llvm/test/CodeGen/AArch64/neon-addlv.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple aarch64-none-linux-gnu < %s | FileCheck %s --check-prefixes=CHECK,CHECK-SD
; RUN: llc -mtriple aarch64-none-linux-gnu -global-isel -global-isel-abort=2 2>&1 < %s | FileCheck %s --check-prefixes=CHECK,CHECK-GI

; CHECK-GI:         warning: Instruction selection used fallback path for uaddlv_v8i8_urshr

declare <4 x i16>  @llvm.aarch64.neon.uaddlp.v4i16.v8i8(<8 x i8>) nounwind readnone
declare <8 x i16>  @llvm.aarch64.neon.uaddlp.v8i16.v16i8(<16 x i8>) nounwind readnone
declare <4 x i32>  @llvm.aarch64.neon.uaddlp.v4i32.v8i16(<8 x i16>) nounwind readnone
declare <2 x i64>  @llvm.aarch64.neon.uaddlp.v2i64.v4i32(<4 x i32>) nounwind readnone
declare <2 x i32>  @llvm.aarch64.neon.uaddlp.v2i32.v4i16(<4 x i16>) nounwind readnone

declare <4 x i16>  @llvm.aarch64.neon.saddlp.v4i16.v8i8(<8 x i8>) nounwind readnone
declare <8 x i16>  @llvm.aarch64.neon.saddlp.v8i16.v16i8(<16 x i8>) nounwind readnone
declare <4 x i32>  @llvm.aarch64.neon.saddlp.v4i32.v8i16(<8 x i16>) nounwind readnone
declare <2 x i64>  @llvm.aarch64.neon.saddlp.v2i64.v4i32(<4 x i32>) nounwind readnone
declare <2 x i32>  @llvm.aarch64.neon.saddlp.v2i32.v4i16(<4 x i16>) nounwind readnone

declare i16 @llvm.vector.reduce.add.v4i16(<4 x i16>) nounwind readnone
declare i16 @llvm.vector.reduce.add.v8i16(<8 x i16>) nounwind readnone
declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>) nounwind readnone
declare i64 @llvm.vector.reduce.add.v2i64(<2 x i64>) nounwind readnone
declare i32 @llvm.vector.reduce.add.v2i32(<2 x i32>) nounwind readnone

define i16 @uaddlv4h_from_v8i8(ptr %A) nounwind {
; CHECK-LABEL: uaddlv4h_from_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    uaddlv h0, v0.8b
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <8 x i8>, ptr %A
  %tmp3 = call <4 x i16> @llvm.aarch64.neon.uaddlp.v4i16.v8i8(<8 x i8> %tmp1)
  %tmp5 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %tmp3)
  ret i16 %tmp5
}

define i16 @uaddlv16b_from_v16i8(ptr %A) nounwind {
; CHECK-LABEL: uaddlv16b_from_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    uaddlv h0, v0.16b
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <16 x i8>, ptr %A
  %tmp3 = call <8 x i16> @llvm.aarch64.neon.uaddlp.v8i16.v16i8(<16 x i8> %tmp1)
  %tmp5 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %tmp3)
  ret i16 %tmp5
}

define i32 @uaddlv8h_from_v8i16(ptr %A) nounwind {
; CHECK-LABEL: uaddlv8h_from_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    uaddlv s0, v0.8h
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <8 x i16>, ptr %A
  %tmp3 = call <4 x i32> @llvm.aarch64.neon.uaddlp.v4i32.v8i16(<8 x i16> %tmp1)
  %tmp5 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %tmp3)
  ret i32 %tmp5
}

define i64 @uaddlv4s_from_v4i32(ptr %A) nounwind {
; CHECK-LABEL: uaddlv4s_from_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    uaddlv d0, v0.4s
; CHECK-NEXT:    fmov x0, d0
; CHECK-NEXT:    ret
  %tmp1 = load <4 x i32>, ptr %A
  %tmp3 = call <2 x i64> @llvm.aarch64.neon.uaddlp.v2i64.v4i32(<4 x i32> %tmp1)
  %tmp5 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %tmp3)
  ret i64 %tmp5
}

define i32 @uaddlv4h_from_v4i16(ptr %A) nounwind {
; CHECK-LABEL: uaddlv4h_from_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    uaddlv s0, v0.4h
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <4 x i16>, ptr %A
  %tmp3 = call <2 x i32> @llvm.aarch64.neon.uaddlp.v2i32.v4i16(<4 x i16> %tmp1)
  %tmp5 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %tmp3)
  ret i32 %tmp5
}



