[RISCV] Make all vector binops use the _TU pseudo form

[reames]

Summary

This continues towards the goal spelled out in https://discourse.llvm.org/t/riscv-transition-in-vector-pseudo-structure-policy-variants/71295. This patch switches all the binary operations (no widen, no narrow, but both int and FP) to use the _TU + implicit_def passthrough form.

Change is mechanical, but I want a second set of eyes on the test diffs. I see mostly minor improvements and scheduling changes.

Note: This only changes the unmasked variants. Masked variants will still go through doPeepholeMaskedRVV and end up in the unsuffixed/TA form. Fixing that will be a separate change.

[reames]

Noticed this set of _TUs is one of the ones *without* a policy operand, need to fix that first.

[reames]

In D152940#4425231, @reames wrote:

Noticed this set of _TUs is one of the ones *without* a policy operand, need to fix that first.

Actually, no, I don't. This patch is fine. At the moment, we're only using _TU for either "tail undisturbed" or "tail undefined". Given that, the lack of the policy flag isn't a problem - the pass through operand tells us which is which. It will be once we start forming "tail agnostic" instances, but that's another patch, and I can group the policy operand addition with the consumer who needs it.

[craig.topper]

LGTM

[craig.topper]

llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td

91

This should be result_type

115

result_type

166

result_type

[reames]

Address review comments.

[craig.topper]

LGTM

llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td

class VPatBinarySDNode_VV<SDPatternOperator vop,
                           OutPatFrag avl,
                           VReg op_reg_class,
                           bit isSEWAware = 0> :
     Pat<(result_type (vop
                      (op_type op_reg_class:$rs1),
                      (op_type op_reg_class:$rs2))),
         (!cast<Instruction>(
                      !if(isSEWAware,
-                         instruction_name#"_VV_"# vlmul.MX#"_E"#!shl(1, log2sew),
-                         instruction_name#"_VV_"# vlmul.MX))
+                         instruction_name#"_VV_"# vlmul.MX#"_E"#!shl(1, log2sew)#"_TU",
+                         instruction_name#"_VV_"# vlmul.MX#"_TU"))
+                     (op_type (IMPLICIT_DEF)),
                      op_reg_class:$rs1,
                      op_reg_class:$rs2,
                      avl, log2sew)>;
 
 class VPatBinarySDNode_XI<SDPatternOperator vop,
                           string instruction_name,
                           string suffix,
                           ValueType result_type,
                           ValueType vop_type,
                           int log2sew,
                           LMULInfo vlmul,
                           OutPatFrag avl,
                           VReg vop_reg_class,
                           ComplexPattern SplatPatKind,
                           DAGOperand xop_kind,
                           bit isSEWAware = 0> :
     Pat<(result_type (vop
                      (vop_type vop_reg_class:$rs1),
                      (vop_type (SplatPatKind xop_kind:$rs2)))),
         (!cast<Instruction>(
                      !if(isSEWAware,
-                         instruction_name#_#suffix#_# vlmul.MX#"_E"#!shl(1, log2sew),
-                         instruction_name#_#suffix#_# vlmul.MX))
+                         instruction_name#_#suffix#_# vlmul.MX#"_E"#!shl(1, log2sew)#"_TU",
+                         instruction_name#_#suffix#_# vlmul.MX#"_TU"))
+                     (vop_type (IMPLICIT_DEF)),
                      vop_reg_class:$rs1,
                      xop_kind:$rs2,
                      avl, log2sew)>;
 
 multiclass VPatBinarySDNode_VV_VX<SDPatternOperator vop, string instruction_name,
                                   list<VTypeInfo> vtilist = AllIntegerVectors,
                                   bit isSEWAware = 0> {
   foreach vti = vtilist in {
class VPatBinarySDNode_VF<SDPatternOperator vop,
                           OutPatFrag avl,
                           VReg vop_reg_class,
                           DAGOperand xop_kind,
                           bit isSEWAware = 0> :
     Pat<(result_type (vop (vop_type vop_reg_class:$rs1),
                           (vop_type (SplatFPOp xop_kind:$rs2)))),
         (!cast<Instruction>(
                      !if(isSEWAware,
-                         instruction_name#"_"#vlmul.MX#"_E"#!shl(1, log2sew),
-                         instruction_name#"_"#vlmul.MX))
+                         instruction_name#"_"#vlmul.MX#"_E"#!shl(1, log2sew)#"_TU",
+                         instruction_name#"_"#vlmul.MX#"_TU"))
+                     (vop_type (IMPLICIT_DEF)),
                      vop_reg_class:$rs1,
                      (xop_type xop_kind:$rs2),
                      avl, log2sew)>;
 
 multiclass VPatBinaryFPSDNode_VV_VF<SDPatternOperator vop, string instruction_name,
                                     bit isSEWAware = 0> {
   foreach vti = AllFloatVectors in {
     let Predicates = GetVTypePredicates<vti>.Predicates in {

 multiclass VPatBinaryFPSDNode_R_VF<SDPatternOperator vop, string instruction_name,
                                    bit isSEWAware = 0> {
   foreach fvti = AllFloatVectors in
     let Predicates = GetVTypePredicates<fvti>.Predicates in
     def : Pat<(fvti.Vector (vop (fvti.Vector (SplatFPOp fvti.Scalar:$rs2)),
                                 (fvti.Vector fvti.RegClass:$rs1))),
               (!cast<Instruction>(
                            !if(isSEWAware,
-                             instruction_name#"_V"#fvti.ScalarSuffix#"_"#fvti.LMul.MX#"_E"#fvti.SEW,
-                             instruction_name#"_V"#fvti.ScalarSuffix#"_"#fvti.LMul.MX))
+                             instruction_name#"_V"#fvti.ScalarSuffix#"_"#fvti.LMul.MX#"_E"#fvti.SEW#"_TU",
+                             instruction_name#"_V"#fvti.ScalarSuffix#"_"#fvti.LMul.MX#"_TU"))
+                           (fvti.Vector (IMPLICIT_DEF)),
                            fvti.RegClass:$rs1,
                            (fvti.Scalar fvti.ScalarRegClass:$rs2),
                            fvti.AVL, fvti.Log2SEW)>;
 }
 
 multiclass VPatIntegerSetCCSDNode_VV<string instruction_name,
                                      CondCode cc> {
   foreach vti = AllIntegerVectors in {

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

 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 349525
 ; RV32-NEXT:    addi a0, a0, 1365
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
-; RV32-NEXT:    vsrl.vx v9, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v10, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v10, v10, a2
-; RV32-NEXT:    vor.vv v9, v10, v9
-; RV32-NEXT:    vsrl.vi v10, v8, 24
+; RV32-NEXT:    vsll.vx v9, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v10, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v10, v10, a2
+; RV32-NEXT:    vor.vv v9, v9, v10
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v11, (a3), zero
+; RV32-NEXT:    vlse64.v v10, (a3), zero
 ; RV32-NEXT:    lui a4, 4080
-; RV32-NEXT:    vand.vx v10, v10, a4
-; RV32-NEXT:    vsrl.vi v12, v8, 8
-; RV32-NEXT:    vand.vv v12, v12, v11
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vor.vv v9, v10, v9
-; RV32-NEXT:    vsll.vx v10, v8, a0
-; RV32-NEXT:    vand.vx v12, v8, a2
-; RV32-NEXT:    vsll.vx v12, v12, a1
-; RV32-NEXT:    vor.vv v10, v10, v12
-; RV32-NEXT:    vand.vx v12, v8, a4
-; RV32-NEXT:    vsll.vi v12, v12, 24
-; RV32-NEXT:    vand.vv v8, v8, v11
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v12, v8
-; RV32-NEXT:    vlse64.v v11, (a3), zero
-; RV32-NEXT:    vor.vv v8, v10, v8
-; RV32-NEXT:    vor.vv v8, v8, v9
+; RV32-NEXT:    vand.vx v11, v8, a4
+; RV32-NEXT:    vsll.vi v11, v11, 24
+; RV32-NEXT:    vand.vv v12, v8, v10
+; RV32-NEXT:    vsll.vi v12, v12, 8
+; RV32-NEXT:    vor.vv v11, v11, v12
+; RV32-NEXT:    vor.vv v9, v9, v11
+; RV32-NEXT:    vsrl.vx v11, v8, a0
+; RV32-NEXT:    vsrl.vx v12, v8, a2
+; RV32-NEXT:    vand.vx v12, v12, a1
+; RV32-NEXT:    vor.vv v11, v12, v11
+; RV32-NEXT:    vsrl.vi v12, v8, 24
+; RV32-NEXT:    vand.vx v12, v12, a4
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v10
+; RV32-NEXT:    vor.vv v8, v8, v12
+; RV32-NEXT:    vlse64.v v10, (a3), zero
+; RV32-NEXT:    vor.vv v8, v8, v11
+; RV32-NEXT:    vor.vv v8, v9, v8
 ; RV32-NEXT:    vsrl.vi v9, v8, 4
-; RV32-NEXT:    vand.vv v9, v9, v11
-; RV32-NEXT:    vand.vv v8, v8, v11
+; RV32-NEXT:    vand.vv v9, v9, v10
+; RV32-NEXT:    vand.vv v8, v8, v10
 ; RV32-NEXT:    vlse64.v v10, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 4
 ; RV32-NEXT:    vor.vv v8, v9, v8
 ; RV32-NEXT:    vsrl.vi v9, v8, 2
 ; RV32-NEXT:    vand.vv v9, v9, v10
 ; RV32-NEXT:    vand.vv v8, v8, v10
 ; RV32-NEXT:    vlse64.v v10, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 2
 ; RV32-NEXT:    vor.vv v8, v9, v8
 ; RV32-NEXT:    vsrl.vi v9, v8, 1
 ; RV32-NEXT:    vand.vv v9, v9, v10
 ; RV32-NEXT:    vand.vv v8, v8, v10
 ; RV32-NEXT:    vadd.vv v8, v8, v8
 ; RV32-NEXT:    vor.vv v8, v9, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bitreverse_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
-; RV64-NEXT:    vsrl.vx v9, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v10, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v10, v10, a2
+; RV64-NEXT:    vand.vx v9, v8, a0
+; RV64-NEXT:    vsll.vi v9, v9, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    vsll.vi v10, v10, 8
+; RV64-NEXT:    vor.vv v9, v9, v10
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v10, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v11, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v11, v11, a4
+; RV64-NEXT:    vor.vv v10, v10, v11
 ; RV64-NEXT:    vor.vv v9, v10, v9
-; RV64-NEXT:    vsrl.vi v10, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v10, v10, a3
-; RV64-NEXT:    vsrl.vi v11, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v11, v11, a4
+; RV64-NEXT:    vsrl.vx v10, v8, a2
+; RV64-NEXT:    vsrl.vx v11, v8, a4
+; RV64-NEXT:    vand.vx v11, v11, a3
 ; RV64-NEXT:    vor.vv v10, v11, v10
-; RV64-NEXT:    vor.vv v9, v10, v9
-; RV64-NEXT:    vand.vx v10, v8, a3
-; 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:    vsrl.vi v11, v8, 24
+; RV64-NEXT:    vand.vx v11, v11, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v11
 ; RV64-NEXT:    vor.vv v8, v8, v10
-; RV64-NEXT:    vor.vv v8, v8, v9
+; 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:    vsll.vi v8, v8, 4

