i64> %va, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ; RV32-NEXT:    vmul.vv v8, v8, v16, v0.t
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v15i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI30_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <15 x i64> @llvm.vp.ctpop.v15i64(<15 x i64> %va, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }
 
 define <15 x i64> @vp_ctpop_v15i64_unmasked(<15 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    vmul.vv v8, v8, v16
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v15i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI31_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI31_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.ctpop.v15i64(<15 x i64> %va, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }

 ; RV32-NEXT:    vmul.vv v8, v8, v16, v0.t
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI32_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <16 x i64> @llvm.vp.ctpop.v16i64(<16 x i64> %va, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }
 
 define <16 x i64> @vp_ctpop_v16i64_unmasked(<16 x i64> %va, i32 zeroext %evl) {

 ; RV32-NEXT:    vmul.vv v8, v8, v16
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsrl.vx v8, v8, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vp_ctpop_v16i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
-; RV64-NEXT:    lui a0, %hi(.LCPI33_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI33_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.ctpop.v16i64(<16 x i64> %va, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }

 ; RV64-NEXT:    li a2, 16
 ; RV64-NEXT:    vslidedown.vi v24, v0, 2
 ; RV64-NEXT:    mv a1, a0
 ; RV64-NEXT:    bltu a0, a2, .LBB34_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB34_2:
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
-; RV64-NEXT:    lui a1, %hi(.LCPI34_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI34_0)(a1)
-; RV64-NEXT:    lui a2, %hi(.LCPI34_1)
-; RV64-NEXT:    ld a2, %lo(.LCPI34_1)(a2)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a2, 209715
+; RV64-NEXT:    addiw a2, a2, 819
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    vand.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a2, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a3, %hi(.LCPI34_2)
-; RV64-NEXT:    ld a3, %lo(.LCPI34_2)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI34_3)
-; RV64-NEXT:    ld a4, %lo(.LCPI34_3)(a4)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a3, 61681
+; RV64-NEXT:    addiw a3, a3, -241
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
+; RV64-NEXT:    lui a4, 4112
+; RV64-NEXT:    addiw a4, a4, 257
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vmul.vx v8, v8, a4, v0.t
 ; RV64-NEXT:    li a5, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    addi a6, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a6) # Unknown-size Folded Spill
 ; RV64-NEXT:    addi a6, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a6
 ; RV64-NEXT:    addi a0, a0, -1

 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a2, 16
 ; RV64-NEXT:    mv a1, a0
 ; RV64-NEXT:    bltu a0, a2, .LBB35_2
 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB35_2:
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
-; RV64-NEXT:    lui a1, %hi(.LCPI35_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI35_0)(a1)
-; RV64-NEXT:    lui a2, %hi(.LCPI35_1)
-; RV64-NEXT:    ld a2, %lo(.LCPI35_1)(a2)
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vand.vx v24, v24, a1
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a2, 209715
+; RV64-NEXT:    addiw a2, a2, 819
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    vand.vx v24, v8, a2
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a2
 ; RV64-NEXT:    vadd.vv v8, v24, v8
-; RV64-NEXT:    lui a3, %hi(.LCPI35_2)
-; RV64-NEXT:    ld a3, %lo(.LCPI35_2)(a3)
-; RV64-NEXT:    lui a4, %hi(.LCPI35_3)
-; RV64-NEXT:    ld a4, %lo(.LCPI35_3)(a4)
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 61681
+; RV64-NEXT:    addiw a3, a3, -241
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v8, v8, a3
+; RV64-NEXT:    lui a4, 4112
+; RV64-NEXT:    addiw a4, a4, 257
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vmul.vx v8, v8, a4
 ; RV64-NEXT:    li a5, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a5
 ; RV64-NEXT:    addi a6, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a6
 ; RV64-NEXT:    addi a0, a0, -1
 ; RV64-NEXT:    and a0, a0, a6
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctpop.ll

 ; LMULMAX2-RV32-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32-NEXT:    ret
 ;
 ; LMULMAX2-RV64-LABEL: ctpop_v2i64:
 ; LMULMAX2-RV64:       # %bb.0:
 ; LMULMAX2-RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX2-RV64-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI3_0)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI3_0)(a1)
-; LMULMAX2-RV64-NEXT:    lui a2, %hi(.LCPI3_1)
-; LMULMAX2-RV64-NEXT:    ld a2, %lo(.LCPI3_1)(a2)
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v9, v8, 1
+; LMULMAX2-RV64-NEXT:    lui a1, 349525
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v9, v9, a1
 ; LMULMAX2-RV64-NEXT:    vsub.vv v8, v8, v9
-; LMULMAX2-RV64-NEXT:    vand.vx v9, v8, a2
+; LMULMAX2-RV64-NEXT:    lui a1, 209715
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64-NEXT:    vand.vx v9, v8, a1
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    vadd.vv v8, v9, v8
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI3_2)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI3_2)(a1)
-; LMULMAX2-RV64-NEXT:    lui a2, %hi(.LCPI3_3)
-; LMULMAX2-RV64-NEXT:    ld a2, %lo(.LCPI3_3)(a2)
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v9, v8, 4
 ; LMULMAX2-RV64-NEXT:    vadd.vv v8, v8, v9
+; LMULMAX2-RV64-NEXT:    lui a1, 61681
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64-NEXT:    lui a1, 4112
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    li a1, 56
 ; LMULMAX2-RV64-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64-NEXT:    ret
 ;
 ; LMULMAX1-RV32-LABEL: ctpop_v2i64:
 ; LMULMAX1-RV32:       # %bb.0:
 ; LMULMAX1-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma

 ; LMULMAX1-RV32-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX1-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX1-RV32-NEXT:    ret
 ;
 ; LMULMAX1-RV64-LABEL: ctpop_v2i64:
 ; LMULMAX1-RV64:       # %bb.0:
 ; LMULMAX1-RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX1-RV64-NEXT:    vle64.v v8, (a0)
-; LMULMAX1-RV64-NEXT:    lui a1, %hi(.LCPI3_0)
-; LMULMAX1-RV64-NEXT:    ld a1, %lo(.LCPI3_0)(a1)
-; LMULMAX1-RV64-NEXT:    lui a2, %hi(.LCPI3_1)
-; LMULMAX1-RV64-NEXT:    ld a2, %lo(.LCPI3_1)(a2)
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v9, v8, 1
+; LMULMAX1-RV64-NEXT:    lui a1, 349525
+; LMULMAX1-RV64-NEXT:    addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX1-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a1
 ; LMULMAX1-RV64-NEXT:    vsub.vv v8, v8, v9
-; LMULMAX1-RV64-NEXT:    vand.vx v9, v8, a2
+; LMULMAX1-RV64-NEXT:    lui a1, 209715
+; LMULMAX1-RV64-NEXT:    addiw a1, a1, 819
+; LMULMAX1-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX1-RV64-NEXT:    add a1, a1, a2
+; LMULMAX1-RV64-NEXT:    vand.vx v9, v8, a1
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX1-RV64-NEXT:    vand.vx v8, v8, a2
+; LMULMAX1-RV64-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX1-RV64-NEXT:    vadd.vv v8, v9, v8
-; LMULMAX1-RV64-NEXT:    lui a1, %hi(.LCPI3_2)
-; LMULMAX1-RV64-NEXT:    ld a1, %lo(.LCPI3_2)(a1)
-; LMULMAX1-RV64-NEXT:    lui a2, %hi(.LCPI3_3)
-; LMULMAX1-RV64-NEXT:    ld a2, %lo(.LCPI3_3)(a2)
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v9, v8, 4
 ; LMULMAX1-RV64-NEXT:    vadd.vv v8, v8, v9
+; LMULMAX1-RV64-NEXT:    lui a1, 61681
+; LMULMAX1-RV64-NEXT:    addiw a1, a1, -241
+; LMULMAX1-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX1-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX1-RV64-NEXT:    vand.vx v8, v8, a1
-; LMULMAX1-RV64-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX1-RV64-NEXT:    lui a1, 4112
+; LMULMAX1-RV64-NEXT:    addiw a1, a1, 257
+; LMULMAX1-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX1-RV64-NEXT:    add a1, a1, a2
+; LMULMAX1-RV64-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX1-RV64-NEXT:    li a1, 56
 ; LMULMAX1-RV64-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX1-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX1-RV64-NEXT:    ret
   %a = load <2 x i64>, ptr %x
   %b = load <2 x i64>, ptr %y
   %c = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %a)
   store <2 x i64> %c, ptr %x

 ; LMULMAX2-RV32-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32-NEXT:    ret
 ;
 ; LMULMAX2-RV64-LABEL: ctpop_v4i64:
 ; LMULMAX2-RV64:       # %bb.0:
 ; LMULMAX2-RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX2-RV64-NEXT:    vle64.v v8, (a0)
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI7_0)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI7_0)(a1)
-; LMULMAX2-RV64-NEXT:    lui a2, %hi(.LCPI7_1)
-; LMULMAX2-RV64-NEXT:    ld a2, %lo(.LCPI7_1)(a2)
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v10, v8, 1
+; LMULMAX2-RV64-NEXT:    lui a1, 349525
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64-NEXT:    vsub.vv v8, v8, v10
-; LMULMAX2-RV64-NEXT:    vand.vx v10, v8, a2
+; LMULMAX2-RV64-NEXT:    lui a1, 209715
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64-NEXT:    vand.vx v10, v8, a1
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    vadd.vv v8, v10, v8
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI7_2)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI7_2)(a1)
-; LMULMAX2-RV64-NEXT:    lui a2, %hi(.LCPI7_3)
-; LMULMAX2-RV64-NEXT:    ld a2, %lo(.LCPI7_3)(a2)
 ; LMULMAX2-RV64-NEXT:    vsrl.vi v10, v8, 4
 ; LMULMAX2-RV64-NEXT:    vadd.vv v8, v8, v10
+; LMULMAX2-RV64-NEXT:    lui a1, 61681
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64-NEXT:    lui a1, 4112
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    li a1, 56
 ; LMULMAX2-RV64-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64-NEXT:    ret
 ;
 ; LMULMAX1-RV32-LABEL: ctpop_v4i64:
 ; LMULMAX1-RV32:       # %bb.0:
 ; LMULMAX1-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma

 ; LMULMAX1-RV32-NEXT:    vsrl.vx v9, v9, a2
 ; LMULMAX1-RV32-NEXT:    vse64.v v9, (a0)
 ; LMULMAX1-RV32-NEXT:    vse64.v v8, (a1)
 ; LMULMAX1-RV32-NEXT:    ret
 ;
 ; LMULMAX1-RV64-LABEL: ctpop_v4i64:
 ; LMULMAX1-RV64:       # %bb.0:
 ; LMULMAX1-RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; LMULMAX1-RV64-NEXT:    vle64.v v8, (a0)
 ; LMULMAX1-RV64-NEXT:    addi a1, a0, 16
-; LMULMAX1-RV64-NEXT:    vle64.v v9, (a1)
-; LMULMAX1-RV64-NEXT:    lui a2, %hi(.LCPI7_0)
-; LMULMAX1-RV64-NEXT:    ld a2, %lo(.LCPI7_0)(a2)
-; LMULMAX1-RV64-NEXT:    lui a3, %hi(.LCPI7_1)
-; LMULMAX1-RV64-NEXT:    ld a3, %lo(.LCPI7_1)(a3)
-; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 1
-; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, a2
-; LMULMAX1-RV64-NEXT:    vsub.vv v9, v9, v10
-; LMULMAX1-RV64-NEXT:    vand.vx v10, v9, a3
-; LMULMAX1-RV64-NEXT:    vsrl.vi v9, v9, 2
-; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a3
-; LMULMAX1-RV64-NEXT:    vadd.vv v9, v10, v9
-; LMULMAX1-RV64-NEXT:    lui a4, %hi(.LCPI7_2)
-; LMULMAX1-RV64-NEXT:    ld a4, %lo(.LCPI7_2)(a4)
-; LMULMAX1-RV64-NEXT:    lui a5, %hi(.LCPI7_3)
-; LMULMAX1-RV64-NEXT:    ld a5, %lo(.LCPI7_3)(a5)
-; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 4
-; LMULMAX1-RV64-NEXT:    vadd.vv v9, v9, v10
-; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a4
-; LMULMAX1-RV64-NEXT:    vmul.vx v9, v9, a5
-; LMULMAX1-RV64-NEXT:    li a6, 56
-; LMULMAX1-RV64-NEXT:    vsrl.vx v9, v9, a6
+; LMULMAX1-RV64-NEXT:    vle64.v v8, (a1)
+; LMULMAX1-RV64-NEXT:    vle64.v v9, (a0)
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v8, 1
+; LMULMAX1-RV64-NEXT:    lui a2, 349525
+; LMULMAX1-RV64-NEXT:    addiw a2, a2, 1365
+; LMULMAX1-RV64-NEXT:    slli a3, a2, 32
+; LMULMAX1-RV64-NEXT:    add a2, a2, a3
 ; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, a2
 ; LMULMAX1-RV64-NEXT:    vsub.vv v8, v8, v10
+; LMULMAX1-RV64-NEXT:    lui a3, 209715
+; LMULMAX1-RV64-NEXT:    addiw a3, a3, 819
+; LMULMAX1-RV64-NEXT:    slli a4, a3, 32
+; LMULMAX1-RV64-NEXT:    add a3, a3, a4
 ; LMULMAX1-RV64-NEXT:    vand.vx v10, v8, a3
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v8, v8, 2
 ; LMULMAX1-RV64-NEXT:    vand.vx v8, v8, a3
 ; LMULMAX1-RV64-NEXT:    vadd.vv v8, v10, v8
 ; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v8, 4
 ; LMULMAX1-RV64-NEXT:    vadd.vv v8, v8, v10
+; LMULMAX1-RV64-NEXT:    lui a4, 61681
+; LMULMAX1-RV64-NEXT:    addiw a4, a4, -241
+; LMULMAX1-RV64-NEXT:    slli a5, a4, 32
+; LMULMAX1-RV64-NEXT:    add a4, a4, a5
 ; LMULMAX1-RV64-NEXT:    vand.vx v8, v8, a4
+; LMULMAX1-RV64-NEXT:    lui a5, 4112
+; LMULMAX1-RV64-NEXT:    addiw a5, a5, 257
+; LMULMAX1-RV64-NEXT:    slli a6, a5, 32
+; LMULMAX1-RV64-NEXT:    add a5, a5, a6
 ; LMULMAX1-RV64-NEXT:    vmul.vx v8, v8, a5
+; LMULMAX1-RV64-NEXT:    li a6, 56
 ; LMULMAX1-RV64-NEXT:    vsrl.vx v8, v8, a6
-; LMULMAX1-RV64-NEXT:    vse64.v v8, (a0)
-; LMULMAX1-RV64-NEXT:    vse64.v v9, (a1)
+; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 1
+; LMULMAX1-RV64-NEXT:    vand.vx v10, v10, a2
+; LMULMAX1-RV64-NEXT:    vsub.vv v9, v9, v10
+; LMULMAX1-RV64-NEXT:    vand.vx v10, v9, a3
+; LMULMAX1-RV64-NEXT:    vsrl.vi v9, v9, 2
+; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a3
+; LMULMAX1-RV64-NEXT:    vadd.vv v9, v10, v9
+; LMULMAX1-RV64-NEXT:    vsrl.vi v10, v9, 4
+; LMULMAX1-RV64-NEXT:    vadd.vv v9, v9, v10
+; LMULMAX1-RV64-NEXT:    vand.vx v9, v9, a4
+; LMULMAX1-RV64-NEXT:    vmul.vx v9, v9, a5
+; LMULMAX1-RV64-NEXT:    vsrl.vx v9, v9, a6
+; LMULMAX1-RV64-NEXT:    vse64.v v9, (a0)
+; LMULMAX1-RV64-NEXT:    vse64.v v8, (a1)
 ; LMULMAX1-RV64-NEXT:    ret
   %a = load <4 x i64>, ptr %x
   %b = load <4 x i64>, ptr %y
   %c = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> %a)
   store <4 x i64> %c, ptr %x
   ret void
 }
 declare <4 x i64> @llvm.ctpop.v4i64(<4 x i64>)

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz-vp.ll

 ;
 ; RV64-LABEL: vp_cttz_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v9, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI24_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI24_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI24_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI24_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.cttz.v2i64(<2 x i64> %va, i1 false, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }
 
 define <2 x i64> @vp_cttz_v2i64_unmasked(<2 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_v2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v9
-; RV64-NEXT:    lui a0, %hi(.LCPI25_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI25_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI25_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI25_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI25_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.cttz.v2i64(<2 x i64> %va, i1 false, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v10, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI26_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI26_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI26_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI26_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.cttz.v4i64(<4 x i64> %va, i1 false, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }
 
 define <4 x i64> @vp_cttz_v4i64_unmasked(<4 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_v4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v10
-; RV64-NEXT:    lui a0, %hi(.LCPI27_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI27_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI27_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI27_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI27_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.cttz.v4i64(<4 x i64> %va, i1 false, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v12, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI28_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI28_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI28_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI28_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.cttz.v8i64(<8 x i64> %va, i1 false, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }
 
 define <8 x i64> @vp_cttz_v8i64_unmasked(<8 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_v8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v12
-; RV64-NEXT:    lui a0, %hi(.LCPI29_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI29_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI29_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI29_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI29_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.cttz.v8i64(<8 x i64> %va, i1 false, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_v15i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI30_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI30_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI30_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI30_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <15 x i64> @llvm.vp.cttz.v15i64(<15 x i64> %va, i1 false, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }
 
 define <15 x i64> @vp_cttz_v15i64_unmasked(<15 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_v15i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI31_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI31_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI31_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI31_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI31_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.cttz.v15i64(<15 x i64> %va, i1 false, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI32_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI32_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI32_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI32_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <16 x i64> @llvm.vp.cttz.v16i64(<16 x i64> %va, i1 false, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }
 
 define <16 x i64> @vp_cttz_v16i64_unmasked(<16 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_v16i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI33_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI33_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI33_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI33_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI33_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.cttz.v16i64(<16 x i64> %va, i1 false, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }

 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB34_2:
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a1, %hi(.LCPI34_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI34_0)(a1)
-; RV64-NEXT:    lui a3, %hi(.LCPI34_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI34_1)(a3)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a3, a1, 32
+; RV64-NEXT:    add a1, a1, a3
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a4, %hi(.LCPI34_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI34_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI34_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI34_3)(a5)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    addi a7, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a7) # Unknown-size Folded Spill
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1