define i16 @saddlv4h_from_v8i8(ptr %A) nounwind {
; CHECK-LABEL: saddlv4h_from_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    saddlv h0, v0.8b
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <8 x i8>, ptr %A
  %tmp3 = call <4 x i16> @llvm.aarch64.neon.saddlp.v4i16.v8i8(<8 x i8> %tmp1)
  %tmp5 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %tmp3)
  ret i16 %tmp5
}

define i16 @saddlv16b_from_v16i8(ptr %A) nounwind {
; CHECK-LABEL: saddlv16b_from_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    saddlv h0, v0.16b
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <16 x i8>, ptr %A
  %tmp3 = call <8 x i16> @llvm.aarch64.neon.saddlp.v8i16.v16i8(<16 x i8> %tmp1)
  %tmp5 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %tmp3)
  ret i16 %tmp5
}

define i32 @saddlv8h_from_v8i16(ptr %A) nounwind {
; CHECK-LABEL: saddlv8h_from_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    saddlv s0, v0.8h
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <8 x i16>, ptr %A
  %tmp3 = call <4 x i32> @llvm.aarch64.neon.saddlp.v4i32.v8i16(<8 x i16> %tmp1)
  %tmp5 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %tmp3)
  ret i32 %tmp5
}

define i64 @saddlv4s_from_v4i32(ptr %A) nounwind {
; CHECK-LABEL: saddlv4s_from_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    saddlv d0, v0.4s
; CHECK-NEXT:    fmov x0, d0
; CHECK-NEXT:    ret
  %tmp1 = load <4 x i32>, ptr %A
  %tmp3 = call <2 x i64> @llvm.aarch64.neon.saddlp.v2i64.v4i32(<4 x i32> %tmp1)
  %tmp5 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %tmp3)
  ret i64 %tmp5
}

define i32 @saddlv4h_from_v4i16(ptr %A) nounwind {
; CHECK-LABEL: saddlv4h_from_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    saddlv s0, v0.4h
; CHECK-NEXT:    fmov w0, s0
; CHECK-NEXT:    ret
  %tmp1 = load <4 x i16>, ptr %A
  %tmp3 = call <2 x i32> @llvm.aarch64.neon.saddlp.v2i32.v4i16(<4 x i16> %tmp1)
  %tmp5 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %tmp3)
  ret i32 %tmp5
}

declare i32 @llvm.aarch64.neon.uaddlv.i32.v8i8(<8 x i8>) nounwind readnone

define i32 @uaddlv_known_bits_v8i8(<8 x i8> %a) {
; CHECK-SD-LABEL: uaddlv_known_bits_v8i8:
; CHECK-SD:       // %bb.0:
; CHECK-SD-NEXT:    uaddlv h0, v0.8b
; CHECK-SD-NEXT:    fmov w0, s0
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: uaddlv_known_bits_v8i8:
; CHECK-GI:       // %bb.0:
; CHECK-GI-NEXT:    uaddlv h0, v0.8b
; CHECK-GI-NEXT:    fmov w8, s0
; CHECK-GI-NEXT:    and w0, w8, #0xffff
; CHECK-GI-NEXT:    ret
  %tmp1 = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v8i8(<8 x i8> %a)
  %tmp2 = and i32 %tmp1, 65535
  ret i32 %tmp2
}

declare i32 @llvm.aarch64.neon.uaddlv.i32.v16i8(<16 x i8>) nounwind readnone

define i32 @uaddlv_known_bits_v16i8(<16 x i8> %a) {
; CHECK-SD-LABEL: uaddlv_known_bits_v16i8:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    uaddlv h0, v0.16b
; CHECK-SD-NEXT:    fmov w0, s0
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: uaddlv_known_bits_v16i8:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    uaddlv h0, v0.16b
; CHECK-GI-NEXT:    fmov w8, s0
; CHECK-GI-NEXT:    and w0, w8, #0xffff
; CHECK-GI-NEXT:    ret
entry:
  %vaddlv.i = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v16i8(<16 x i8> %a)
  %0 = and i32 %vaddlv.i, 65535
  ret i32 %0
}