 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 349525
 ; RV32-NEXT:    addi a0, a0, 1365
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
-; RV32-NEXT:    vsrl.vx v10, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v12, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v12, v12, a2
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vsrl.vi v12, v8, 24
+; RV32-NEXT:    vsll.vx v10, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v12, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v12, v12, a2
+; RV32-NEXT:    vor.vv v10, v10, v12
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v14, (a3), zero
+; RV32-NEXT:    vlse64.v v12, (a3), zero
 ; RV32-NEXT:    lui a4, 4080
-; RV32-NEXT:    vand.vx v12, v12, a4
-; RV32-NEXT:    vsrl.vi v16, v8, 8
-; RV32-NEXT:    vand.vv v16, v16, v14
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vsll.vx v12, v8, a0
-; RV32-NEXT:    vand.vx v16, v8, a2
-; RV32-NEXT:    vsll.vx v16, v16, a1
-; RV32-NEXT:    vor.vv v12, v12, v16
-; RV32-NEXT:    vand.vx v16, v8, a4
-; RV32-NEXT:    vsll.vi v16, v16, 24
-; RV32-NEXT:    vand.vv v8, v8, v14
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v16, v8
-; RV32-NEXT:    vlse64.v v14, (a3), zero
-; RV32-NEXT:    vor.vv v8, v12, v8
-; RV32-NEXT:    vor.vv v8, v8, v10
+; RV32-NEXT:    vand.vx v14, v8, a4
+; RV32-NEXT:    vsll.vi v14, v14, 24
+; RV32-NEXT:    vand.vv v16, v8, v12
+; RV32-NEXT:    vsll.vi v16, v16, 8
+; RV32-NEXT:    vor.vv v14, v14, v16
+; RV32-NEXT:    vor.vv v10, v10, v14
+; RV32-NEXT:    vsrl.vx v14, v8, a0
+; RV32-NEXT:    vsrl.vx v16, v8, a2
+; RV32-NEXT:    vand.vx v16, v16, a1
+; RV32-NEXT:    vor.vv v14, v16, v14
+; RV32-NEXT:    vsrl.vi v16, v8, 24
+; RV32-NEXT:    vand.vx v16, v16, a4
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v12
+; RV32-NEXT:    vor.vv v8, v8, v16
+; RV32-NEXT:    vlse64.v v12, (a3), zero
+; RV32-NEXT:    vor.vv v8, v8, v14
+; RV32-NEXT:    vor.vv v8, v10, v8
 ; RV32-NEXT:    vsrl.vi v10, v8, 4
-; RV32-NEXT:    vand.vv v10, v10, v14
-; RV32-NEXT:    vand.vv v8, v8, v14
+; RV32-NEXT:    vand.vv v10, v10, v12
+; RV32-NEXT:    vand.vv v8, v8, v12
 ; RV32-NEXT:    vlse64.v v12, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 4
 ; RV32-NEXT:    vor.vv v8, v10, v8
 ; RV32-NEXT:    vsrl.vi v10, v8, 2
 ; RV32-NEXT:    vand.vv v10, v10, v12
 ; RV32-NEXT:    vand.vv v8, v8, v12
 ; RV32-NEXT:    vlse64.v v12, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 2
 ; RV32-NEXT:    vor.vv v8, v10, v8
 ; RV32-NEXT:    vsrl.vi v10, v8, 1
 ; RV32-NEXT:    vand.vv v10, v10, v12
 ; RV32-NEXT:    vand.vv v8, v8, v12
 ; RV32-NEXT:    vadd.vv v8, v8, v8
 ; RV32-NEXT:    vor.vv v8, v10, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bitreverse_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
-; RV64-NEXT:    vsrl.vx v10, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v12, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v12, v12, a2
+; RV64-NEXT:    vand.vx v10, v8, a0
+; RV64-NEXT:    vsll.vi v10, v10, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    vsll.vi v12, v12, 8
+; RV64-NEXT:    vor.vv v10, v10, v12
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v12, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v14, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v14, v14, a4
+; RV64-NEXT:    vor.vv v12, v12, v14
 ; RV64-NEXT:    vor.vv v10, v12, v10
-; RV64-NEXT:    vsrl.vi v12, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v12, v12, a3
-; RV64-NEXT:    vsrl.vi v14, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v14, v14, a4
+; RV64-NEXT:    vsrl.vx v12, v8, a2
+; RV64-NEXT:    vsrl.vx v14, v8, a4
+; RV64-NEXT:    vand.vx v14, v14, a3
 ; RV64-NEXT:    vor.vv v12, v14, v12
-; RV64-NEXT:    vor.vv v10, v12, v10
-; RV64-NEXT:    vand.vx v12, v8, a3
-; 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:    vsrl.vi v14, v8, 24
+; RV64-NEXT:    vand.vx v14, v14, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v14
 ; RV64-NEXT:    vor.vv v8, v8, v12
-; RV64-NEXT:    vor.vv v8, v8, v10
+; 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:    vsll.vi v8, v8, 4

