; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s

define <vscale x 1 x i1> @intrinsic_vmflt_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vv_nxv1bf16_nxv1bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16(
    <vscale x 1 x bfloat> %0,
    <vscale x 1 x bfloat> %1,
    iXLen %2)

  ret <vscale x 1 x i1> %a
}

define <vscale x 1 x i1> @intrinsic_vmflt_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv1bf16_nxv1bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
; CHECK-NEXT:    vmv1r.v v11, v0
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
; CHECK-NEXT:    vmv1r.v v0, v11
; CHECK-NEXT:    ret
entry:
  %mask = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16(
    <vscale x 1 x bfloat> %1,
    <vscale x 1 x bfloat> %2,
    iXLen %4)
  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16(
    <vscale x 1 x i1> %0,
    <vscale x 1 x bfloat> %2,
    <vscale x 1 x bfloat> %3,
    <vscale x 1 x i1> %mask,
    iXLen %4)

  ret <vscale x 1 x i1> %a
}

define <vscale x 2 x i1> @intrinsic_vmflt_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vv_nxv2bf16_nxv2bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16(
    <vscale x 2 x bfloat> %0,
    <vscale x 2 x bfloat> %1,
    iXLen %2)

  ret <vscale x 2 x i1> %a
}

define <vscale x 2 x i1> @intrinsic_vmflt_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv2bf16_nxv2bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
; CHECK-NEXT:    vmv1r.v v11, v0
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
; CHECK-NEXT:    vmv1r.v v0, v11
; CHECK-NEXT:    ret
entry:
  %mask = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16(
    <vscale x 2 x bfloat> %1,
    <vscale x 2 x bfloat> %2,
    iXLen %4)
  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16(
    <vscale x 2 x i1> %0,
    <vscale x 2 x bfloat> %2,
    <vscale x 2 x bfloat> %3,
    <vscale x 2 x i1> %mask,
    iXLen %4)

  ret <vscale x 2 x i1> %a
}

define <vscale x 4 x i1> @intrinsic_vmflt_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vv_nxv4bf16_nxv4bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16(
    <vscale x 4 x bfloat> %0,
    <vscale x 4 x bfloat> %1,
    iXLen %2)

  ret <vscale x 4 x i1> %a
}

define <vscale x 4 x i1> @intrinsic_vmflt_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv4bf16_nxv4bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
; CHECK-NEXT:    vmv1r.v v11, v0
; CHECK-NEXT:    vmflt.vv v0, v8, v9
; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
; CHECK-NEXT:    vmv.v.v v0, v11
; CHECK-NEXT:    ret
entry:
  %mask = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16(
    <vscale x 4 x bfloat> %1,
    <vscale x 4 x bfloat> %2,
    iXLen %4)
  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16(
    <vscale x 4 x i1> %0,
    <vscale x 4 x bfloat> %2,
    <vscale x 4 x bfloat> %3,
    <vscale x 4 x i1> %mask,
    iXLen %4)

  ret <vscale x 4 x i1> %a
}

define <vscale x 8 x i1> @intrinsic_vmflt_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vv_nxv8bf16_nxv8bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
; CHECK-NEXT:    vmflt.vv v0, v8, v10
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16(
    <vscale x 8 x bfloat> %0,
    <vscale x 8 x bfloat> %1,
    iXLen %2)

  ret <vscale x 8 x i1> %a
}

define <vscale x 8 x i1> @intrinsic_vmflt_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv8bf16_nxv8bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
; CHECK-NEXT:    vmv1r.v v14, v0
; CHECK-NEXT:    vmflt.vv v0, v8, v10
; CHECK-NEXT:    vmflt.vv v14, v10, v12, v0.t
; CHECK-NEXT:    vmv1r.v v0, v14
; CHECK-NEXT:    ret
entry:
  %mask = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16(
    <vscale x 8 x bfloat> %1,
    <vscale x 8 x bfloat> %2,
    iXLen %4)
  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16(
    <vscale x 8 x i1> %0,
    <vscale x 8 x bfloat> %2,
    <vscale x 8 x bfloat> %3,
    <vscale x 8 x i1> %mask,
    iXLen %4)

  ret <vscale x 8 x i1> %a
}

define <vscale x 16 x i1> @intrinsic_vmflt_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vv_nxv16bf16_nxv16bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
; CHECK-NEXT:    vmflt.vv v0, v8, v12
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16(
    <vscale x 16 x bfloat> %0,
    <vscale x 16 x bfloat> %1,
    iXLen %2)