define dso_local <8 x i8> @uaddlv_v8i8_dup(<8 x i8> %a) {
; CHECK-SD-LABEL: uaddlv_v8i8_dup:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    uaddlv h0, v0.8b
; CHECK-SD-NEXT:    dup v0.8h, v0.h[0]
; CHECK-SD-NEXT:    rshrn v0.8b, v0.8h, #3
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: uaddlv_v8i8_dup:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    uaddlv h0, v0.8b
; CHECK-GI-NEXT:    fmov w8, s0
; CHECK-GI-NEXT:    dup v0.8h, w8
; CHECK-GI-NEXT:    rshrn v0.8b, v0.8h, #3
; CHECK-GI-NEXT:    ret
entry:
  %vaddlv.i = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v8i8(<8 x i8> %a)
  %0 = trunc i32 %vaddlv.i to i16
  %vecinit.i = insertelement <8 x i16> undef, i16 %0, i64 0
  %vecinit7.i = shufflevector <8 x i16> %vecinit.i, <8 x i16> poison, <8 x i32> zeroinitializer
  %vrshrn_n2 = tail call <8 x i8> @llvm.aarch64.neon.rshrn.v8i8(<8 x i16> %vecinit7.i, i32 3)
  ret <8 x i8> %vrshrn_n2
}

declare <8 x i8> @llvm.aarch64.neon.rshrn.v8i8(<8 x i16>, i32)
declare i64 @llvm.aarch64.neon.urshl.i64(i64, i64)

define <8 x i8> @uaddlv_v8i8_urshr(<8 x i8> %a) {
; CHECK-LABEL: uaddlv_v8i8_urshr:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    uaddlv h0, v0.8b
; CHECK-NEXT:    urshr d0, d0, #3
; CHECK-NEXT:    dup v0.8b, v0.b[0]
; CHECK-NEXT:    ret
entry:
  %vaddlv.i = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v8i8(<8 x i8> %a)
  %0 = and i32 %vaddlv.i, 65535
  %conv = zext i32 %0 to i64
  %vrshr_n = tail call i64 @llvm.aarch64.neon.urshl.i64(i64 %conv, i64 -3)
  %conv1 = trunc i64 %vrshr_n to i8
  %vecinit.i = insertelement <8 x i8> undef, i8 %conv1, i64 0
  %vecinit7.i = shufflevector <8 x i8> %vecinit.i, <8 x i8> poison, <8 x i32> zeroinitializer
  ret <8 x i8> %vecinit7.i
}

define <4 x i32> @uaddlv_dup_v4i16(<4 x i16> %a) {
; CHECK-LABEL: uaddlv_dup_v4i16:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    uaddlv s0, v0.4h
; CHECK-NEXT:    dup v0.4s, v0.s[0]
; CHECK-NEXT:    ushr v0.4s, v0.4s, #3
; CHECK-NEXT:    ret
entry:
  %vaddlv.i = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v4i16(<4 x i16> %a)
  %vecinit.i = insertelement <4 x i32> undef, i32 %vaddlv.i, i64 0
  %vecinit7.i = shufflevector <4 x i32> %vecinit.i, <4 x i32> poison, <4 x i32> zeroinitializer
  %vshr_n = lshr <4 x i32> %vecinit7.i, <i32 3, i32 3, i32 3, i32 3>
  ret <4 x i32> %vshr_n
}

define <4 x i32> @uaddlv_dup_v8i16(<8 x i16> %a) {
; CHECK-LABEL: uaddlv_dup_v8i16:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    uaddlv s0, v0.8h
; CHECK-NEXT:    dup v0.4s, v0.s[0]
; CHECK-NEXT:    ushr v0.4s, v0.4s, #3
; CHECK-NEXT:    ret
entry:
  %vaddlv.i = tail call i32 @llvm.aarch64.neon.uaddlv.i32.v8i16(<8 x i16> %a)
  %vecinit.i = insertelement <4 x i32> undef, i32 %vaddlv.i, i64 0
  %vecinit7.i = shufflevector <4 x i32> %vecinit.i, <4 x i32> poison, <4 x i32> zeroinitializer
  %vshr_n = lshr <4 x i32> %vecinit7.i, <i32 3, i32 3, i32 3, i32 3>
  ret <4 x i32> %vshr_n
}

declare i32 @llvm.aarch64.neon.uaddlv.i32.v8i16(<8 x i16>)
declare i32 @llvm.aarch64.neon.uaddlv.i32.v4i16(<4 x i16>)