 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 349525
 ; RV32-NEXT:    addi a0, a0, 1365
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
-; RV32-NEXT:    vsrl.vx v12, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v16, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v16, v16, a2
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vsrl.vi v16, v8, 24
+; RV32-NEXT:    vsll.vx v12, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v16, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v16, v16, a2
+; RV32-NEXT:    vor.vv v12, v12, v16
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v20, (a3), zero
+; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    lui a4, 4080
-; RV32-NEXT:    vand.vx v16, v16, a4
-; RV32-NEXT:    vsrl.vi v24, v8, 8
-; RV32-NEXT:    vand.vv v24, v24, v20
-; RV32-NEXT:    vor.vv v16, v24, v16
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vsll.vx v16, v8, a0
-; RV32-NEXT:    vand.vx v24, v8, a2
-; RV32-NEXT:    vsll.vx v24, v24, a1
-; RV32-NEXT:    vor.vv v16, v16, v24
-; RV32-NEXT:    vand.vx v24, v8, a4
-; RV32-NEXT:    vsll.vi v24, v24, 24
-; RV32-NEXT:    vand.vv v8, v8, v20
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v24, v8
-; RV32-NEXT:    vlse64.v v20, (a3), zero
-; RV32-NEXT:    vor.vv v8, v16, v8
-; RV32-NEXT:    vor.vv v8, v8, v12
+; RV32-NEXT:    vand.vx v20, v8, a4
+; RV32-NEXT:    vsll.vi v20, v20, 24
+; RV32-NEXT:    vand.vv v24, v8, v16
+; RV32-NEXT:    vsll.vi v24, v24, 8
+; RV32-NEXT:    vor.vv v20, v20, v24
+; RV32-NEXT:    vor.vv v12, v12, v20
+; RV32-NEXT:    vsrl.vx v20, v8, a0
+; RV32-NEXT:    vsrl.vx v24, v8, a2
+; RV32-NEXT:    vand.vx v24, v24, a1
+; RV32-NEXT:    vor.vv v20, v24, v20
+; RV32-NEXT:    vsrl.vi v24, v8, 24
+; RV32-NEXT:    vand.vx v24, v24, a4
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v16
+; RV32-NEXT:    vor.vv v8, v8, v24
+; RV32-NEXT:    vlse64.v v16, (a3), zero
+; RV32-NEXT:    vor.vv v8, v8, v20
+; RV32-NEXT:    vor.vv v8, v12, v8
 ; RV32-NEXT:    vsrl.vi v12, v8, 4
-; RV32-NEXT:    vand.vv v12, v12, v20
-; RV32-NEXT:    vand.vv v8, v8, v20
+; RV32-NEXT:    vand.vv v12, v12, v16
+; RV32-NEXT:    vand.vv v8, v8, v16
 ; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 4
 ; RV32-NEXT:    vor.vv v8, v12, v8
 ; RV32-NEXT:    vsrl.vi v12, v8, 2
 ; RV32-NEXT:    vand.vv v12, v12, v16
 ; RV32-NEXT:    vand.vv v8, v8, v16
 ; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 2
 ; RV32-NEXT:    vor.vv v8, v12, v8
 ; RV32-NEXT:    vsrl.vi v12, v8, 1
 ; RV32-NEXT:    vand.vv v12, v12, v16
 ; RV32-NEXT:    vand.vv v8, v8, v16
 ; RV32-NEXT:    vadd.vv v8, v8, v8
 ; RV32-NEXT:    vor.vv v8, v12, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bitreverse_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
-; RV64-NEXT:    vsrl.vx v12, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v16, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v16, v16, a2
+; RV64-NEXT:    vand.vx v12, v8, a0
+; RV64-NEXT:    vsll.vi v12, v12, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    vsll.vi v16, v16, 8
+; RV64-NEXT:    vor.vv v12, v12, v16
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v16, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v20, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v20, v20, a4
+; RV64-NEXT:    vor.vv v16, v16, v20
 ; RV64-NEXT:    vor.vv v12, v16, v12
-; RV64-NEXT:    vsrl.vi v16, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v16, v16, a3
-; RV64-NEXT:    vsrl.vi v20, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v20, v20, a4
+; RV64-NEXT:    vsrl.vx v16, v8, a2
+; RV64-NEXT:    vsrl.vx v20, v8, a4
+; RV64-NEXT:    vand.vx v20, v20, a3
 ; RV64-NEXT:    vor.vv v16, v20, v16
-; RV64-NEXT:    vor.vv v12, v16, v12
-; RV64-NEXT:    vand.vx v16, v8, a3
-; 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:    vsrl.vi v20, v8, 24
+; RV64-NEXT:    vand.vx v20, v20, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v20
 ; RV64-NEXT:    vor.vv v8, v8, v16
-; RV64-NEXT:    vor.vv v8, v8, v12
+; 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:    vsll.vi v8, v8, 4

 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 349525
 ; RV32-NEXT:    addi a0, a0, 1365
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
-; RV32-NEXT:    vsrl.vx v16, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v24, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v24, v24, a2
-; RV32-NEXT:    vor.vv v16, v24, v16
+; RV32-NEXT:    vsll.vx v16, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v24, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v24, v24, a2
+; RV32-NEXT:    vor.vv v16, v16, v24
 ; RV32-NEXT:    addi a3, sp, 16
 ; RV32-NEXT:    vs8r.v v16, (a3) # Unknown-size Folded Spill
-; RV32-NEXT:    vsrl.vi v0, v8, 24
 ; RV32-NEXT:    addi a3, sp, 8
 ; RV32-NEXT:    vlse64.v v24, (a3), zero
 ; RV32-NEXT:    lui a4, 4080
-; RV32-NEXT:    vand.vx v0, v0, a4
-; RV32-NEXT:    vsrl.vi v16, v8, 8
-; RV32-NEXT:    vand.vv v16, v16, v24
-; RV32-NEXT:    vor.vv v16, v16, v0
+; RV32-NEXT:    vand.vx v0, v8, a4
+; RV32-NEXT:    vsll.vi v0, v0, 24
+; RV32-NEXT:    vand.vv v16, v8, v24
+; RV32-NEXT:    vsll.vi v16, v16, 8
+; RV32-NEXT:    vor.vv v16, v0, v16
 ; RV32-NEXT:    addi a5, sp, 16
 ; RV32-NEXT:    vl8r.v v0, (a5) # Unknown-size Folded Reload
-; RV32-NEXT:    vor.vv v16, v16, v0
+; RV32-NEXT:    vor.vv v16, v0, v16
 ; RV32-NEXT:    vs8r.v v16, (a5) # Unknown-size Folded Spill
-; RV32-NEXT:    vand.vx v0, v8, a2
-; RV32-NEXT:    vsll.vx v0, v0, a1
-; RV32-NEXT:    vsll.vx v16, v8, a0
-; RV32-NEXT:    vor.vv v0, v16, v0
-; RV32-NEXT:    vand.vv v16, v8, v24
+; RV32-NEXT:    vsrl.vx v0, v8, a2
+; RV32-NEXT:    vand.vx v0, v0, a1
+; RV32-NEXT:    vsrl.vx v16, v8, a0
+; RV32-NEXT:    vor.vv v0, v0, v16
+; RV32-NEXT:    vsrl.vi v16, v8, 8
+; RV32-NEXT:    vand.vv v16, v16, v24
+; RV32-NEXT:    vsrl.vi v8, v8, 24
 ; RV32-NEXT:    vand.vx v8, v8, a4
-; RV32-NEXT:    vsll.vi v8, v8, 24
-; RV32-NEXT:    vsll.vi v16, v16, 8
-; RV32-NEXT:    vor.vv v8, v8, v16
+; RV32-NEXT:    vor.vv v8, v16, v8
 ; RV32-NEXT:    vlse64.v v16, (a3), zero
-; RV32-NEXT:    vor.vv v8, v0, v8
+; RV32-NEXT:    vor.vv v8, v8, v0
 ; RV32-NEXT:    addi a0, sp, 16
 ; RV32-NEXT:    vl8r.v v24, (a0) # Unknown-size Folded Reload
-; RV32-NEXT:    vor.vv v8, v8, v24
+; RV32-NEXT:    vor.vv v8, v24, v8
 ; RV32-NEXT:    vsrl.vi v24, v8, 4
 ; RV32-NEXT:    vand.vv v24, v24, v16
 ; RV32-NEXT:    vand.vv v8, v8, v16
 ; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    vsll.vi v8, v8, 4
 ; RV32-NEXT:    vor.vv v8, v24, v8
 ; RV32-NEXT:    vsrl.vi v24, v8, 2
 ; RV32-NEXT:    vand.vv v24, v24, v16

 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    slli a0, a0, 3
 ; RV32-NEXT:    add sp, sp, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bitreverse_nxv8i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
-; RV64-NEXT:    vsrl.vx v16, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v24, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v24, v24, a2
+; RV64-NEXT:    vand.vx v16, v8, a0
+; RV64-NEXT:    vsll.vi v16, v16, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v24, v8, a1
+; RV64-NEXT:    vsll.vi v24, v24, 8
+; RV64-NEXT:    vor.vv v16, v16, v24
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v24, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v0, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v0, v0, a4
+; RV64-NEXT:    vor.vv v24, v24, v0
 ; RV64-NEXT:    vor.vv v16, v24, v16
-; RV64-NEXT:    vsrl.vi v24, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v24, v24, a3
-; RV64-NEXT:    vsrl.vi v0, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v0, v0, a4
+; RV64-NEXT:    vsrl.vx v24, v8, a2
+; RV64-NEXT:    vsrl.vx v0, v8, a4
+; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
-; RV64-NEXT:    vor.vv v16, v24, v16
-; RV64-NEXT:    vand.vx v24, v8, a3
-; 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:    vsrl.vi v0, v8, 24
+; RV64-NEXT:    vand.vx v0, v0, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v0
 ; RV64-NEXT:    vor.vv v8, v8, v24
-; RV64-NEXT:    vor.vv v8, v8, v16
+; 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:    vsll.vi v8, v8, 4