  ret <vscale x 16 x i1> %a
}

define <vscale x 16 x i1> @intrinsic_vmflt_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv16bf16_nxv16bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
; CHECK-NEXT:    vmv1r.v v20, v0
; CHECK-NEXT:    vmflt.vv v0, v8, v12
; CHECK-NEXT:    vmflt.vv v20, v12, v16, v0.t
; CHECK-NEXT:    vmv1r.v v0, v20
; CHECK-NEXT:    ret
entry:
  %mask = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16(
    <vscale x 16 x bfloat> %1,
    <vscale x 16 x bfloat> %2,
    iXLen %4)
  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16(
    <vscale x 16 x i1> %0,
    <vscale x 16 x bfloat> %2,
    <vscale x 16 x bfloat> %3,
    <vscale x 16 x i1> %mask,
    iXLen %4)

  ret <vscale x 16 x i1> %a
}

define <vscale x 1 x i1> @intrinsic_vmflt_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vf_nxv1bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
; CHECK-NEXT:    vmflt.vf v0, v8, fa0
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16(
    <vscale x 1 x bfloat> %0,
    bfloat %1,
    iXLen %2)

  ret <vscale x 1 x i1> %a
}

define <vscale x 1 x i1> @intrinsic_vmflt_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv1bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
; CHECK-NEXT:    vmv1r.v v10, v0
; CHECK-NEXT:    vmv1r.v v0, v9
; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
; CHECK-NEXT:    vmv1r.v v0, v10
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16(
    <vscale x 1 x i1> %0,
    <vscale x 1 x bfloat> %1,
    bfloat %2,
    <vscale x 1 x i1> %3,
    iXLen %4)

  ret <vscale x 1 x i1> %a
}

define <vscale x 2 x i1> @intrinsic_vmflt_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vf_nxv2bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
; CHECK-NEXT:    vmflt.vf v0, v8, fa0
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16(
    <vscale x 2 x bfloat> %0,
    bfloat %1,
    iXLen %2)

  ret <vscale x 2 x i1> %a
}

define <vscale x 2 x i1> @intrinsic_vmflt_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv2bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
; CHECK-NEXT:    vmv1r.v v10, v0
; CHECK-NEXT:    vmv1r.v v0, v9
; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
; CHECK-NEXT:    vmv1r.v v0, v10
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16(
    <vscale x 2 x i1> %0,
    <vscale x 2 x bfloat> %1,
    bfloat %2,
    <vscale x 2 x i1> %3,
    iXLen %4)

  ret <vscale x 2 x i1> %a
}

define <vscale x 4 x i1> @intrinsic_vmflt_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vf_nxv4bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
; CHECK-NEXT:    vmflt.vf v0, v8, fa0
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16(
    <vscale x 4 x bfloat> %0,
    bfloat %1,
    iXLen %2)

  ret <vscale x 4 x i1> %a
}

define <vscale x 4 x i1> @intrinsic_vmflt_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv4bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
; CHECK-NEXT:    vmv1r.v v10, v0
; CHECK-NEXT:    vmv1r.v v0, v9
; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
; CHECK-NEXT:    vmv.v.v v0, v10
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16(
    <vscale x 4 x i1> %0,
    <vscale x 4 x bfloat> %1,
    bfloat %2,
    <vscale x 4 x i1> %3,
    iXLen %4)

  ret <vscale x 4 x i1> %a
}

define <vscale x 8 x i1> @intrinsic_vmflt_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vf_nxv8bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
; CHECK-NEXT:    vmflt.vf v0, v8, fa0
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16(
    <vscale x 8 x bfloat> %0,
    bfloat %1,
    iXLen %2)

  ret <vscale x 8 x i1> %a
}

define <vscale x 8 x i1> @intrinsic_vmflt_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv8bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
; CHECK-NEXT:    vmv1r.v v11, v0
; CHECK-NEXT:    vmv1r.v v0, v10
; CHECK-NEXT:    vmflt.vf v11, v8, fa0, v0.t
; CHECK-NEXT:    vmv1r.v v0, v11
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16(
    <vscale x 8 x i1> %0,
    <vscale x 8 x bfloat> %1,
    bfloat %2,
    <vscale x 8 x i1> %3,
    iXLen %4)

  ret <vscale x 8 x i1> %a
}

define <vscale x 16 x i1> @intrinsic_vmflt_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
; CHECK-LABEL: intrinsic_vmflt_vf_nxv16bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
; CHECK-NEXT:    vmflt.vf v0, v8, fa0
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16(
    <vscale x 16 x bfloat> %0,
    bfloat %1,
    iXLen %2)

  ret <vscale x 16 x i1> %a
}

define <vscale x 16 x i1> @intrinsic_vmflt_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv16bf16_bf16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
; CHECK-NEXT:    vmv1r.v v13, v0
; CHECK-NEXT:    vmv1r.v v0, v12
; CHECK-NEXT:    vmflt.vf v13, v8, fa0, v0.t
; CHECK-NEXT:    vmv1r.v v0, v13
; CHECK-NEXT:    ret
entry:
  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16(
    <vscale x 16 x i1> %0,
    <vscale x 16 x bfloat> %1,
    bfloat %2,
    <vscale x 16 x i1> %3,
    iXLen %4)

  ret <vscale x 16 x i1> %a
}