 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB35_2:
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v24, v8, a2
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v24
-; RV64-NEXT:    lui a1, %hi(.LCPI35_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI35_0)(a1)
-; RV64-NEXT:    lui a3, %hi(.LCPI35_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI35_1)(a3)
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a3, a1, 32
+; RV64-NEXT:    add a1, a1, a3
 ; RV64-NEXT:    vand.vx v24, v24, a1
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v8, a3
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vadd.vv v8, v24, v8
-; RV64-NEXT:    lui a4, %hi(.LCPI35_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI35_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI35_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI35_3)(a5)
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1
 ; RV64-NEXT:    and a0, a0, a7
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v9, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI60_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI60_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI60_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI60_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v9, v0.t
-; RV64-NEXT:    vand.vx v9, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v9, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI60_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI60_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI60_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI60_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v9, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <2 x i64> @llvm.vp.cttz.v2i64(<2 x i64> %va, i1 true, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }
 
 define <2 x i64> @vp_cttz_zero_undef_v2i64_unmasked(<2 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v2i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
 ; RV64-NEXT:    vsub.vx v9, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v9
-; RV64-NEXT:    lui a0, %hi(.LCPI61_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI61_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI61_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI61_1)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v9, v9, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v9
-; RV64-NEXT:    vand.vx v9, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v9, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v9, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI61_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI61_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI61_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI61_3)(a1)
 ; RV64-NEXT:    vsrl.vi v9, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v9
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <2 x i1> poison, i1 true, i32 0
   %m = shufflevector <2 x i1> %head, <2 x i1> poison, <2 x i32> zeroinitializer
   %v = call <2 x i64> @llvm.vp.cttz.v2i64(<2 x i64> %va, i1 true, <2 x i1> %m, i32 %evl)
   ret <2 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v10, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI62_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI62_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI62_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI62_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v10, v0.t
-; RV64-NEXT:    vand.vx v10, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v10, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI62_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI62_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI62_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI62_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v10, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <4 x i64> @llvm.vp.cttz.v4i64(<4 x i64> %va, i1 true, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }
 
 define <4 x i64> @vp_cttz_zero_undef_v4i64_unmasked(<4 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v4i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
 ; RV64-NEXT:    vsub.vx v10, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v10
-; RV64-NEXT:    lui a0, %hi(.LCPI63_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI63_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI63_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI63_1)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v10, v10, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v10
-; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v10, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v10, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI63_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI63_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI63_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI63_3)(a1)
 ; RV64-NEXT:    vsrl.vi v10, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v10
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <4 x i1> poison, i1 true, i32 0
   %m = shufflevector <4 x i1> %head, <4 x i1> poison, <4 x i32> zeroinitializer
   %v = call <4 x i64> @llvm.vp.cttz.v4i64(<4 x i64> %va, i1 true, <4 x i1> %m, i32 %evl)
   ret <4 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v12, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI64_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI64_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI64_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI64_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v12, v0.t
-; RV64-NEXT:    vand.vx v12, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v12, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI64_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI64_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI64_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI64_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v12, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <8 x i64> @llvm.vp.cttz.v8i64(<8 x i64> %va, i1 true, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }
 
 define <8 x i64> @vp_cttz_zero_undef_v8i64_unmasked(<8 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v8i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
 ; RV64-NEXT:    vsub.vx v12, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v12
-; RV64-NEXT:    lui a0, %hi(.LCPI65_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI65_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI65_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI65_1)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v12, v12, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v12
-; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v12, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v12, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI65_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI65_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI65_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI65_3)(a1)
 ; RV64-NEXT:    vsrl.vi v12, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v12
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <8 x i1> poison, i1 true, i32 0
   %m = shufflevector <8 x i1> %head, <8 x i1> poison, <8 x i32> zeroinitializer
   %v = call <8 x i64> @llvm.vp.cttz.v8i64(<8 x i64> %va, i1 true, <8 x i1> %m, i32 %evl)
   ret <8 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v15i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI66_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI66_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI66_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI66_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI66_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI66_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI66_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI66_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <15 x i64> @llvm.vp.cttz.v15i64(<15 x i64> %va, i1 true, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }
 
 define <15 x i64> @vp_cttz_zero_undef_v15i64_unmasked(<15 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v15i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI67_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI67_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI67_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI67_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI67_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI67_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI67_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI67_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <15 x i1> poison, i1 true, i32 0
   %m = shufflevector <15 x i1> %head, <15 x i1> poison, <15 x i32> zeroinitializer
   %v = call <15 x i64> @llvm.vp.cttz.v15i64(<15 x i64> %va, i1 true, <15 x i1> %m, i32 %evl)
   ret <15 x i64> %v
 }

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI68_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI68_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI68_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI68_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
-; RV64-NEXT:    vand.vx v16, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
-; RV64-NEXT:    vand.vx v8, v8, a1, v0.t
+; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a0, %hi(.LCPI68_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI68_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI68_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI68_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0, v0.t
-; RV64-NEXT:    vmul.vx v8, v8, a1, v0.t
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0, v0.t
 ; RV64-NEXT:    ret
   %v = call <16 x i64> @llvm.vp.cttz.v16i64(<16 x i64> %va, i1 true, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }
 
 define <16 x i64> @vp_cttz_zero_undef_v16i64_unmasked(<16 x i64> %va, i32 zeroext %evl) {

 ;
 ; RV64-LABEL: vp_cttz_zero_undef_v16i64_unmasked:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 1
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a1
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    lui a0, %hi(.LCPI69_0)
-; RV64-NEXT:    ld a0, %lo(.LCPI69_0)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI69_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI69_1)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1
+; RV64-NEXT:    lui a0, 349525
+; RV64-NEXT:    addiw a0, a0, 1365
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v16, v16, a0
 ; RV64-NEXT:    vsub.vv v8, v8, v16
-; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    lui a0, 209715
+; RV64-NEXT:    addiw a0, a0, 819
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vand.vx v16, v8, a0
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
-; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vand.vx v8, v8, a0
 ; RV64-NEXT:    vadd.vv v8, v16, v8
-; RV64-NEXT:    lui a0, %hi(.LCPI69_2)
-; RV64-NEXT:    ld a0, %lo(.LCPI69_2)(a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI69_3)
-; RV64-NEXT:    ld a1, %lo(.LCPI69_3)(a1)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v16
+; RV64-NEXT:    lui a0, 61681
+; RV64-NEXT:    addiw a0, a0, -241
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    vand.vx v8, v8, a0
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 4112
+; RV64-NEXT:    addiw a0, a0, 257
+; RV64-NEXT:    slli a1, a0, 32
+; RV64-NEXT:    add a0, a0, a1
+; RV64-NEXT:    vmul.vx v8, v8, a0
 ; RV64-NEXT:    li a0, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a0
 ; RV64-NEXT:    ret
   %head = insertelement <16 x i1> poison, i1 true, i32 0
   %m = shufflevector <16 x i1> %head, <16 x i1> poison, <16 x i32> zeroinitializer
   %v = call <16 x i64> @llvm.vp.cttz.v16i64(<16 x i64> %va, i1 true, <16 x i1> %m, i32 %evl)
   ret <16 x i64> %v
 }

 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB70_2:
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v16, v8, a2, v0.t
 ; RV64-NEXT:    vnot.v v8, v8, v0.t
 ; RV64-NEXT:    vand.vv v8, v8, v16, v0.t
-; RV64-NEXT:    lui a1, %hi(.LCPI70_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI70_0)(a1)
-; RV64-NEXT:    lui a3, %hi(.LCPI70_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI70_1)(a3)
 ; RV64-NEXT:    vsrl.vi v16, v8, 1, v0.t
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a3, a1, 32
+; RV64-NEXT:    add a1, a1, a3
 ; RV64-NEXT:    vand.vx v16, v16, a1, v0.t
 ; RV64-NEXT:    vsub.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v16, v8, a3, v0.t
 ; RV64-NEXT:    vsrl.vi v8, v8, 2, v0.t
 ; RV64-NEXT:    vand.vx v8, v8, a3, v0.t
 ; RV64-NEXT:    vadd.vv v8, v16, v8, v0.t
-; RV64-NEXT:    lui a4, %hi(.LCPI70_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI70_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI70_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI70_3)(a5)
 ; RV64-NEXT:    vsrl.vi v16, v8, 4, v0.t
 ; RV64-NEXT:    vadd.vv v8, v8, v16, v0.t
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4, v0.t
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5, v0.t
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6, v0.t
 ; RV64-NEXT:    addi a7, sp, 16
 ; RV64-NEXT:    vs8r.v v8, (a7) # Unknown-size Folded Spill
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1

 ; RV64-NEXT:  # %bb.1:
 ; RV64-NEXT:    li a1, 16
 ; RV64-NEXT:  .LBB71_2:
 ; RV64-NEXT:    li a2, 1
 ; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
 ; RV64-NEXT:    vsub.vx v24, v8, a2
 ; RV64-NEXT:    vnot.v v8, v8
 ; RV64-NEXT:    vand.vv v8, v8, v24
-; RV64-NEXT:    lui a1, %hi(.LCPI71_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI71_0)(a1)
-; RV64-NEXT:    lui a3, %hi(.LCPI71_1)
-; RV64-NEXT:    ld a3, %lo(.LCPI71_1)(a3)
 ; RV64-NEXT:    vsrl.vi v24, v8, 1
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a3, a1, 32
+; RV64-NEXT:    add a1, a1, a3
 ; RV64-NEXT:    vand.vx v24, v24, a1
 ; RV64-NEXT:    vsub.vv v8, v8, v24
+; RV64-NEXT:    lui a3, 209715
+; RV64-NEXT:    addiw a3, a3, 819
+; RV64-NEXT:    slli a4, a3, 32
+; RV64-NEXT:    add a3, a3, a4
 ; RV64-NEXT:    vand.vx v24, v8, a3
 ; RV64-NEXT:    vsrl.vi v8, v8, 2
 ; RV64-NEXT:    vand.vx v8, v8, a3
 ; RV64-NEXT:    vadd.vv v8, v24, v8
-; RV64-NEXT:    lui a4, %hi(.LCPI71_2)
-; RV64-NEXT:    ld a4, %lo(.LCPI71_2)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI71_3)
-; RV64-NEXT:    ld a5, %lo(.LCPI71_3)(a5)
 ; RV64-NEXT:    vsrl.vi v24, v8, 4
 ; RV64-NEXT:    vadd.vv v8, v8, v24
+; RV64-NEXT:    lui a4, 61681
+; RV64-NEXT:    addiw a4, a4, -241
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    vand.vx v8, v8, a4
+; RV64-NEXT:    lui a5, 4112
+; RV64-NEXT:    addiw a5, a5, 257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    vmul.vx v8, v8, a5
 ; RV64-NEXT:    li a6, 56
 ; RV64-NEXT:    vsrl.vx v8, v8, a6
 ; RV64-NEXT:    addi a7, a0, -16
 ; RV64-NEXT:    sltu a0, a0, a7
 ; RV64-NEXT:    addi a0, a0, -1
 ; RV64-NEXT:    and a0, a0, a7
 ; RV64-NEXT:    vsetvli zero, a0, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-cttz.ll