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

 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw zero, 12(sp)
 ; RV32-NEXT:    lui a0, 1044480
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
-; RV32-NEXT:    vsrl.vx v9, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v10, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v10, v10, a2
-; RV32-NEXT:    vor.vv v9, v10, v9
-; RV32-NEXT:    vsrl.vi v10, v8, 24
+; RV32-NEXT:    vsll.vx v9, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v10, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v10, v10, a2
+; RV32-NEXT:    vor.vv v9, v9, v10
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v11, (a3), zero
+; RV32-NEXT:    vlse64.v v10, (a3), zero
 ; RV32-NEXT:    lui a3, 4080
-; RV32-NEXT:    vand.vx v10, v10, a3
-; RV32-NEXT:    vsrl.vi v12, v8, 8
-; RV32-NEXT:    vand.vv v12, v12, v11
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vor.vv v9, v10, v9
-; RV32-NEXT:    vsll.vx v10, v8, a0
-; RV32-NEXT:    vand.vx v12, v8, a2
-; RV32-NEXT:    vsll.vx v12, v12, a1
-; RV32-NEXT:    vor.vv v10, v10, v12
-; RV32-NEXT:    vand.vx v12, v8, a3
-; RV32-NEXT:    vsll.vi v12, v12, 24
-; RV32-NEXT:    vand.vv v8, v8, v11
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v12, v8
-; RV32-NEXT:    vor.vv v8, v10, v8
-; RV32-NEXT:    vor.vv v8, v8, v9
+; RV32-NEXT:    vand.vx v11, v8, a3
+; RV32-NEXT:    vsll.vi v11, v11, 24
+; RV32-NEXT:    vand.vv v12, v8, v10
+; RV32-NEXT:    vsll.vi v12, v12, 8
+; RV32-NEXT:    vor.vv v11, v11, v12
+; RV32-NEXT:    vor.vv v9, v9, v11
+; RV32-NEXT:    vsrl.vx v11, v8, a0
+; RV32-NEXT:    vsrl.vx v12, v8, a2
+; RV32-NEXT:    vand.vx v12, v12, a1
+; RV32-NEXT:    vor.vv v11, v12, v11
+; RV32-NEXT:    vsrl.vi v12, v8, 24
+; RV32-NEXT:    vand.vx v12, v12, a3
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v10
+; RV32-NEXT:    vor.vv v8, v8, v12
+; RV32-NEXT:    vor.vv v8, v8, v11
+; RV32-NEXT:    vor.vv v8, v9, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bswap_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
-; RV64-NEXT:    vsrl.vx v9, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v10, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v10, v10, a2
+; RV64-NEXT:    vand.vx v9, v8, a0
+; RV64-NEXT:    vsll.vi v9, v9, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v10, v8, a1
+; RV64-NEXT:    vsll.vi v10, v10, 8
+; RV64-NEXT:    vor.vv v9, v9, v10
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v10, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v11, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v11, v11, a4
+; RV64-NEXT:    vor.vv v10, v10, v11
 ; RV64-NEXT:    vor.vv v9, v10, v9
-; RV64-NEXT:    vsrl.vi v10, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v10, v10, a3
-; RV64-NEXT:    vsrl.vi v11, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v11, v11, a4
+; RV64-NEXT:    vsrl.vx v10, v8, a2
+; RV64-NEXT:    vsrl.vx v11, v8, a4
+; RV64-NEXT:    vand.vx v11, v11, a3
 ; RV64-NEXT:    vor.vv v10, v11, v10
-; RV64-NEXT:    vor.vv v9, v10, v9
-; RV64-NEXT:    vand.vx v10, v8, a3
-; 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:    vsrl.vi v11, v8, 24
+; RV64-NEXT:    vand.vx v11, v11, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v11
 ; RV64-NEXT:    vor.vv v8, v8, v10
-; RV64-NEXT:    vor.vv v8, v8, v9
+; RV64-NEXT:    vor.vv v8, v9, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 1 x i64> @llvm.bswap.nxv1i64(<vscale x 1 x i64> %va)
   ret <vscale x 1 x i64> %a
 }
 declare <vscale x 1 x i64> @llvm.bswap.nxv1i64(<vscale x 1 x i64>)
 
 define <vscale x 2 x i64> @bswap_nxv2i64(<vscale x 2 x i64> %va) {
 ; RV32-LABEL: bswap_nxv2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw zero, 12(sp)
 ; RV32-NEXT:    lui a0, 1044480
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
-; RV32-NEXT:    vsrl.vx v10, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v12, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v12, v12, a2
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vsrl.vi v12, v8, 24
+; RV32-NEXT:    vsll.vx v10, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v12, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v12, v12, a2
+; RV32-NEXT:    vor.vv v10, v10, v12
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v14, (a3), zero
+; RV32-NEXT:    vlse64.v v12, (a3), zero
 ; RV32-NEXT:    lui a3, 4080
-; RV32-NEXT:    vand.vx v12, v12, a3
-; RV32-NEXT:    vsrl.vi v16, v8, 8
-; RV32-NEXT:    vand.vv v16, v16, v14
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vor.vv v10, v12, v10
-; RV32-NEXT:    vsll.vx v12, v8, a0
-; RV32-NEXT:    vand.vx v16, v8, a2
-; RV32-NEXT:    vsll.vx v16, v16, a1
-; RV32-NEXT:    vor.vv v12, v12, v16
-; RV32-NEXT:    vand.vx v16, v8, a3
-; RV32-NEXT:    vsll.vi v16, v16, 24
-; RV32-NEXT:    vand.vv v8, v8, v14
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v16, v8
-; RV32-NEXT:    vor.vv v8, v12, v8
-; RV32-NEXT:    vor.vv v8, v8, v10
+; RV32-NEXT:    vand.vx v14, v8, a3
+; RV32-NEXT:    vsll.vi v14, v14, 24
+; RV32-NEXT:    vand.vv v16, v8, v12
+; RV32-NEXT:    vsll.vi v16, v16, 8
+; RV32-NEXT:    vor.vv v14, v14, v16
+; RV32-NEXT:    vor.vv v10, v10, v14
+; RV32-NEXT:    vsrl.vx v14, v8, a0
+; RV32-NEXT:    vsrl.vx v16, v8, a2
+; RV32-NEXT:    vand.vx v16, v16, a1
+; RV32-NEXT:    vor.vv v14, v16, v14
+; RV32-NEXT:    vsrl.vi v16, v8, 24
+; RV32-NEXT:    vand.vx v16, v16, a3
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v12
+; RV32-NEXT:    vor.vv v8, v8, v16
+; RV32-NEXT:    vor.vv v8, v8, v14
+; RV32-NEXT:    vor.vv v8, v10, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bswap_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m2, ta, ma
-; RV64-NEXT:    vsrl.vx v10, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v12, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v12, v12, a2
+; RV64-NEXT:    vand.vx v10, v8, a0
+; RV64-NEXT:    vsll.vi v10, v10, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v12, v8, a1
+; RV64-NEXT:    vsll.vi v12, v12, 8
+; RV64-NEXT:    vor.vv v10, v10, v12
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v12, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v14, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v14, v14, a4
+; RV64-NEXT:    vor.vv v12, v12, v14
 ; RV64-NEXT:    vor.vv v10, v12, v10
-; RV64-NEXT:    vsrl.vi v12, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v12, v12, a3
-; RV64-NEXT:    vsrl.vi v14, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v14, v14, a4
+; RV64-NEXT:    vsrl.vx v12, v8, a2
+; RV64-NEXT:    vsrl.vx v14, v8, a4
+; RV64-NEXT:    vand.vx v14, v14, a3
 ; RV64-NEXT:    vor.vv v12, v14, v12
-; RV64-NEXT:    vor.vv v10, v12, v10
-; RV64-NEXT:    vand.vx v12, v8, a3
-; 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:    vsrl.vi v14, v8, 24
+; RV64-NEXT:    vand.vx v14, v14, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v14
 ; RV64-NEXT:    vor.vv v8, v8, v12
-; RV64-NEXT:    vor.vv v8, v8, v10
+; RV64-NEXT:    vor.vv v8, v10, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 2 x i64> @llvm.bswap.nxv2i64(<vscale x 2 x i64> %va)
   ret <vscale x 2 x i64> %a
 }
 declare <vscale x 2 x i64> @llvm.bswap.nxv2i64(<vscale x 2 x i64>)
 
