; RUN: llc < %s -mtriple=ve -mattr=+vpu | FileCheck %s

;;; Test vector floating compare intrinsic instructions
;;;
;;; Note:
;;;   We test VFCMP*vvl, VFCMP*vvl_v, VFCMP*rvl, VFCMP*rvl_v, VFCMP*vvml_v,
;;;   VFCMP*rvml_v, PVFCMP*vvl, PVFCMP*vvl_v, PVFCMP*rvl, PVFCMP*rvl_v,
;;;   PVFCMP*vvml_v, and PVFCMP*rvml_v instructions.

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vvvl(<256 x double> %0, <256 x double> %1) {
; CHECK-LABEL: vfcmpd_vvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.d %v0, %v0, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vvvl(<256 x double> %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vvvl(<256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfcmpd_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.d %v2, %v0, %v1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vsvl(double %0, <256 x double> %1) {
; CHECK-LABEL: vfcmpd_vsvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 256
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.d %v0, %s0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vsvl(double %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vsvl(double, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vsvvl(double %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfcmpd_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.d %v1, %s0, %v0
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vsvvl(double %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vsvvl(double, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfcmpd_vvvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.d %v2, %v0, %v1, %vm1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vvvmvl(<256 x double>, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmpd_vsvmvl(double %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfcmpd_vsvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.d %v1, %s0, %v0, %vm1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.vfcmpd.vsvmvl(double %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmpd.vsvmvl(double, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vvvl(<256 x double> %0, <256 x double> %1) {
; CHECK-LABEL: vfcmps_vvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.s %v0, %v0, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vvvl(<256 x double> %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vvvl(<256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfcmps_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.s %v2, %v0, %v1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vsvl(float %0, <256 x double> %1) {
; CHECK-LABEL: vfcmps_vsvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 256
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.s %v0, %s0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vsvl(float %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vsvl(float, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vsvvl(float %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfcmps_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.s %v1, %s0, %v0
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vsvvl(float %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vsvvl(float, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfcmps_vvvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfcmp.s %v2, %v0, %v1, %vm1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vvvmvl(<256 x double>, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfcmps_vsvmvl(float %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfcmps_vsvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfcmp.s %v1, %s0, %v0, %vm1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.vfcmps.vsvmvl(float %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.vfcmps.vsvmvl(float, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vvvl(<256 x double> %0, <256 x double> %1) {
; CHECK-LABEL: pvfcmp_vvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvfcmp %v0, %v0, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vvvl(<256 x double> %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vvvl(<256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: pvfcmp_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvfcmp %v2, %v0, %v1
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vsvl(i64 %0, <256 x double> %1) {
; CHECK-LABEL: pvfcmp_vsvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 256
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    pvfcmp %v0, %s0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vsvl(i64 %0, <256 x double> %1, i32 256)
  ret <256 x double> %3
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vsvl(i64, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vsvvl(i64 %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: pvfcmp_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    pvfcmp %v1, %s0, %v0
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %4 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vsvvl(i64 %0, <256 x double> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vsvvl(i64, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vvvMvl(<256 x double> %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: pvfcmp_vvvMvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvfcmp %v2, %v0, %v1, %vm2
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v2
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vvvMvl(<256 x double> %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vvvMvl(<256 x double>, <256 x double>, <512 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfcmp_vsvMvl(i64 %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: pvfcmp_vsvMvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    pvfcmp %v1, %s0, %v0, %vm2
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %5 = tail call fast <256 x double> @llvm.ve.vl.pvfcmp.vsvMvl(i64 %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3, i32 128)
  ret <256 x double> %5
}

; Function Attrs: nounwind readnone
declare <256 x double> @llvm.ve.vl.pvfcmp.vsvMvl(i64, <256 x double>, <512 x i1>, <256 x double>, i32)