 ; LMULMAX2-RV64I-LABEL: cttz_v2i64:
 ; LMULMAX2-RV64I:       # %bb.0:
 ; LMULMAX2-RV64I-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX2-RV64I-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    li a1, 1
 ; LMULMAX2-RV64I-NEXT:    vsub.vx v9, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vnot.v v8, v8
 ; LMULMAX2-RV64I-NEXT:    vand.vv v8, v8, v9
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI3_0)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI3_0)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI3_1)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI3_1)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 1
+; LMULMAX2-RV64I-NEXT:    lui a1, 349525
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v9, v9, a1
 ; LMULMAX2-RV64I-NEXT:    vsub.vv v8, v8, v9
-; LMULMAX2-RV64I-NEXT:    vand.vx v9, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 209715
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v9, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v9, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI3_2)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI3_2)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI3_3)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI3_3)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v9, v8, 4
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v8, v9
+; LMULMAX2-RV64I-NEXT:    lui a1, 61681
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 4112
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    li a1, 56
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    ret
 ;
 ; LMULMAX2-RV32F-LABEL: cttz_v2i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 2, e64, m1, ta, ma

 ; LMULMAX2-RV64I-LABEL: cttz_v4i64:
 ; LMULMAX2-RV64I:       # %bb.0:
 ; LMULMAX2-RV64I-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
 ; LMULMAX2-RV64I-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    li a1, 1
 ; LMULMAX2-RV64I-NEXT:    vsub.vx v10, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vnot.v v8, v8
 ; LMULMAX2-RV64I-NEXT:    vand.vv v8, v8, v10
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI7_0)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI7_0)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI7_1)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI7_1)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 1
+; LMULMAX2-RV64I-NEXT:    lui a1, 349525
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v10, v10, a1
 ; LMULMAX2-RV64I-NEXT:    vsub.vv v8, v8, v10
-; LMULMAX2-RV64I-NEXT:    vand.vx v10, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 209715
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 819
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v10, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v8, v8, 2
-; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v10, v8
-; LMULMAX2-RV64I-NEXT:    lui a1, %hi(.LCPI7_2)
-; LMULMAX2-RV64I-NEXT:    ld a1, %lo(.LCPI7_2)(a1)
-; LMULMAX2-RV64I-NEXT:    lui a2, %hi(.LCPI7_3)
-; LMULMAX2-RV64I-NEXT:    ld a2, %lo(.LCPI7_3)(a2)
 ; LMULMAX2-RV64I-NEXT:    vsrl.vi v10, v8, 4
 ; LMULMAX2-RV64I-NEXT:    vadd.vv v8, v8, v10
+; LMULMAX2-RV64I-NEXT:    lui a1, 61681
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, -241
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
 ; LMULMAX2-RV64I-NEXT:    vand.vx v8, v8, a1
-; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a2
+; LMULMAX2-RV64I-NEXT:    lui a1, 4112
+; LMULMAX2-RV64I-NEXT:    addiw a1, a1, 257
+; LMULMAX2-RV64I-NEXT:    slli a2, a1, 32
+; LMULMAX2-RV64I-NEXT:    add a1, a1, a2
+; LMULMAX2-RV64I-NEXT:    vmul.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    li a1, 56
 ; LMULMAX2-RV64I-NEXT:    vsrl.vx v8, v8, a1
 ; LMULMAX2-RV64I-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64I-NEXT:    ret
 ;
 ; LMULMAX2-RV32F-LABEL: cttz_v4i64:
 ; LMULMAX2-RV32F:       # %bb.0:
 ; LMULMAX2-RV32F-NEXT:    vsetivli zero, 4, e64, m2, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-int.ll

 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vv v8, v8, v9
 ; RV32-NEXT:    vse64.v v8, (a0)
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: mulhu_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; RV64-NEXT:    lui a1, %hi(.LCPI69_0)
-; RV64-NEXT:    addi a1, a1, %lo(.LCPI69_0)
-; RV64-NEXT:    vlse64.v v8, (a1), zero
-; RV64-NEXT:    lui a1, %hi(.LCPI69_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI69_1)(a1)
-; RV64-NEXT:    vle64.v v9, (a0)
+; RV64-NEXT:    vle64.v v8, (a0)
+; RV64-NEXT:    lui a1, 838861
+; RV64-NEXT:    addiw a1, a1, -819
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    vmv.v.x v9, a1
+; RV64-NEXT:    lui a1, 699051
+; RV64-NEXT:    addiw a1, a1, -1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
-; RV64-NEXT:    vmv.s.x v8, a1
+; RV64-NEXT:    vmv.s.x v9, a1
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
-; RV64-NEXT:    vmulhu.vv v8, v9, v8
+; RV64-NEXT:    vmulhu.vv v8, v8, v9
 ; RV64-NEXT:    vid.v v9
 ; RV64-NEXT:    vadd.vi v9, v9, 1
 ; RV64-NEXT:    vsrl.vv v8, v8, v9
 ; RV64-NEXT:    vse64.v v8, (a0)
 ; RV64-NEXT:    ret
   %a = load <2 x i64>, ptr %x
   %b = udiv <2 x i64> %a, <i64 3, i64 5>
   store <2 x i64> %b, ptr %x

 ; RV32-NEXT:    vsrl.vx v9, v10, a1
 ; RV32-NEXT:    vadd.vv v8, v8, v9
 ; RV32-NEXT:    vse64.v v8, (a0)
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: mulhs_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; RV64-NEXT:    lui a1, %hi(.LCPI74_0)
-; RV64-NEXT:    addi a1, a1, %lo(.LCPI74_0)
-; RV64-NEXT:    vlse64.v v8, (a1), zero
-; RV64-NEXT:    lui a1, %hi(.LCPI74_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI74_1)(a1)
-; RV64-NEXT:    vle64.v v9, (a0)
+; RV64-NEXT:    lui a1, 349525
+; RV64-NEXT:    addiw a1, a1, 1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    lui a2, %hi(.LCPI74_0)
+; RV64-NEXT:    ld a2, %lo(.LCPI74_0)(a2)
+; RV64-NEXT:    vle64.v v8, (a0)
+; RV64-NEXT:    vmv.v.x v9, a1
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
-; RV64-NEXT:    vmv.s.x v8, a1
+; RV64-NEXT:    vmv.s.x v9, a2
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
-; RV64-NEXT:    vmulh.vv v8, v9, v8
+; RV64-NEXT:    vmulh.vv v9, v8, v9
 ; RV64-NEXT:    vid.v v10
 ; RV64-NEXT:    vrsub.vi v11, v10, 0
-; RV64-NEXT:    vmadd.vv v11, v9, v8
+; RV64-NEXT:    vmadd.vv v11, v8, v9
 ; RV64-NEXT:    li a1, 63
 ; RV64-NEXT:    vsrl.vx v8, v11, a1
 ; RV64-NEXT:    vsra.vv v9, v11, v10
 ; RV64-NEXT:    vadd.vv v8, v9, v8
 ; RV64-NEXT:    vse64.v v8, (a0)
 ; RV64-NEXT:    ret
   %a = load <2 x i64>, ptr %x
   %b = sdiv <2 x i64> %a, <i64 3, i64 -3>

 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; LMULMAX1-RV64-NEXT:    vmv.s.x v11, a2
 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; LMULMAX1-RV64-NEXT:    vmulhu.vv v11, v9, v11
 ; LMULMAX1-RV64-NEXT:    vsub.vv v9, v9, v11
 ; LMULMAX1-RV64-NEXT:    vmulhu.vv v9, v9, v10
 ; LMULMAX1-RV64-NEXT:    vadd.vv v9, v9, v11
 ; LMULMAX1-RV64-NEXT:    vid.v v10
-; LMULMAX1-RV64-NEXT:    lui a2, %hi(.LCPI184_2)
-; LMULMAX1-RV64-NEXT:    addi a2, a2, %lo(.LCPI184_2)
-; LMULMAX1-RV64-NEXT:    vlse64.v v11, (a2), zero
-; LMULMAX1-RV64-NEXT:    lui a2, %hi(.LCPI184_3)
-; LMULMAX1-RV64-NEXT:    ld a2, %lo(.LCPI184_3)(a2)
-; LMULMAX1-RV64-NEXT:    vadd.vi v12, v10, 2
-; LMULMAX1-RV64-NEXT:    vsrl.vv v9, v9, v12
+; LMULMAX1-RV64-NEXT:    vadd.vi v11, v10, 2
+; LMULMAX1-RV64-NEXT:    vsrl.vv v9, v9, v11
+; LMULMAX1-RV64-NEXT:    lui a2, 838861
+; LMULMAX1-RV64-NEXT:    addiw a2, a2, -819
+; LMULMAX1-RV64-NEXT:    slli a3, a2, 32
+; LMULMAX1-RV64-NEXT:    add a2, a2, a3
+; LMULMAX1-RV64-NEXT:    vmv.v.x v11, a2
+; LMULMAX1-RV64-NEXT:    lui a2, 699051
+; LMULMAX1-RV64-NEXT:    addiw a2, a2, -1365
+; LMULMAX1-RV64-NEXT:    slli a3, a2, 32
+; LMULMAX1-RV64-NEXT:    add a2, a2, a3
 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; LMULMAX1-RV64-NEXT:    vmv.s.x v11, a2
 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; LMULMAX1-RV64-NEXT:    vmulhu.vv v8, v8, v11
 ; LMULMAX1-RV64-NEXT:    vadd.vi v10, v10, 1
 ; LMULMAX1-RV64-NEXT:    vsrl.vv v8, v8, v10
 ; LMULMAX1-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX1-RV64-NEXT:    vse64.v v9, (a1)