 define <vscale x 4 x i64> @bswap_nxv4i64(<vscale x 4 x i64> %va) {
 ; RV32-LABEL: bswap_nxv4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw zero, 12(sp)
 ; RV32-NEXT:    lui a0, 1044480
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
-; RV32-NEXT:    vsrl.vx v12, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v16, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v16, v16, a2
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vsrl.vi v16, v8, 24
+; RV32-NEXT:    vsll.vx v12, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v16, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v16, v16, a2
+; RV32-NEXT:    vor.vv v12, v12, v16
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v20, (a3), zero
+; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    lui a3, 4080
-; RV32-NEXT:    vand.vx v16, v16, a3
-; RV32-NEXT:    vsrl.vi v24, v8, 8
-; RV32-NEXT:    vand.vv v24, v24, v20
-; RV32-NEXT:    vor.vv v16, v24, v16
-; RV32-NEXT:    vor.vv v12, v16, v12
-; RV32-NEXT:    vsll.vx v16, v8, a0
-; RV32-NEXT:    vand.vx v24, v8, a2
-; RV32-NEXT:    vsll.vx v24, v24, a1
-; RV32-NEXT:    vor.vv v16, v16, v24
-; RV32-NEXT:    vand.vx v24, v8, a3
-; RV32-NEXT:    vsll.vi v24, v24, 24
-; RV32-NEXT:    vand.vv v8, v8, v20
-; RV32-NEXT:    vsll.vi v8, v8, 8
-; RV32-NEXT:    vor.vv v8, v24, v8
-; RV32-NEXT:    vor.vv v8, v16, v8
-; RV32-NEXT:    vor.vv v8, v8, v12
+; RV32-NEXT:    vand.vx v20, v8, a3
+; RV32-NEXT:    vsll.vi v20, v20, 24
+; RV32-NEXT:    vand.vv v24, v8, v16
+; RV32-NEXT:    vsll.vi v24, v24, 8
+; RV32-NEXT:    vor.vv v20, v20, v24
+; RV32-NEXT:    vor.vv v12, v12, v20
+; RV32-NEXT:    vsrl.vx v20, v8, a0
+; RV32-NEXT:    vsrl.vx v24, v8, a2
+; RV32-NEXT:    vand.vx v24, v24, a1
+; RV32-NEXT:    vor.vv v20, v24, v20
+; RV32-NEXT:    vsrl.vi v24, v8, 24
+; RV32-NEXT:    vand.vx v24, v24, a3
+; RV32-NEXT:    vsrl.vi v8, v8, 8
+; RV32-NEXT:    vand.vv v8, v8, v16
+; RV32-NEXT:    vor.vv v8, v8, v24
+; RV32-NEXT:    vor.vv v8, v8, v20
+; RV32-NEXT:    vor.vv v8, v12, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bswap_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m4, ta, ma
-; RV64-NEXT:    vsrl.vx v12, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v16, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v16, v16, a2
+; RV64-NEXT:    vand.vx v12, v8, a0
+; RV64-NEXT:    vsll.vi v12, v12, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v16, v8, a1
+; RV64-NEXT:    vsll.vi v16, v16, 8
+; RV64-NEXT:    vor.vv v12, v12, v16
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v16, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v20, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v20, v20, a4
+; RV64-NEXT:    vor.vv v16, v16, v20
 ; RV64-NEXT:    vor.vv v12, v16, v12
-; RV64-NEXT:    vsrl.vi v16, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v16, v16, a3
-; RV64-NEXT:    vsrl.vi v20, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v20, v20, a4
+; RV64-NEXT:    vsrl.vx v16, v8, a2
+; RV64-NEXT:    vsrl.vx v20, v8, a4
+; RV64-NEXT:    vand.vx v20, v20, a3
 ; RV64-NEXT:    vor.vv v16, v20, v16
-; RV64-NEXT:    vor.vv v12, v16, v12
-; RV64-NEXT:    vand.vx v16, v8, a3
-; 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:    vsrl.vi v20, v8, 24
+; RV64-NEXT:    vand.vx v20, v20, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v20
 ; RV64-NEXT:    vor.vv v8, v8, v16
-; RV64-NEXT:    vor.vv v8, v8, v12
+; RV64-NEXT:    vor.vv v8, v12, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 4 x i64> @llvm.bswap.nxv4i64(<vscale x 4 x i64> %va)
   ret <vscale x 4 x i64> %a
 }
 declare <vscale x 4 x i64> @llvm.bswap.nxv4i64(<vscale x 4 x i64>)
 
 define <vscale x 8 x i64> @bswap_nxv8i64(<vscale x 8 x i64> %va) {
 ; RV32-LABEL: bswap_nxv8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    slli a0, a0, 3
 ; RV32-NEXT:    sub sp, sp, a0
 ; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
 ; RV32-NEXT:    sw zero, 12(sp)
 ; RV32-NEXT:    lui a0, 1044480
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    li a0, 56
 ; RV32-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
-; RV32-NEXT:    vsrl.vx v16, v8, a0
-; RV32-NEXT:    li a1, 40
-; RV32-NEXT:    vsrl.vx v24, v8, a1
-; RV32-NEXT:    lui a2, 16
-; RV32-NEXT:    addi a2, a2, -256
-; RV32-NEXT:    vand.vx v24, v24, a2
-; RV32-NEXT:    vor.vv v16, v24, v16
+; RV32-NEXT:    vsll.vx v16, v8, a0
+; RV32-NEXT:    lui a1, 16
+; RV32-NEXT:    addi a1, a1, -256
+; RV32-NEXT:    vand.vx v24, v8, a1
+; RV32-NEXT:    li a2, 40
+; RV32-NEXT:    vsll.vx v24, v24, a2
+; RV32-NEXT:    vor.vv v16, v16, v24
 ; RV32-NEXT:    addi a3, sp, 16
 ; RV32-NEXT:    vs8r.v v16, (a3) # Unknown-size Folded Spill
-; RV32-NEXT:    vsrl.vi v0, v8, 24
 ; RV32-NEXT:    addi a3, sp, 8
-; RV32-NEXT:    vlse64.v v24, (a3), zero
+; RV32-NEXT:    vlse64.v v16, (a3), zero
 ; RV32-NEXT:    lui a3, 4080
-; RV32-NEXT:    vand.vx v0, v0, a3
-; RV32-NEXT:    vsrl.vi v16, v8, 8
-; RV32-NEXT:    vand.vv v16, v16, v24
-; RV32-NEXT:    vor.vv v16, v16, v0
+; RV32-NEXT:    vand.vx v0, v8, a3
+; RV32-NEXT:    vsll.vi v0, v0, 24
+; RV32-NEXT:    vand.vv v24, v8, v16
+; RV32-NEXT:    vsll.vi v24, v24, 8
+; RV32-NEXT:    vor.vv v24, v0, v24
 ; RV32-NEXT:    addi a4, sp, 16
 ; RV32-NEXT:    vl8r.v v0, (a4) # Unknown-size Folded Reload
-; RV32-NEXT:    vor.vv v16, v16, v0
-; RV32-NEXT:    vs8r.v v16, (a4) # Unknown-size Folded Spill
-; RV32-NEXT:    vand.vx v0, v8, a2
-; RV32-NEXT:    vsll.vx v0, v0, a1
-; RV32-NEXT:    vsll.vx v16, v8, a0
-; RV32-NEXT:    vor.vv v16, v16, v0
-; RV32-NEXT:    vand.vv v24, v8, v24
+; RV32-NEXT:    vor.vv v24, v0, v24
+; RV32-NEXT:    vs8r.v v24, (a4) # Unknown-size Folded Spill
+; RV32-NEXT:    vsrl.vx v0, v8, a2
+; RV32-NEXT:    vand.vx v0, v0, a1
+; RV32-NEXT:    vsrl.vx v24, v8, a0
+; RV32-NEXT:    vor.vv v24, v0, v24
+; RV32-NEXT:    vsrl.vi v0, v8, 8
+; RV32-NEXT:    vand.vv v16, v0, v16
+; RV32-NEXT:    vsrl.vi v8, v8, 24
 ; RV32-NEXT:    vand.vx v8, v8, a3
-; RV32-NEXT:    vsll.vi v8, v8, 24
-; RV32-NEXT:    vsll.vi v24, v24, 8
-; RV32-NEXT:    vor.vv v8, v8, v24
 ; RV32-NEXT:    vor.vv v8, v16, v8
+; RV32-NEXT:    vor.vv v8, v8, v24
 ; RV32-NEXT:    addi a0, sp, 16
 ; RV32-NEXT:    vl8r.v v16, (a0) # Unknown-size Folded Reload
-; RV32-NEXT:    vor.vv v8, v8, v16
+; RV32-NEXT:    vor.vv v8, v16, v8
 ; RV32-NEXT:    csrr a0, vlenb
 ; RV32-NEXT:    slli a0, a0, 3
 ; RV32-NEXT:    add sp, sp, a0
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: bswap_nxv8i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    li a0, 56
+; RV64-NEXT:    lui a0, 4080
 ; RV64-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
-; RV64-NEXT:    vsrl.vx v16, v8, a0
-; RV64-NEXT:    li a1, 40
-; RV64-NEXT:    vsrl.vx v24, v8, a1
-; RV64-NEXT:    lui a2, 16
-; RV64-NEXT:    addiw a2, a2, -256
-; RV64-NEXT:    vand.vx v24, v24, a2
+; RV64-NEXT:    vand.vx v16, v8, a0
+; RV64-NEXT:    vsll.vi v16, v16, 24
+; RV64-NEXT:    li a1, 255
+; RV64-NEXT:    slli a1, a1, 24
+; RV64-NEXT:    vand.vx v24, v8, a1
+; RV64-NEXT:    vsll.vi v24, v24, 8
+; RV64-NEXT:    vor.vv v16, v16, v24
+; RV64-NEXT:    li a2, 56
+; RV64-NEXT:    vsll.vx v24, v8, a2
+; RV64-NEXT:    lui a3, 16
+; RV64-NEXT:    addiw a3, a3, -256
+; RV64-NEXT:    vand.vx v0, v8, a3
+; RV64-NEXT:    li a4, 40
+; RV64-NEXT:    vsll.vx v0, v0, a4
+; RV64-NEXT:    vor.vv v24, v24, v0
 ; RV64-NEXT:    vor.vv v16, v24, v16
-; RV64-NEXT:    vsrl.vi v24, v8, 24
-; RV64-NEXT:    lui a3, 4080
-; RV64-NEXT:    vand.vx v24, v24, a3
-; RV64-NEXT:    vsrl.vi v0, v8, 8
-; RV64-NEXT:    li a4, 255
-; RV64-NEXT:    slli a4, a4, 24
-; RV64-NEXT:    vand.vx v0, v0, a4
+; RV64-NEXT:    vsrl.vx v24, v8, a2
+; RV64-NEXT:    vsrl.vx v0, v8, a4
+; RV64-NEXT:    vand.vx v0, v0, a3
 ; RV64-NEXT:    vor.vv v24, v0, v24
-; RV64-NEXT:    vor.vv v16, v24, v16
-; RV64-NEXT:    vand.vx v24, v8, a3
-; 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:    vsrl.vi v0, v8, 24
+; RV64-NEXT:    vand.vx v0, v0, a0
+; RV64-NEXT:    vsrl.vi v8, v8, 8
+; RV64-NEXT:    vand.vx v8, v8, a1
+; RV64-NEXT:    vor.vv v8, v8, v0
 ; RV64-NEXT:    vor.vv v8, v8, v24
-; RV64-NEXT:    vor.vv v8, v8, v16
+; RV64-NEXT:    vor.vv v8, v16, v8
 ; RV64-NEXT:    ret
   %a = call <vscale x 8 x i64> @llvm.bswap.nxv8i64(<vscale x 8 x i64> %va)
   ret <vscale x 8 x i64> %a
 }
 declare <vscale x 8 x i64> @llvm.bswap.nxv8i64(<vscale x 8 x i64>)

