llvm/test/CodeGen/RISCV/rvv/vp-vaaddu.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc < %s -mtriple=riscv64 -mattr=+v | FileCheck %s

declare <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i8(<vscale x 2 x i8>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i16(<vscale x 2 x i16>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i16> @llvm.vp.trunc.nxv2i16.nxv2i32(<vscale x 2 x i32>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i32(<vscale x 2 x i32>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16>, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16>, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32>, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
declare <vscale x 2 x i32> @llvm.vp.lshr.nxv2i32(<vscale x 2 x i32>, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)

define <vscale x 2 x i8> @vaaddu_1(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_1:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v8, v9, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %xz, <vscale x 2 x i16> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %a, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

define <vscale x 2 x i8> @vaaddu_2(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v8, v9, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %xz, <vscale x 2 x i16> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> splat (i16 1), <vscale x 2 x i16> %a, <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

define <vscale x 2 x i8> @vaaddu_3(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_3:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v9, v8, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %xz, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %a, <vscale x 2 x i16> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

define <vscale x 2 x i8> @vaaddu_4(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_4:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v9, v8, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %xz, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %yz, <vscale x 2 x i16> %a, <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

define <vscale x 2 x i8> @vaaddu_5(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_5:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v9, v8, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> splat (i16 1), <vscale x 2 x i16> %xz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %a, <vscale x 2 x i16> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

define <vscale x 2 x i8> @vaaddu_6(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_6:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v9, v8, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i16> @llvm.vp.zext.nxv2i16.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> splat (i16 1), <vscale x 2 x i16> %xz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i16> @llvm.vp.add.nxv2i16(<vscale x 2 x i16> %yz, <vscale x 2 x i16> %a, <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i16> @llvm.vp.lshr.nxv2i16(<vscale x 2 x i16> %b, <vscale x 2 x i16> splat (i16 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i16(<vscale x 2 x i16> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

; Test where the size is reduced by 4x instead of 2x.
define <vscale x 2 x i8> @vaaddu_7(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_7:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vsetvli zero, a0, e8, mf4, ta, ma
; CHECK-NEXT:    vaaddu.vv v8, v8, v9, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %xz, <vscale x 2 x i32> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %a, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i32> @llvm.vp.lshr.nxv2i32(<vscale x 2 x i32> %b, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i32(<vscale x 2 x i32> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}

; Test where the zext can't be completely removed.
define <vscale x 2 x i16> @vaaddu_8(<vscale x 2 x i8> %x, <vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
; CHECK-NEXT:    vzext.vf2 v10, v8, v0.t
; CHECK-NEXT:    csrwi vxrm, 0
; CHECK-NEXT:    vzext.vf2 v8, v9, v0.t
; CHECK-NEXT:    vaaddu.vv v8, v10, v8, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i8(<vscale x 2 x i8> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i8(<vscale x 2 x i8> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %xz, <vscale x 2 x i32> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %a, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i32> @llvm.vp.lshr.nxv2i32(<vscale x 2 x i32> %b, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i16> @llvm.vp.trunc.nxv2i16.nxv2i32(<vscale x 2 x i32> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i16> %d
}

; Negative test. The truncate has a smaller type than the zero extend.
; TODO: Could still handle this by truncating after an i16 vaaddu.
define <vscale x 2 x i8> @vaaddu_9(<vscale x 2 x i16> %x, <vscale x 2 x i16> %y, <vscale x 2 x i1> %m, i32 zeroext %vl) {
; CHECK-LABEL: vaaddu_9:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
; CHECK-NEXT:    vwaddu.vv v10, v8, v9, v0.t
; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
; CHECK-NEXT:    vadd.vi v8, v10, 1, v0.t
; CHECK-NEXT:    vsrl.vi v8, v8, 1, v0.t
; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
; CHECK-NEXT:    vnsrl.wi v8, v8, 0, v0.t
; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, ta, ma
; CHECK-NEXT:    vnsrl.wi v8, v8, 0, v0.t
; CHECK-NEXT:    ret
  %xz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i16(<vscale x 2 x i16> %x, <vscale x 2 x i1> %m, i32 %vl)
  %yz = call <vscale x 2 x i32> @llvm.vp.zext.nxv2i32.nxv2i16(<vscale x 2 x i16> %y, <vscale x 2 x i1> %m, i32 %vl)
  %a = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %xz, <vscale x 2 x i32> %yz, <vscale x 2 x i1> %m, i32 %vl)
  %b = call <vscale x 2 x i32> @llvm.vp.add.nxv2i32(<vscale x 2 x i32> %a, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %c = call <vscale x 2 x i32> @llvm.vp.lshr.nxv2i32(<vscale x 2 x i32> %b, <vscale x 2 x i32> splat (i32 1), <vscale x 2 x i1> %m, i32 %vl)
  %d = call <vscale x 2 x i8> @llvm.vp.trunc.nxv2i8.nxv2i32(<vscale x 2 x i32> %c, <vscale x 2 x i1> %m, i32 %vl)
  ret <vscale x 2 x i8> %d
}