 ; LMULMAX2-RV32-NEXT:    vsra.vv v10, v12, v10
 ; LMULMAX2-RV32-NEXT:    vadd.vv v8, v10, v8
 ; LMULMAX2-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV32-NEXT:    ret
 ;
 ; LMULMAX2-RV64-LABEL: mulhs_v4i64:
 ; LMULMAX2-RV64:       # %bb.0:
 ; LMULMAX2-RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; LMULMAX2-RV64-NEXT:    vle64.v v8, (a0)
 ; LMULMAX2-RV64-NEXT:    li a1, 5
 ; LMULMAX2-RV64-NEXT:    vmv.s.x v0, a1
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI188_0)
-; LMULMAX2-RV64-NEXT:    addi a1, a1, %lo(.LCPI188_0)
-; LMULMAX2-RV64-NEXT:    vlse64.v v8, (a1), zero
-; LMULMAX2-RV64-NEXT:    lui a1, %hi(.LCPI188_1)
-; LMULMAX2-RV64-NEXT:    ld a1, %lo(.LCPI188_1)(a1)
-; LMULMAX2-RV64-NEXT:    vle64.v v10, (a0)
+; LMULMAX2-RV64-NEXT:    lui a1, 349525
+; LMULMAX2-RV64-NEXT:    addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT:    lui a2, %hi(.LCPI188_0)
+; LMULMAX2-RV64-NEXT:    ld a2, %lo(.LCPI188_0)(a2)
+; LMULMAX2-RV64-NEXT:    slli a3, a1, 32
+; LMULMAX2-RV64-NEXT:    add a1, a1, a3
+; LMULMAX2-RV64-NEXT:    vmv.v.x v10, a1
+; LMULMAX2-RV64-NEXT:    vmerge.vxm v10, v10, a2, v0
+; LMULMAX2-RV64-NEXT:    vmulh.vv v10, v8, v10
 ; LMULMAX2-RV64-NEXT:    vmv.v.i v12, -1
 ; LMULMAX2-RV64-NEXT:    vmerge.vim v12, v12, 0, v0
-; LMULMAX2-RV64-NEXT:    vmerge.vxm v8, v8, a1, v0
-; LMULMAX2-RV64-NEXT:    vmulh.vv v8, v10, v8
-; LMULMAX2-RV64-NEXT:    vmacc.vv v8, v10, v12
+; LMULMAX2-RV64-NEXT:    vmadd.vv v12, v8, v10
 ; LMULMAX2-RV64-NEXT:    li a1, 63
-; LMULMAX2-RV64-NEXT:    vsrl.vx v10, v8, a1
-; LMULMAX2-RV64-NEXT:    vmv.v.i v12, 1
-; LMULMAX2-RV64-NEXT:    vmerge.vim v12, v12, 0, v0
-; LMULMAX2-RV64-NEXT:    vsra.vv v8, v8, v12
-; LMULMAX2-RV64-NEXT:    vadd.vv v8, v8, v10
+; LMULMAX2-RV64-NEXT:    vsrl.vx v8, v12, a1
+; LMULMAX2-RV64-NEXT:    vmv.v.i v10, 1
+; LMULMAX2-RV64-NEXT:    vmerge.vim v10, v10, 0, v0
+; LMULMAX2-RV64-NEXT:    vsra.vv v10, v12, v10
+; LMULMAX2-RV64-NEXT:    vadd.vv v8, v10, v8
 ; LMULMAX2-RV64-NEXT:    vse64.v v8, (a0)
 ; LMULMAX2-RV64-NEXT:    ret
 ;
 ; LMULMAX1-RV32-LABEL: mulhs_v4i64:
 ; LMULMAX1-RV32:       # %bb.0:
 ; LMULMAX1-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX1-RV32-NEXT:    vle64.v v8, (a0)
 ; LMULMAX1-RV32-NEXT:    addi a1, a0, 16
 ; LMULMAX1-RV32-NEXT:    vle64.v v9, (a1)
 ; LMULMAX1-RV32-NEXT:    lui a2, %hi(.LCPI188_0)
 ; LMULMAX1-RV32-NEXT:    addi a2, a2, %lo(.LCPI188_0)
 ; LMULMAX1-RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; LMULMAX1-RV32-NEXT:    vle32.v v10, (a2)
 ; LMULMAX1-RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX1-RV32-NEXT:    vdiv.vv v9, v9, v10
 ; LMULMAX1-RV32-NEXT:    vdiv.vv v8, v8, v10
 ; LMULMAX1-RV32-NEXT:    vse64.v v8, (a0)
 ; LMULMAX1-RV32-NEXT:    vse64.v v9, (a1)
 ; LMULMAX1-RV32-NEXT:    ret
 ;
 ; LMULMAX1-RV64-LABEL: mulhs_v4i64:
 ; LMULMAX1-RV64:       # %bb.0:
 ; LMULMAX1-RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; LMULMAX1-RV64-NEXT:    vle64.v v8, (a0)
-; LMULMAX1-RV64-NEXT:    lui a1, %hi(.LCPI188_0)
-; LMULMAX1-RV64-NEXT:    addi a1, a1, %lo(.LCPI188_0)
-; LMULMAX1-RV64-NEXT:    vlse64.v v9, (a1), zero
-; LMULMAX1-RV64-NEXT:    lui a1, %hi(.LCPI188_1)
-; LMULMAX1-RV64-NEXT:    ld a1, %lo(.LCPI188_1)(a1)
-; LMULMAX1-RV64-NEXT:    addi a2, a0, 16
-; LMULMAX1-RV64-NEXT:    vle64.v v10, (a2)
+; LMULMAX1-RV64-NEXT:    addi a1, a0, 16
+; LMULMAX1-RV64-NEXT:    lui a2, 349525
+; LMULMAX1-RV64-NEXT:    addiw a2, a2, 1365
+; LMULMAX1-RV64-NEXT:    slli a3, a2, 32
+; LMULMAX1-RV64-NEXT:    add a2, a2, a3
+; LMULMAX1-RV64-NEXT:    lui a3, %hi(.LCPI188_0)
+; LMULMAX1-RV64-NEXT:    ld a3, %lo(.LCPI188_0)(a3)
+; LMULMAX1-RV64-NEXT:    vle64.v v9, (a1)
+; LMULMAX1-RV64-NEXT:    vmv.v.x v10, a2
 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
-; LMULMAX1-RV64-NEXT:    vmv.s.x v9, a1
+; LMULMAX1-RV64-NEXT:    vmv.s.x v10, a3
 ; LMULMAX1-RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
-; LMULMAX1-RV64-NEXT:    vmulh.vv v11, v10, v9
+; LMULMAX1-RV64-NEXT:    vmulh.vv v11, v9, v10
 ; LMULMAX1-RV64-NEXT:    vid.v v12
 ; LMULMAX1-RV64-NEXT:    vrsub.vi v13, v12, 0
-; LMULMAX1-RV64-NEXT:    vmacc.vv v11, v13, v10
-; LMULMAX1-RV64-NEXT:    li a1, 63
-; LMULMAX1-RV64-NEXT:    vsrl.vx v10, v11, a1
+; LMULMAX1-RV64-NEXT:    vmacc.vv v11, v13, v9
+; LMULMAX1-RV64-NEXT:    li a2, 63
+; LMULMAX1-RV64-NEXT:    vsrl.vx v9, v11, a2
 ; LMULMAX1-RV64-NEXT:    vsra.vv v11, v11, v12
-; LMULMAX1-RV64-NEXT:    vadd.vv v10, v11, v10
-; LMULMAX1-RV64-NEXT:    vmulh.vv v9, v8, v9
-; LMULMAX1-RV64-NEXT:    vmacc.vv v9, v8, v13
-; LMULMAX1-RV64-NEXT:    vsrl.vx v8, v9, a1
-; LMULMAX1-RV64-NEXT:    vsra.vv v9, v9, v12
-; LMULMAX1-RV64-NEXT:    vadd.vv v8, v9, v8
+; LMULMAX1-RV64-NEXT:    vadd.vv v9, v11, v9
+; LMULMAX1-RV64-NEXT:    vmulh.vv v10, v8, v10
+; LMULMAX1-RV64-NEXT:    vmacc.vv v10, v8, v13
+; LMULMAX1-RV64-NEXT:    vsrl.vx v8, v10, a2
+; LMULMAX1-RV64-NEXT:    vsra.vv v10, v10, v12
+; LMULMAX1-RV64-NEXT:    vadd.vv v8, v10, v8
 ; LMULMAX1-RV64-NEXT:    vse64.v v8, (a0)
-; LMULMAX1-RV64-NEXT:    vse64.v v10, (a2)
+; LMULMAX1-RV64-NEXT:    vse64.v v9, (a1)
 ; LMULMAX1-RV64-NEXT:    ret
   %a = load <4 x i64>, ptr %x
   %b = sdiv <4 x i64> %a, <i64 3, i64 -3, i64 3, i64 -3>
   store <4 x i64> %b, ptr %x
   ret void
 }
 