llvm/test/CodeGen/RISCV/rvv/calling-conv-fastcc.ll

 ; CHECK-NEXT:    csrr a0, vlenb
 ; CHECK-NEXT:    slli a0, a0, 3
 ; CHECK-NEXT:    add a0, sp, a0
 ; CHECK-NEXT:    addi a0, a0, 16
 ; CHECK-NEXT:    vl8r.v v24, (a0) # Unknown-size Folded Reload
 ; CHECK-NEXT:    vadd.vv v8, v24, v8
 ; CHECK-NEXT:    addi a0, sp, 16
 ; CHECK-NEXT:    vl8r.v v24, (a0) # Unknown-size Folded Reload
-; CHECK-NEXT:    vadd.vv v8, v8, v24
-; CHECK-NEXT:    vadd.vv v24, v0, v16
-; CHECK-NEXT:    vadd.vx v16, v8, a4
-; CHECK-NEXT:    vadd.vx v8, v24, a4
+; CHECK-NEXT:    vadd.vv v24, v8, v24
+; CHECK-NEXT:    vadd.vv v8, v0, v16
+; CHECK-NEXT:    vadd.vx v8, v8, a4
+; CHECK-NEXT:    vadd.vx v16, v24, a4
 ; CHECK-NEXT:    csrr a0, vlenb
 ; CHECK-NEXT:    slli a0, a0, 4
 ; CHECK-NEXT:    add sp, sp, a0
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
   %r = add <vscale x 32 x i32> %x, %y
   %s = add <vscale x 32 x i32> %r, %z
   %head = insertelement <vscale x 32 x i32> poison, i32 %w, i32 0

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

 ; 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
-; RV32F-NEXT:    vrsub.vi v10, v8, 0
-; RV32F-NEXT:    vand.vv v8, v8, v10
+; RV32F-NEXT:    vrsub.vi v9, v8, 0
+; RV32F-NEXT:    vand.vv v9, v8, v9
 ; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v10, v8, v0.t
-; RV32F-NEXT:    vsrl.vi v8, v10, 23
+; RV32F-NEXT:    vfncvt.f.xu.w v10, v9, v0.t
+; RV32F-NEXT:    vsrl.vi v9, v10, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
-; RV32F-NEXT:    vzext.vf2 v10, v8
+; RV32F-NEXT:    vzext.vf2 v10, v9
 ; RV32F-NEXT:    li a1, 127
-; RV32F-NEXT:    vsub.vx v8, v10, a1
+; RV32F-NEXT:    vsub.vx v9, v10, a1
+; RV32F-NEXT:    vmseq.vx v0, v8, zero
 ; RV32F-NEXT:    li a1, 64
-; RV32F-NEXT:    vmv.v.v v0, v9
-; RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32F-NEXT:    vmerge.vxm v8, v9, a1, v0
 ; RV32F-NEXT:    fsrm a0
 ; RV32F-NEXT:    ret
 ;
 ; RV64F-LABEL: cttz_nxv1i64:
 ; RV64F:       # %bb.0:
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV64F-NEXT:    vrsub.vi v9, v8, 0
 ; RV64F-NEXT:    vand.vv v9, v8, v9

 ; RV64F-NEXT:    li a1, 64
 ; RV64F-NEXT:    vmerge.vxm v8, v9, a1, v0
 ; RV64F-NEXT:    fsrm a0
 ; RV64F-NEXT:    ret
 ;
 ; RV32D-LABEL: cttz_nxv1i64:
 ; RV32D:       # %bb.0:
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
-; RV32D-NEXT:    vmseq.vx v9, v8, zero
-; RV32D-NEXT:    vrsub.vi v10, v8, 0
-; RV32D-NEXT:    vand.vv v8, v8, v10
+; RV32D-NEXT:    vrsub.vi v9, v8, 0
+; RV32D-NEXT:    vand.vv v9, v8, v9
 ; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v9, v9, v0.t
 ; RV32D-NEXT:    li a1, 52
-; RV32D-NEXT:    vsrl.vx v8, v8, a1
+; RV32D-NEXT:    vsrl.vx v9, v9, a1
 ; RV32D-NEXT:    li a1, 1023
-; RV32D-NEXT:    vsub.vx v8, v8, a1
+; RV32D-NEXT:    vsub.vx v9, v9, a1
+; RV32D-NEXT:    vmseq.vx v0, v8, zero
 ; RV32D-NEXT:    li a1, 64
-; RV32D-NEXT:    vmv.v.v v0, v9
-; RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32D-NEXT:    vmerge.vxm v8, v9, a1, v0
 ; RV32D-NEXT:    fsrm a0
 ; RV32D-NEXT:    ret
 ;
 ; RV64D-LABEL: cttz_nxv1i64:
 ; RV64D:       # %bb.0:
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV64D-NEXT:    vrsub.vi v9, v8, 0
 ; RV64D-NEXT:    vand.vv v9, v8, v9

 ; 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
-; RV32F-NEXT:    vrsub.vi v12, v8, 0
-; RV32F-NEXT:    vand.vv v8, v8, v12
+; RV32F-NEXT:    vrsub.vi v10, v8, 0
+; RV32F-NEXT:    vand.vv v10, v8, v10
 ; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v11, v8, v0.t
-; RV32F-NEXT:    vsrl.vi v8, v11, 23
+; RV32F-NEXT:    vfncvt.f.xu.w v12, v10, v0.t
+; RV32F-NEXT:    vsrl.vi v10, v12, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
-; RV32F-NEXT:    vzext.vf2 v12, v8
+; RV32F-NEXT:    vzext.vf2 v12, v10
 ; RV32F-NEXT:    li a1, 127
-; RV32F-NEXT:    vsub.vx v8, v12, a1
+; RV32F-NEXT:    vsub.vx v10, v12, a1
+; RV32F-NEXT:    vmseq.vx v0, v8, zero
 ; RV32F-NEXT:    li a1, 64
-; RV32F-NEXT:    vmv1r.v v0, v10
-; RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32F-NEXT:    vmerge.vxm v8, v10, a1, v0
 ; RV32F-NEXT:    fsrm a0
 ; RV32F-NEXT:    ret
 ;
 ; RV64F-LABEL: cttz_nxv2i64:
 ; RV64F:       # %bb.0:
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV64F-NEXT:    vrsub.vi v10, v8, 0
 ; RV64F-NEXT:    vand.vv v10, v8, v10

