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

;;; Test vector floating multiply intrinsic instructions
;;;
;;; Note:
;;;   We test VFMUL*vvl, VFMUL*vvl_v, VFMUL*rvl, VFMUL*rvl_v, VFMUL*vvml_v,
;;;   VFMUL*rvml_v, PVFMUL*vvl, PVFMUL*vvl_v, PVFMUL*rvl, PVFMUL*rvl_v,
;;;   PVFMUL*vvml_v, and PVFMUL*rvml_v instructions.

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuld_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfmuld_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfmul.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.vfmuld.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.vfmuld.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuld_vsvvl(double %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfmuld_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfmul.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.vfmuld.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.vfmuld.vsvvl(double, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuld_vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfmuld_vvvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfmul.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.vfmuld.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.vfmuld.vvvmvl(<256 x double>, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuld_vsvmvl(double %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfmuld_vsvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfmul.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.vfmuld.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.vfmuld.vsvmvl(double, <256 x double>, <256 x i1>, <256 x double>, i32)

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuls_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfmuls_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfmul.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.vfmuls.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.vfmuls.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuls_vsvvl(float %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: vfmuls_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfmul.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.vfmuls.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.vfmuls.vsvvl(float, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuls_vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfmuls_vvvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vfmul.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.vfmuls.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.vfmuls.vvvmvl(<256 x double>, <256 x double>, <256 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vfmuls_vsvmvl(float %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: vfmuls_vsvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    vfmul.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.vfmuls.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.vfmuls.vsvmvl(float, <256 x double>, <256 x i1>, <256 x double>, i32)

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfmul_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: pvfmul_vvvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvfmul %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.pvfmul.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.pvfmul.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32)

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfmul_vsvvl(i64 %0, <256 x double> %1, <256 x double> %2) {
; CHECK-LABEL: pvfmul_vsvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    pvfmul %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.pvfmul.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.pvfmul.vsvvl(i64, <256 x double>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfmul_vvvMvl(<256 x double> %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: pvfmul_vvvMvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvfmul %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.pvfmul.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.pvfmul.vvvMvl(<256 x double>, <256 x double>, <512 x i1>, <256 x double>, i32)

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvfmul_vsvMvl(i64 %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3) {
; CHECK-LABEL: pvfmul_vsvMvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s1, 128
; CHECK-NEXT:    lvl %s1
; CHECK-NEXT:    pvfmul %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.pvfmul.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.pvfmul.vsvMvl(i64, <256 x double>, <512 x i1>, <256 x double>, i32)