 define void @smin_v32i8(ptr %x, ptr %y) {

 ; RV32-NEXT:    vse64.v v8, (a0)
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: mulhu_vx_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    lui a1, %hi(.LCPI317_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI317_0)(a1)
+; RV64-NEXT:    lui a1, 699051
+; RV64-NEXT:    addiw a1, a1, -1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    vmulhu.vx v8, v8, a1
 ; RV64-NEXT:    vsrl.vi v8, v8, 1
 ; RV64-NEXT:    vse64.v v8, (a0)
 ; RV64-NEXT:    ret
   %a = load <2 x i64>, ptr %x
   %b = udiv <2 x i64> %a, <i64 3, i64 3>
   store <2 x i64> %b, ptr %x
   ret void

llvm/test/CodeGen/RISCV/sextw-removal.ll

 ; RV64I-LABEL: test7:
 ; RV64I:       # %bb.0: # %bb
 ; RV64I-NEXT:    addi sp, sp, -48
 ; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
 ; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_0)
-; RV64I-NEXT:    ld s0, %lo(.LCPI6_0)(a2)
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_1)
-; RV64I-NEXT:    ld s1, %lo(.LCPI6_1)(a2)
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_2)
-; RV64I-NEXT:    ld s2, %lo(.LCPI6_2)(a2)
-; RV64I-NEXT:    lui a2, %hi(.LCPI6_3)
-; RV64I-NEXT:    ld s3, %lo(.LCPI6_3)(a2)
 ; RV64I-NEXT:    sraw a0, a0, a1
+; RV64I-NEXT:    lui a1, 349525
+; RV64I-NEXT:    addiw s0, a1, 1365
+; RV64I-NEXT:    slli a1, s0, 32
+; RV64I-NEXT:    add s0, s0, a1
+; RV64I-NEXT:    lui a1, 209715
+; RV64I-NEXT:    addiw s1, a1, 819
+; RV64I-NEXT:    slli a1, s1, 32
+; RV64I-NEXT:    add s1, s1, a1
+; RV64I-NEXT:    lui a1, 61681
+; RV64I-NEXT:    addiw s2, a1, -241
+; RV64I-NEXT:    slli a1, s2, 32
+; RV64I-NEXT:    add s2, s2, a1
+; RV64I-NEXT:    lui a1, 4112
+; RV64I-NEXT:    addiw s3, a1, 257
+; RV64I-NEXT:    slli a1, s3, 32
+; RV64I-NEXT:    add s3, s3, a1
 ; RV64I-NEXT:  .LBB6_1: # %bb2
 ; RV64I-NEXT:    # =>This Inner Loop Header: Depth=1
 ; RV64I-NEXT:    call foo@plt
 ; RV64I-NEXT:    srli a1, a0, 1
 ; RV64I-NEXT:    and a1, a1, s0
 ; RV64I-NEXT:    sub a0, a0, a1
 ; RV64I-NEXT:    and a1, a0, s1
 ; RV64I-NEXT:    srli a0, a0, 2

llvm/test/CodeGen/RISCV/split-udiv-by-constant.ll

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a3, a1
 ; RV32-NEXT:    mul a0, a5, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_3:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI0_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI0_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 699051
+; RV64-NEXT:    addiw a3, a3, -1365
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a4, a3, a2
+; RV64-NEXT:    mulhu a4, a2, a3
 ; RV64-NEXT:    srli a5, a4, 1
 ; RV64-NEXT:    andi a4, a4, -2
-; RV64-NEXT:    lui a6, %hi(.LCPI0_1)
-; RV64-NEXT:    ld a6, %lo(.LCPI0_1)(a6)
+; RV64-NEXT:    lui a6, %hi(.LCPI0_0)
+; RV64-NEXT:    ld a6, %lo(.LCPI0_0)(a6)
 ; RV64-NEXT:    add a4, a4, a5
-; RV64-NEXT:    sub a3, a3, a4
-; RV64-NEXT:    sub a4, a0, a3
+; RV64-NEXT:    sub a2, a2, a4
+; RV64-NEXT:    sub a4, a0, a2
 ; RV64-NEXT:    mul a5, a4, a6
-; RV64-NEXT:    mulhu a6, a4, a2
+; RV64-NEXT:    mulhu a6, a4, a3
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sltu a0, a0, a3
+; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a2
+; RV64-NEXT:    mul a1, a1, a3
 ; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a4, a2
+; RV64-NEXT:    mul a0, a4, a3
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 3
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_5(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_5:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a3, a1
 ; RV32-NEXT:    mul a0, a5, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_5:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI1_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI1_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 838861
+; RV64-NEXT:    addiw a3, a3, -819
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a4, a3, a2
+; RV64-NEXT:    mulhu a4, a2, a3
 ; RV64-NEXT:    srli a5, a4, 2
 ; RV64-NEXT:    andi a4, a4, -4
-; RV64-NEXT:    lui a6, %hi(.LCPI1_1)
-; RV64-NEXT:    ld a6, %lo(.LCPI1_1)(a6)
+; RV64-NEXT:    lui a6, %hi(.LCPI1_0)
+; RV64-NEXT:    ld a6, %lo(.LCPI1_0)(a6)
 ; RV64-NEXT:    add a4, a4, a5
-; RV64-NEXT:    sub a3, a3, a4
-; RV64-NEXT:    sub a4, a0, a3
+; RV64-NEXT:    sub a2, a2, a4
+; RV64-NEXT:    sub a4, a0, a2
 ; RV64-NEXT:    mul a5, a4, a6
-; RV64-NEXT:    mulhu a6, a4, a2
+; RV64-NEXT:    mulhu a6, a4, a3
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sltu a0, a0, a3
+; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a2
+; RV64-NEXT:    mul a1, a1, a3
 ; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a4, a2
+; RV64-NEXT:    mul a0, a4, a3
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 5
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_7(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_7:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a5, a1
 ; RV32-NEXT:    mul a0, a3, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_15:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI4_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI4_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 559241
+; RV64-NEXT:    addiw a3, a3, -1911
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a2, a3, a2
-; RV64-NEXT:    srli a2, a2, 3
-; RV64-NEXT:    slli a4, a2, 4
-; RV64-NEXT:    sub a2, a2, a4
-; RV64-NEXT:    lui a4, %hi(.LCPI4_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI4_1)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI4_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI4_2)(a5)
-; RV64-NEXT:    add a2, a3, a2
+; RV64-NEXT:    mulhu a3, a2, a3
+; RV64-NEXT:    srli a3, a3, 3
+; RV64-NEXT:    slli a4, a3, 4
+; RV64-NEXT:    lui a5, %hi(.LCPI4_0)
+; RV64-NEXT:    ld a5, %lo(.LCPI4_0)(a5)
+; RV64-NEXT:    sub a3, a3, a4
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    sub a3, a0, a2
-; RV64-NEXT:    mul a4, a3, a4
+; RV64-NEXT:    mul a4, a3, a5
+; RV64-NEXT:    lui a5, 978671
+; RV64-NEXT:    addiw a5, a5, -273
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    mulhu a6, a3, a5
 ; RV64-NEXT:    add a4, a6, a4
 ; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
 ; RV64-NEXT:    mul a1, a1, a5
 ; RV64-NEXT:    add a1, a4, a1
 ; RV64-NEXT:    mul a0, a3, a5
 ; RV64-NEXT:    ret

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a3, a1
 ; RV32-NEXT:    mul a0, a5, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_17:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI5_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI5_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 986895
+; RV64-NEXT:    addiw a3, a3, 241
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a4, a3, a2
+; RV64-NEXT:    mulhu a4, a2, a3
 ; RV64-NEXT:    srli a5, a4, 4
 ; RV64-NEXT:    andi a4, a4, -16
-; RV64-NEXT:    lui a6, %hi(.LCPI5_1)
-; RV64-NEXT:    ld a6, %lo(.LCPI5_1)(a6)
+; RV64-NEXT:    lui a6, %hi(.LCPI5_0)
+; RV64-NEXT:    ld a6, %lo(.LCPI5_0)(a6)
 ; RV64-NEXT:    add a4, a4, a5
-; RV64-NEXT:    sub a3, a3, a4
-; RV64-NEXT:    sub a4, a0, a3
+; RV64-NEXT:    sub a2, a2, a4
+; RV64-NEXT:    sub a4, a0, a2
 ; RV64-NEXT:    mul a5, a4, a6
-; RV64-NEXT:    mulhu a6, a4, a2
+; RV64-NEXT:    mulhu a6, a4, a3
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sltu a0, a0, a3
+; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a2
+; RV64-NEXT:    mul a1, a1, a3
 ; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a4, a2
+; RV64-NEXT:    mul a0, a4, a3
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 17
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_255(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_255:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a5, a1
 ; RV32-NEXT:    mul a0, a3, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_255:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI6_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI6_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 526344
+; RV64-NEXT:    addiw a3, a3, 129
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a2, a3, a2
-; RV64-NEXT:    srli a2, a2, 7
-; RV64-NEXT:    slli a4, a2, 8
-; RV64-NEXT:    sub a2, a2, a4
-; RV64-NEXT:    lui a4, %hi(.LCPI6_1)
-; RV64-NEXT:    ld a4, %lo(.LCPI6_1)(a4)
-; RV64-NEXT:    lui a5, %hi(.LCPI6_2)
-; RV64-NEXT:    ld a5, %lo(.LCPI6_2)(a5)
-; RV64-NEXT:    add a2, a3, a2
+; RV64-NEXT:    mulhu a3, a2, a3
+; RV64-NEXT:    srli a3, a3, 7
+; RV64-NEXT:    slli a4, a3, 8
+; RV64-NEXT:    lui a5, %hi(.LCPI6_0)
+; RV64-NEXT:    ld a5, %lo(.LCPI6_0)(a5)
+; RV64-NEXT:    sub a3, a3, a4
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    sub a3, a0, a2
-; RV64-NEXT:    mul a4, a3, a4
+; RV64-NEXT:    mul a4, a3, a5
+; RV64-NEXT:    lui a5, 1044464
+; RV64-NEXT:    addiw a5, a5, -257
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
 ; RV64-NEXT:    mulhu a6, a3, a5
 ; RV64-NEXT:    add a4, a6, a4
 ; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
 ; RV64-NEXT:    mul a1, a1, a5
 ; RV64-NEXT:    add a1, a4, a1
 ; RV64-NEXT:    mul a0, a3, a5
 ; RV64-NEXT:    ret