 ; RV64F-NEXT:    li a1, 64
 ; RV64F-NEXT:    vmerge.vxm v8, v10, a1, v0
 ; RV64F-NEXT:    fsrm a0
 ; RV64F-NEXT:    ret
 ;
 ; RV32D-LABEL: cttz_nxv2i64:
 ; RV32D:       # %bb.0:
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
-; RV32D-NEXT:    vmseq.vx v10, v8, zero
-; RV32D-NEXT:    vrsub.vi v12, v8, 0
-; RV32D-NEXT:    vand.vv v8, v8, v12
+; RV32D-NEXT:    vrsub.vi v10, v8, 0
+; RV32D-NEXT:    vand.vv v10, v8, v10
 ; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v10, v10, v0.t
 ; RV32D-NEXT:    li a1, 52
-; RV32D-NEXT:    vsrl.vx v8, v8, a1
+; RV32D-NEXT:    vsrl.vx v10, v10, a1
 ; RV32D-NEXT:    li a1, 1023
-; RV32D-NEXT:    vsub.vx v8, v8, a1
+; RV32D-NEXT:    vsub.vx v10, v10, a1
+; RV32D-NEXT:    vmseq.vx v0, v8, zero
 ; RV32D-NEXT:    li a1, 64
-; RV32D-NEXT:    vmv1r.v v0, v10
-; RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32D-NEXT:    vmerge.vxm v8, v10, a1, v0
 ; RV32D-NEXT:    fsrm a0
 ; RV32D-NEXT:    ret
 ;
 ; RV64D-LABEL: cttz_nxv2i64:
 ; RV64D:       # %bb.0:
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
 ; RV64D-NEXT:    vrsub.vi v10, v8, 0
 ; RV64D-NEXT:    vand.vv v10, v8, v10

 ; 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
-; RV32F-NEXT:    vrsub.vi v16, v8, 0
-; RV32F-NEXT:    vand.vv v8, v8, v16
+; RV32F-NEXT:    vrsub.vi v12, v8, 0
+; RV32F-NEXT:    vand.vv v12, v8, v12
 ; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v14, v8, v0.t
-; RV32F-NEXT:    vsrl.vi v8, v14, 23
+; RV32F-NEXT:    vfncvt.f.xu.w v16, v12, v0.t
+; RV32F-NEXT:    vsrl.vi v12, v16, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
-; RV32F-NEXT:    vzext.vf2 v16, v8
+; RV32F-NEXT:    vzext.vf2 v16, v12
 ; RV32F-NEXT:    li a1, 127
-; RV32F-NEXT:    vsub.vx v8, v16, a1
+; RV32F-NEXT:    vsub.vx v12, v16, a1
+; RV32F-NEXT:    vmseq.vx v0, v8, zero
 ; RV32F-NEXT:    li a1, 64
-; RV32F-NEXT:    vmv1r.v v0, v12
-; RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32F-NEXT:    vmerge.vxm v8, v12, a1, v0
 ; RV32F-NEXT:    fsrm a0
 ; RV32F-NEXT:    ret
 ;
 ; RV64F-LABEL: cttz_nxv4i64:
 ; RV64F:       # %bb.0:
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV64F-NEXT:    vrsub.vi v12, v8, 0
 ; RV64F-NEXT:    vand.vv v12, v8, v12

 ; RV64F-NEXT:    li a1, 64
 ; RV64F-NEXT:    vmerge.vxm v8, v12, a1, v0
 ; RV64F-NEXT:    fsrm a0
 ; RV64F-NEXT:    ret
 ;
 ; RV32D-LABEL: cttz_nxv4i64:
 ; RV32D:       # %bb.0:
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
-; RV32D-NEXT:    vmseq.vx v12, v8, zero
-; RV32D-NEXT:    vrsub.vi v16, v8, 0
-; RV32D-NEXT:    vand.vv v8, v8, v16
+; RV32D-NEXT:    vrsub.vi v12, v8, 0
+; RV32D-NEXT:    vand.vv v12, v8, v12
 ; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v12, v12, v0.t
 ; RV32D-NEXT:    li a1, 52
-; RV32D-NEXT:    vsrl.vx v8, v8, a1
+; RV32D-NEXT:    vsrl.vx v12, v12, a1
 ; RV32D-NEXT:    li a1, 1023
-; RV32D-NEXT:    vsub.vx v8, v8, a1
+; RV32D-NEXT:    vsub.vx v12, v12, a1
+; RV32D-NEXT:    vmseq.vx v0, v8, zero
 ; RV32D-NEXT:    li a1, 64
-; RV32D-NEXT:    vmv1r.v v0, v12
-; RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32D-NEXT:    vmerge.vxm v8, v12, a1, v0
 ; RV32D-NEXT:    fsrm a0
 ; RV32D-NEXT:    ret
 ;
 ; RV64D-LABEL: cttz_nxv4i64:
 ; RV64D:       # %bb.0:
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
 ; RV64D-NEXT:    vrsub.vi v12, v8, 0
 ; RV64D-NEXT:    vand.vv v12, v8, v12

 ; 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
-; RV32F-NEXT:    vrsub.vi v24, v8, 0
-; RV32F-NEXT:    vand.vv v8, v8, v24
+; RV32F-NEXT:    vrsub.vi v16, v8, 0
+; RV32F-NEXT:    vand.vv v16, v8, v16
 ; RV32F-NEXT:    vmset.m v0
 ; RV32F-NEXT:    fsrmi a0, 1
 ; RV32F-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
-; RV32F-NEXT:    vfncvt.f.xu.w v20, v8, v0.t
-; RV32F-NEXT:    vsrl.vi v8, v20, 23
+; RV32F-NEXT:    vfncvt.f.xu.w v24, v16, v0.t
+; RV32F-NEXT:    vsrl.vi v16, v24, 23
 ; RV32F-NEXT:    vsetvli zero, zero, e64, m8, ta, ma
-; RV32F-NEXT:    vzext.vf2 v24, v8
+; RV32F-NEXT:    vzext.vf2 v24, v16
 ; RV32F-NEXT:    li a1, 127
-; RV32F-NEXT:    vsub.vx v8, v24, a1
+; RV32F-NEXT:    vsub.vx v16, v24, a1
+; RV32F-NEXT:    vmseq.vx v0, v8, zero
 ; RV32F-NEXT:    li a1, 64
-; RV32F-NEXT:    vmv1r.v v0, v16
-; RV32F-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32F-NEXT:    vmerge.vxm v8, v16, a1, v0
 ; RV32F-NEXT:    fsrm a0
 ; RV32F-NEXT:    ret
 ;
 ; RV64F-LABEL: cttz_nxv8i64:
 ; RV64F:       # %bb.0:
 ; RV64F-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64F-NEXT:    vrsub.vi v16, v8, 0
 ; RV64F-NEXT:    vand.vv v16, v8, v16

 ; RV64F-NEXT:    li a1, 64
 ; RV64F-NEXT:    vmerge.vxm v8, v16, a1, v0
 ; RV64F-NEXT:    fsrm a0
 ; RV64F-NEXT:    ret
 ;
 ; RV32D-LABEL: cttz_nxv8i64:
 ; RV32D:       # %bb.0:
 ; RV32D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
-; RV32D-NEXT:    vmseq.vx v16, v8, zero
-; RV32D-NEXT:    vrsub.vi v24, v8, 0
-; RV32D-NEXT:    vand.vv v8, v8, v24
+; RV32D-NEXT:    vrsub.vi v16, v8, 0
+; RV32D-NEXT:    vand.vv v16, v8, v16
 ; RV32D-NEXT:    vmset.m v0
 ; RV32D-NEXT:    fsrmi a0, 1
-; RV32D-NEXT:    vfcvt.f.xu.v v8, v8, v0.t
+; RV32D-NEXT:    vfcvt.f.xu.v v16, v16, v0.t
 ; RV32D-NEXT:    li a1, 52
-; RV32D-NEXT:    vsrl.vx v8, v8, a1
+; RV32D-NEXT:    vsrl.vx v16, v16, a1
 ; RV32D-NEXT:    li a1, 1023
-; RV32D-NEXT:    vsub.vx v8, v8, a1
+; RV32D-NEXT:    vsub.vx v16, v16, a1
+; RV32D-NEXT:    vmseq.vx v0, v8, zero
 ; RV32D-NEXT:    li a1, 64
-; RV32D-NEXT:    vmv1r.v v0, v16
-; RV32D-NEXT:    vmerge.vxm v8, v8, a1, v0
+; RV32D-NEXT:    vmerge.vxm v8, v16, a1, v0
 ; RV32D-NEXT:    fsrm a0
 ; RV32D-NEXT:    ret
 ;
 ; RV64D-LABEL: cttz_nxv8i64:
 ; RV64D:       # %bb.0:
 ; RV64D-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64D-NEXT:    vrsub.vi v16, v8, 0
 ; RV64D-NEXT:    vand.vv v16, v8, v16

llvm/test/CodeGen/RISCV/rvv/stepvector.ll

 ; RV32-NEXT:    vid.v v24
 ; RV32-NEXT:    vmul.vv v8, v24, v8
 ; RV32-NEXT:    vadd.vv v16, v8, v16
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: mul_bigimm_stepvector_nxv16i64:
 ; RV64:       # %bb.0: # %entry
-; RV64-NEXT:    csrr a0, vlenb
-; RV64-NEXT:    lui a1, 1987
-; RV64-NEXT:    addiw a1, a1, -731
-; RV64-NEXT:    slli a1, a1, 12
-; RV64-NEXT:    addi a1, a1, -683
-; RV64-NEXT:    mul a0, a0, a1
-; RV64-NEXT:    vsetvli a2, zero, e64, m8, ta, ma
+; RV64-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64-NEXT:    vid.v v8
-; RV64-NEXT:    vmul.vx v8, v8, a1
+; RV64-NEXT:    lui a0, 1987
+; RV64-NEXT:    addiw a0, a0, -731
+; RV64-NEXT:    slli a0, a0, 12
+; RV64-NEXT:    addi a0, a0, -683
+; RV64-NEXT:    vmul.vx v8, v8, a0
+; RV64-NEXT:    csrr a1, vlenb
+; RV64-NEXT:    mul a0, a1, a0
 ; RV64-NEXT:    vadd.vx v16, v8, a0
 ; RV64-NEXT:    ret
 entry:
   %0 = insertelement <vscale x 16 x i64> poison, i64 33333333333, i32 0
   %1 = shufflevector <vscale x 16 x i64> %0, <vscale x 16 x i64> poison, <vscale x 16 x i32> zeroinitializer
   %2 = call <vscale x 16 x i64> @llvm.experimental.stepvector.nxv16i64()
   %3 = mul <vscale x 16 x i64> %2, %1
   ret <vscale x 16 x i64> %3

