nuudlman/llvm: llvm/test/CodeGen/VE/VELIntrinsics/vpcnt.ll

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

;;; Test vector population count intrinsic instructions
;;;
;;; Note:
;;;   We test VPCNT*vl, VPCNT*vl_v, VPCNT*vml_v, PVPCNT*vl, PVPCNT*vl_v, PVPCNT*vml_v instructions.

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vpcnt_vvl(<256 x double> %0) {
; CHECK-LABEL: vpcnt_vvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vpcnt %v0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %2 = tail call fast <256 x double> @llvm.ve.vl.vpcnt.vvl(<256 x double> %0, i32 256)
  ret <256 x double> %2
}

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

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

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @vpcnt_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
; CHECK-LABEL: vpcnt_vvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    vpcnt %v1, %v0, %vm1
; 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.vpcnt.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntlo_vvl(<256 x double> %0) {
; CHECK-LABEL: pvpcntlo_vvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.lo %v0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcntlo.vvl(<256 x double> %0, i32 256)
  ret <256 x double> %2
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntlo_vvvl(<256 x double> %0, <256 x double> %1) {
; CHECK-LABEL: pvpcntlo_vvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.lo %v1, %v0
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.pvpcntlo.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
  ret <256 x double> %3
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntlo_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
; CHECK-LABEL: pvpcntlo_vvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.lo %v1, %v0, %vm1
; 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.pvpcntlo.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntup_vvl(<256 x double> %0) {
; CHECK-LABEL: pvpcntup_vvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.up %v0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcntup.vvl(<256 x double> %0, i32 256)
  ret <256 x double> %2
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntup_vvvl(<256 x double> %0, <256 x double> %1) {
; CHECK-LABEL: pvpcntup_vvvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.up %v1, %v0
; CHECK-NEXT:    lea %s16, 256
; CHECK-NEXT:    lvl %s16
; CHECK-NEXT:    vor %v0, (0)1, %v1
; CHECK-NEXT:    b.l.t (, %s10)
  %3 = tail call fast <256 x double> @llvm.ve.vl.pvpcntup.vvvl(<256 x double> %0, <256 x double> %1, i32 128)
  ret <256 x double> %3
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcntup_vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2) {
; CHECK-LABEL: pvpcntup_vvmvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt.up %v1, %v0, %vm1
; 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.pvpcntup.vvmvl(<256 x double> %0, <256 x i1> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcnt_vvl(<256 x double> %0) {
; CHECK-LABEL: pvpcnt_vvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 256
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt %v0, %v0
; CHECK-NEXT:    b.l.t (, %s10)
  %2 = tail call fast <256 x double> @llvm.ve.vl.pvpcnt.vvl(<256 x double> %0, i32 256)
  ret <256 x double> %2
}

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

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

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

; Function Attrs: nounwind readnone
define fastcc <256 x double> @pvpcnt_vvMvl(<256 x double> %0, <512 x i1> %1, <256 x double> %2) {
; CHECK-LABEL: pvpcnt_vvMvl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lea %s0, 128
; CHECK-NEXT:    lvl %s0
; CHECK-NEXT:    pvpcnt %v1, %v0, %vm2
; 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.pvpcnt.vvMvl(<256 x double> %0, <512 x i1> %1, <256 x double> %2, i32 128)
  ret <256 x double> %4
}

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