 ; RV32-NEXT:    sub a1, a1, a0
 ; RV32-NEXT:    mul a1, a1, a4
 ; RV32-NEXT:    add a1, a3, a1
 ; RV32-NEXT:    mul a0, a5, a4
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_257:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI7_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI7_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 1044496
+; RV64-NEXT:    addiw a3, a3, -255
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a4, a3, a2
+; RV64-NEXT:    mulhu a4, a2, a3
 ; RV64-NEXT:    srli a5, a4, 8
 ; RV64-NEXT:    andi a4, a4, -256
-; RV64-NEXT:    lui a6, %hi(.LCPI7_1)
-; RV64-NEXT:    ld a6, %lo(.LCPI7_1)(a6)
+; RV64-NEXT:    lui a6, %hi(.LCPI7_0)
+; RV64-NEXT:    ld a6, %lo(.LCPI7_0)(a6)
 ; RV64-NEXT:    add a4, a4, a5
-; RV64-NEXT:    sub a3, a3, a4
-; RV64-NEXT:    sub a4, a0, a3
+; RV64-NEXT:    sub a2, a2, a4
+; RV64-NEXT:    sub a4, a0, a2
 ; RV64-NEXT:    mul a5, a4, a6
-; RV64-NEXT:    mulhu a6, a4, a2
+; RV64-NEXT:    mulhu a6, a4, a3
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sltu a0, a0, a3
+; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a2
+; RV64-NEXT:    mul a1, a1, a3
 ; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a4, a2
+; RV64-NEXT:    mul a0, a4, a3
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 257
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_65535(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_65535:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    sub a1, a4, a0
 ; RV32-NEXT:    slli a0, a3, 16
 ; RV32-NEXT:    neg a2, a3
 ; RV32-NEXT:    sub a0, a2, a0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_65535:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI8_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI8_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 524296
+; RV64-NEXT:    addiw a3, a3, 1
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a2, a3, a2
-; RV64-NEXT:    srli a2, a2, 15
-; RV64-NEXT:    slli a4, a2, 16
-; RV64-NEXT:    sub a2, a2, a4
-; RV64-NEXT:    add a2, a3, a2
+; RV64-NEXT:    mulhu a3, a2, a3
+; RV64-NEXT:    srli a3, a3, 15
+; RV64-NEXT:    slli a4, a3, 16
+; RV64-NEXT:    sub a3, a3, a4
+; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    sub a3, a0, a2
 ; RV64-NEXT:    lui a4, 983039
 ; RV64-NEXT:    slli a4, a4, 4
 ; RV64-NEXT:    addi a4, a4, -1
 ; RV64-NEXT:    slli a4, a4, 16
-; RV64-NEXT:    addi a5, a4, -2
-; RV64-NEXT:    mul a5, a3, a5
-; RV64-NEXT:    addi a4, a4, -1
-; RV64-NEXT:    mulhu a6, a3, a4
-; RV64-NEXT:    add a5, a6, a5
+; RV64-NEXT:    addi a4, a4, -2
+; RV64-NEXT:    mul a4, a3, a4
+; RV64-NEXT:    lui a5, 1048560
+; RV64-NEXT:    addiw a5, a5, -1
+; RV64-NEXT:    slli a6, a5, 32
+; RV64-NEXT:    add a5, a5, a6
+; RV64-NEXT:    mulhu a6, a3, a5
+; RV64-NEXT:    add a4, a6, a4
 ; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a4
-; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a3, a4
+; RV64-NEXT:    mul a1, a1, a5
+; RV64-NEXT:    add a1, a4, a1
+; RV64-NEXT:    mul a0, a3, a5
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 65535
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_65537(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_65537:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    sub a0, a3, a5
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_65537:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    add a2, a0, a1
 ; RV64-NEXT:    sltu a3, a2, a0
 ; RV64-NEXT:    add a2, a2, a3
-; RV64-NEXT:    lui a3, 983041
-; RV64-NEXT:    slli a3, a3, 4
-; RV64-NEXT:    addi a3, a3, -1
-; RV64-NEXT:    slli a3, a3, 16
-; RV64-NEXT:    addi a4, a3, 1
+; RV64-NEXT:    lui a3, 1048560
+; RV64-NEXT:    addiw a4, a3, 1
+; RV64-NEXT:    slli a5, a4, 32
+; RV64-NEXT:    add a4, a4, a5
 ; RV64-NEXT:    mulhu a5, a2, a4
-; RV64-NEXT:    lui a6, 1048560
-; RV64-NEXT:    and a6, a5, a6
+; RV64-NEXT:    and a3, a5, a3
 ; RV64-NEXT:    srli a5, a5, 16
+; RV64-NEXT:    add a3, a3, a5
+; RV64-NEXT:    sub a2, a2, a3
+; RV64-NEXT:    sub a3, a0, a2
+; RV64-NEXT:    lui a5, 983041
+; RV64-NEXT:    slli a5, a5, 4
+; RV64-NEXT:    addi a5, a5, -1
+; RV64-NEXT:    slli a5, a5, 16
+; RV64-NEXT:    mul a5, a3, a5
+; RV64-NEXT:    mulhu a6, a3, a4
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sub a2, a2, a5
-; RV64-NEXT:    sub a5, a0, a2
-; RV64-NEXT:    mul a3, a5, a3
-; RV64-NEXT:    mulhu a6, a5, a4
-; RV64-NEXT:    add a3, a6, a3
 ; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
 ; RV64-NEXT:    mul a1, a1, a4
-; RV64-NEXT:    add a1, a3, a1
-; RV64-NEXT:    mul a0, a5, a4
+; RV64-NEXT:    add a1, a5, a1
+; RV64-NEXT:    mul a0, a3, a4
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 65537
   ret iXLen2 %a
 }
 
 define iXLen2 @test_udiv_12(iXLen2 %x) nounwind {
 ; RV32-LABEL: test_udiv_12:
 ; RV32:       # %bb.0:

 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_udiv_12:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    slli a2, a1, 62
 ; RV64-NEXT:    srli a0, a0, 2
 ; RV64-NEXT:    or a0, a0, a2
 ; RV64-NEXT:    srli a1, a1, 2
-; RV64-NEXT:    lui a2, %hi(.LCPI10_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI10_0)(a2)
-; RV64-NEXT:    add a3, a0, a1
-; RV64-NEXT:    sltu a4, a3, a0
+; RV64-NEXT:    add a2, a0, a1
+; RV64-NEXT:    sltu a3, a2, a0
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    lui a3, 699051
+; RV64-NEXT:    addiw a3, a3, -1365
+; RV64-NEXT:    slli a4, a3, 32
 ; RV64-NEXT:    add a3, a3, a4
-; RV64-NEXT:    mulhu a4, a3, a2
+; RV64-NEXT:    mulhu a4, a2, a3
 ; RV64-NEXT:    srli a5, a4, 1
 ; RV64-NEXT:    andi a4, a4, -2
-; RV64-NEXT:    lui a6, %hi(.LCPI10_1)
-; RV64-NEXT:    ld a6, %lo(.LCPI10_1)(a6)
+; RV64-NEXT:    lui a6, %hi(.LCPI10_0)
+; RV64-NEXT:    ld a6, %lo(.LCPI10_0)(a6)
 ; RV64-NEXT:    add a4, a4, a5
-; RV64-NEXT:    sub a3, a3, a4
-; RV64-NEXT:    sub a4, a0, a3
+; RV64-NEXT:    sub a2, a2, a4
+; RV64-NEXT:    sub a4, a0, a2
 ; RV64-NEXT:    mul a5, a4, a6
-; RV64-NEXT:    mulhu a6, a4, a2
+; RV64-NEXT:    mulhu a6, a4, a3
 ; RV64-NEXT:    add a5, a6, a5
-; RV64-NEXT:    sltu a0, a0, a3
+; RV64-NEXT:    sltu a0, a0, a2
 ; RV64-NEXT:    sub a1, a1, a0
-; RV64-NEXT:    mul a1, a1, a2
+; RV64-NEXT:    mul a1, a1, a3
 ; RV64-NEXT:    add a1, a5, a1
-; RV64-NEXT:    mul a0, a4, a2
+; RV64-NEXT:    mul a0, a4, a3
 ; RV64-NEXT:    ret
   %a = udiv iXLen2 %x, 12
   ret iXLen2 %a
 }

llvm/test/CodeGen/RISCV/split-urem-by-constant.ll

 ; RV32-NEXT:    andi a1, a1, -2
 ; RV32-NEXT:    add a1, a1, a2
 ; RV32-NEXT:    sub a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_3:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI0_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI0_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 699051
+; RV64-NEXT:    addiw a1, a1, -1365
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a2, a1, 1
 ; RV64-NEXT:    andi a1, a1, -2
 ; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    sub a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 3
   ret iXLen2 %a