 ; RV32-NEXT:    vid.v v8
 ; RV32-NEXT:    vsll.vi v8, v8, 2
 ; RV32-NEXT:    vadd.vv v16, v8, v16
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: shl_stepvector_nxv16i64:
 ; RV64:       # %bb.0: # %entry
-; RV64-NEXT:    csrr a0, vlenb
-; RV64-NEXT:    slli a0, a0, 2
-; RV64-NEXT:    vsetvli a1, zero, e64, m8, ta, ma
+; RV64-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
 ; RV64-NEXT:    vid.v v8
 ; RV64-NEXT:    vsll.vi v8, v8, 2
+; RV64-NEXT:    csrr a0, vlenb
+; RV64-NEXT:    slli a0, a0, 2
 ; RV64-NEXT:    vadd.vx v16, v8, a0
 ; RV64-NEXT:    ret
 entry:
   %0 = insertelement <vscale x 16 x i64> poison, i64 2, i32 0
   %1 = shufflevector <vscale x 16 x i64> %0, <vscale x 16 x i64> poison, <vscale x 16 x i32> zeroinitializer
   %2 = call <vscale x 16 x i64> @llvm.experimental.stepvector.nxv16i64()
   %3 = shl <vscale x 16 x i64> %2, %1
   ret <vscale x 16 x i64> %3
 }

llvm/test/CodeGen/RISCV/rvv/vector-interleave.ll

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b)
   ret <vscale x 8 x i32> %res
 }
 
 define <vscale x 4 x i64> @vector_interleave_nxv4i64_nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b) {
 ; CHECK-LABEL: vector_interleave_nxv4i64_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    srli a0, a0, 2
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; CHECK-NEXT:    vid.v v12
-; CHECK-NEXT:    vand.vi v13, v12, 1
-; CHECK-NEXT:    vmsne.vi v0, v13, 0
 ; CHECK-NEXT:    vsrl.vi v16, v12, 1
+; CHECK-NEXT:    vand.vi v12, v12, 1
+; CHECK-NEXT:    vmsne.vi v0, v12, 0
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
 ; CHECK-NEXT:    vadd.vx v16, v16, a0, v0.t
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-NEXT:    vrgatherei16.vv v12, v8, v16, v0.t
 ; CHECK-NEXT:    vmv.v.v v8, v12
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x i64> @llvm.experimental.vector.interleave2.nxv4i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b)
   ret <vscale x 4 x i64> %res
 }

 ; CHECK-NEXT:    slli a0, a0, 4
 ; CHECK-NEXT:    sub sp, sp, a0
 ; CHECK-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
 ; CHECK-NEXT:    csrr a0, vlenb
 ; CHECK-NEXT:    slli a0, a0, 3
 ; CHECK-NEXT:    add a0, sp, a0
 ; CHECK-NEXT:    addi a0, a0, 16
 ; CHECK-NEXT:    vs8r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    srli a0, a0, 1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, mu
 ; CHECK-NEXT:    vid.v v24
-; CHECK-NEXT:    vand.vi v26, v24, 1
-; CHECK-NEXT:    vmsne.vi v0, v26, 0
 ; CHECK-NEXT:    vsrl.vi v2, v24, 1
+; CHECK-NEXT:    vand.vi v24, v24, 1
+; CHECK-NEXT:    vmsne.vi v0, v24, 0
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 1
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 3
 ; CHECK-NEXT:    add a1, sp, a1
 ; CHECK-NEXT:    addi a1, a1, 16
 ; CHECK-NEXT:    vl8r.v v8, (a1) # Unknown-size Folded Reload
 ; CHECK-NEXT:    vadd.vx v2, v2, a0, v0.t
 ; CHECK-NEXT:    vmv4r.v v12, v16
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m8, ta, ma

 ; CHECK-NEXT:    ret
   %res = call <vscale x 8 x float> @llvm.experimental.vector.interleave2.nxv8f32(<vscale x 4 x float> %a, <vscale x 4 x float> %b)
   ret <vscale x 8 x float> %res
 }
 
 define <vscale x 4 x double> @vector_interleave_nxv4f64_nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b) {
 ; CHECK-LABEL: vector_interleave_nxv4f64_nxv2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    srli a0, a0, 2
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; CHECK-NEXT:    vid.v v12
-; CHECK-NEXT:    vand.vi v13, v12, 1
-; CHECK-NEXT:    vmsne.vi v0, v13, 0
 ; CHECK-NEXT:    vsrl.vi v16, v12, 1
+; CHECK-NEXT:    vand.vi v12, v12, 1
+; CHECK-NEXT:    vmsne.vi v0, v12, 0
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
 ; CHECK-NEXT:    vadd.vx v16, v16, a0, v0.t
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m4, ta, ma
 ; CHECK-NEXT:    vrgatherei16.vv v12, v8, v16, v0.t
 ; CHECK-NEXT:    vmv.v.v v8, v12
 ; CHECK-NEXT:    ret
   %res = call <vscale x 4 x double> @llvm.experimental.vector.interleave2.nxv4f64(<vscale x 2 x double> %a, <vscale x 2 x double> %b)
   ret <vscale x 4 x double> %res
 }

 ; CHECK-NEXT:    slli a0, a0, 4
 ; CHECK-NEXT:    sub sp, sp, a0
 ; CHECK-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
 ; CHECK-NEXT:    csrr a0, vlenb
 ; CHECK-NEXT:    slli a0, a0, 3
 ; CHECK-NEXT:    add a0, sp, a0
 ; CHECK-NEXT:    addi a0, a0, 16
 ; CHECK-NEXT:    vs8r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-NEXT:    csrr a0, vlenb
-; CHECK-NEXT:    srli a0, a0, 1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, mu
 ; CHECK-NEXT:    vid.v v24
-; CHECK-NEXT:    vand.vi v26, v24, 1
-; CHECK-NEXT:    vmsne.vi v0, v26, 0
 ; CHECK-NEXT:    vsrl.vi v2, v24, 1
+; CHECK-NEXT:    vand.vi v24, v24, 1
+; CHECK-NEXT:    vmsne.vi v0, v24, 0
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 1
 ; CHECK-NEXT:    csrr a1, vlenb
 ; CHECK-NEXT:    slli a1, a1, 3
 ; CHECK-NEXT:    add a1, sp, a1
 ; CHECK-NEXT:    addi a1, a1, 16
 ; CHECK-NEXT:    vl8r.v v8, (a1) # Unknown-size Folded Reload
 ; CHECK-NEXT:    vadd.vx v2, v2, a0, v0.t
 ; CHECK-NEXT:    vmv4r.v v12, v16
 ; CHECK-NEXT:    vsetvli zero, zero, e64, m8, ta, ma

llvm/test/CodeGen/RISCV/rvv/vreductions-fp-sdnode.ll

; CHECK-NEXT:    ret
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float>)
 
 define float @vreduce_fmin_nxv32f32(<vscale x 32 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv32f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, 523264
-; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
-; CHECK-NEXT:    vmv.s.x v24, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vfredmin.vs v8, v8, v24
+; CHECK-NEXT:    lui a0, 523264
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double>)
 
; CHECK-NEXT:    ret
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half>)
 
 define half @vreduce_fmax_nxv64f16(<vscale x 64 x half> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv64f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    li a0, -512
-; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, ma
-; CHECK-NEXT:    vmv.s.x v24, a0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vfredmax.vs v8, v8, v24
+; CHECK-NEXT:    li a0, -512
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float>)
 
; CHECK-NEXT:    ret
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float>)
 
 define float @vreduce_fmax_nxv32f32(<vscale x 32 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv32f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, 1047552
-; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, ma
-; CHECK-NEXT:    vmv.s.x v24, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vfredmax.vs v8, v8, v24
+; CHECK-NEXT:    lui a0, 1047552
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double>)