 ; RV32-NEXT:    andi a1, a1, -4
 ; RV32-NEXT:    add a1, a1, a2
 ; RV32-NEXT:    sub a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_5:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI1_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI1_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 838861
+; RV64-NEXT:    addiw a1, a1, -819
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a2, a1, 2
 ; RV64-NEXT:    andi a1, a1, -4
 ; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    sub a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 5
   ret iXLen2 %a

 ; RV32-NEXT:    slli a2, a1, 4
 ; RV32-NEXT:    sub a1, a1, a2
 ; RV32-NEXT:    add a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_15:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI4_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI4_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 559241
+; RV64-NEXT:    addiw a1, a1, -1911
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a1, a1, 3
 ; RV64-NEXT:    slli a2, a1, 4
 ; RV64-NEXT:    sub a1, a1, a2
 ; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 15
   ret iXLen2 %a

 ; RV32-NEXT:    andi a1, a1, -16
 ; RV32-NEXT:    add a1, a1, a2
 ; RV32-NEXT:    sub a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_17:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI5_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI5_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 986895
+; RV64-NEXT:    addiw a1, a1, 241
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a2, a1, 4
 ; RV64-NEXT:    andi a1, a1, -16
 ; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    sub a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 17
   ret iXLen2 %a

 ; RV32-NEXT:    slli a2, a1, 8
 ; RV32-NEXT:    sub a1, a1, a2
 ; RV32-NEXT:    add a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_255:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI6_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI6_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 526344
+; RV64-NEXT:    addiw a1, a1, 129
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a1, a1, 7
 ; RV64-NEXT:    slli a2, a1, 8
 ; RV64-NEXT:    sub a1, a1, a2
 ; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 255
   ret iXLen2 %a

 ; RV32-NEXT:    andi a1, a1, -256
 ; RV32-NEXT:    add a1, a1, a2
 ; RV32-NEXT:    sub a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_257:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI7_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI7_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 1044496
+; RV64-NEXT:    addiw a1, a1, -255
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a2, a1, 8
 ; RV64-NEXT:    andi a1, a1, -256
 ; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    sub a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 257
   ret iXLen2 %a

 ; RV32-NEXT:    slli a2, a1, 16
 ; RV32-NEXT:    sub a1, a1, a2
 ; RV32-NEXT:    add a0, a0, a1
 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_65535:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    lui a2, %hi(.LCPI8_0)
-; RV64-NEXT:    ld a2, %lo(.LCPI8_0)(a2)
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    mulhu a1, a0, a2
+; RV64-NEXT:    lui a1, 524296
+; RV64-NEXT:    addiw a1, a1, 1
+; RV64-NEXT:    slli a2, a1, 32
+; RV64-NEXT:    add a1, a1, a2
+; RV64-NEXT:    mulhu a1, a0, a1
 ; RV64-NEXT:    srli a1, a1, 15
 ; RV64-NEXT:    slli a2, a1, 16
 ; RV64-NEXT:    sub a1, a1, a2
 ; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 65535
   ret iXLen2 %a

 ; RV32-NEXT:    li a1, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_65537:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    add a1, a0, a1
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    add a0, a1, a0
-; RV64-NEXT:    lui a1, 983041
-; RV64-NEXT:    slli a1, a1, 4
-; RV64-NEXT:    addi a1, a1, -1
-; RV64-NEXT:    slli a1, a1, 16
-; RV64-NEXT:    addi a1, a1, 1
-; RV64-NEXT:    mulhu a1, a0, a1
-; RV64-NEXT:    lui a2, 1048560
-; RV64-NEXT:    and a2, a1, a2
-; RV64-NEXT:    srli a1, a1, 16
-; RV64-NEXT:    add a1, a2, a1
+; RV64-NEXT:    lui a1, 1048560
+; RV64-NEXT:    addiw a2, a1, 1
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    mulhu a2, a0, a2
+; RV64-NEXT:    and a1, a2, a1
+; RV64-NEXT:    srli a2, a2, 16
+; RV64-NEXT:    add a1, a1, a2
 ; RV64-NEXT:    sub a0, a0, a1
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 65537
   ret iXLen2 %a
 }
 
 define iXLen2 @test_urem_12(iXLen2 %x) nounwind {

 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_urem_12:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    slli a2, a1, 62
 ; RV64-NEXT:    srli a3, a0, 2
 ; RV64-NEXT:    or a2, a3, a2
 ; RV64-NEXT:    srli a1, a1, 2
-; RV64-NEXT:    lui a3, %hi(.LCPI10_0)
-; RV64-NEXT:    ld a3, %lo(.LCPI10_0)(a3)
 ; RV64-NEXT:    add a1, a2, a1
 ; RV64-NEXT:    sltu a2, a1, a2
 ; RV64-NEXT:    add a1, a1, a2
-; RV64-NEXT:    mulhu a2, a1, a3
+; RV64-NEXT:    lui a2, 699051
+; RV64-NEXT:    addiw a2, a2, -1365
+; RV64-NEXT:    slli a3, a2, 32
+; RV64-NEXT:    add a2, a2, a3
+; RV64-NEXT:    mulhu a2, a1, a2
 ; RV64-NEXT:    srli a3, a2, 1
 ; RV64-NEXT:    andi a2, a2, -2
 ; RV64-NEXT:    add a2, a2, a3
 ; RV64-NEXT:    sub a1, a1, a2
 ; RV64-NEXT:    slli a1, a1, 2
 ; RV64-NEXT:    andi a0, a0, 3
 ; RV64-NEXT:    or a0, a1, a0
 ; RV64-NEXT:    li a1, 0
 ; RV64-NEXT:    ret
   %a = urem iXLen2 %x, 12
   ret iXLen2 %a
 }
 

llvm/test/CodeGen/RISCV/srem-seteq-illegal-types.ll

 ; RV64-NEXT:    srai s2, a2, 31
 ; RV64-NEXT:    li a1, 7
 ; RV64-NEXT:    call __moddi3@plt
 ; RV64-NEXT:    mv s3, a0
 ; RV64-NEXT:    li a1, -5
 ; RV64-NEXT:    mv a0, s1
 ; RV64-NEXT:    call __moddi3@plt
 ; RV64-NEXT:    mv s1, a0
-; RV64-NEXT:    lui a0, %hi(.LCPI3_0)
-; RV64-NEXT:    ld a1, %lo(.LCPI3_0)(a0)
+; RV64-NEXT:    lui a0, 699051
+; RV64-NEXT:    addiw a1, a0, -1365
+; RV64-NEXT:    slli a0, a1, 32
+; RV64-NEXT:    add a1, a1, a0
 ; RV64-NEXT:    mv a0, s2
 ; RV64-NEXT:    call __muldi3@plt
-; RV64-NEXT:    lui a1, %hi(.LCPI3_1)
-; RV64-NEXT:    ld a1, %lo(.LCPI3_1)(a1)
+; RV64-NEXT:    lui a1, %hi(.LCPI3_0)
+; RV64-NEXT:    ld a1, %lo(.LCPI3_0)(a1)
 ; RV64-NEXT:    add a0, a0, a1
 ; RV64-NEXT:    slli a2, a0, 63
 ; RV64-NEXT:    srli a0, a0, 1
 ; RV64-NEXT:    or a0, a0, a2
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    addi s1, s1, -2
 ; RV64-NEXT:    seqz a1, s1
 ; RV64-NEXT:    addi s3, s3, -1

 ; RV64M-NEXT:    srli a5, a4, 63
 ; RV64M-NEXT:    srai a4, a4, 1
 ; RV64M-NEXT:    add a4, a4, a5
 ; RV64M-NEXT:    slli a5, a4, 3
 ; RV64M-NEXT:    add a3, a3, a4
 ; RV64M-NEXT:    sub a3, a3, a5
 ; RV64M-NEXT:    addi a3, a3, -1
 ; RV64M-NEXT:    seqz a3, a3
-; RV64M-NEXT:    lui a4, %hi(.LCPI3_2)
-; RV64M-NEXT:    ld a4, %lo(.LCPI3_2)(a4)
-; RV64M-NEXT:    lui a5, %hi(.LCPI3_3)
-; RV64M-NEXT:    ld a5, %lo(.LCPI3_3)(a5)
+; RV64M-NEXT:    lui a4, 699051
+; RV64M-NEXT:    addiw a4, a4, -1365
+; RV64M-NEXT:    slli a5, a4, 32
+; RV64M-NEXT:    add a4, a4, a5
+; RV64M-NEXT:    lui a5, %hi(.LCPI3_2)
+; RV64M-NEXT:    ld a5, %lo(.LCPI3_2)(a5)
 ; RV64M-NEXT:    addi a2, a2, -2
 ; RV64M-NEXT:    seqz a2, a2
 ; RV64M-NEXT:    mul a1, a1, a4
 ; RV64M-NEXT:    add a1, a1, a5
 ; RV64M-NEXT:    slli a4, a1, 63
 ; RV64M-NEXT:    srli a1, a1, 1
 ; RV64M-NEXT:    or a1, a1, a4
 ; RV64M-NEXT:    sltu a1, a5, a1