llvm/test/CodeGen/AArch64/vqabs.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=aarch64 %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-SD
; RUN: llc -mtriple=aarch64 -global-isel %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-GI

define <16 x i8> @vqabs_test16(<16 x i8> %A) nounwind {
; CHECK-LABEL: vqabs_test16:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    movi v1.16b, #128
; CHECK-NEXT:    movi v2.16b, #127
; CHECK-NEXT:    neg v3.16b, v0.16b
; CHECK-NEXT:    cmeq v1.16b, v0.16b, v1.16b
; CHECK-NEXT:    bsl v1.16b, v2.16b, v3.16b
; CHECK-NEXT:    cmgt v2.16b, v0.16b, #0
; CHECK-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <16 x i8> %A, zeroinitializer
  %1 = icmp eq <16 x i8> %A, <i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128, i8 -128>
  %2 = sub nsw <16 x i8> zeroinitializer, %A
  %3 = select <16 x i1> %1, <16 x i8> <i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127, i8 127>, <16 x i8> %2
  %4 = select <16 x i1> %0, <16 x i8> %A, <16 x i8> %3
  ret <16 x i8> %4
}

define <8 x i16> @vqabs_test8(<8 x i16> %A) nounwind {
; CHECK-SD-LABEL: vqabs_test8:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    movi v1.8h, #128, lsl #8
; CHECK-SD-NEXT:    neg v2.8h, v0.8h
; CHECK-SD-NEXT:    cmgt v3.8h, v0.8h, #0
; CHECK-SD-NEXT:    cmeq v1.8h, v0.8h, v1.8h
; CHECK-SD-NEXT:    bic v2.16b, v2.16b, v1.16b
; CHECK-SD-NEXT:    bic v1.8h, #128, lsl #8
; CHECK-SD-NEXT:    orr v1.16b, v1.16b, v2.16b
; CHECK-SD-NEXT:    bif v0.16b, v1.16b, v3.16b
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: vqabs_test8:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    movi v1.8h, #128, lsl #8
; CHECK-GI-NEXT:    mvni v2.8h, #128, lsl #8
; CHECK-GI-NEXT:    neg v3.8h, v0.8h
; CHECK-GI-NEXT:    cmeq v1.8h, v0.8h, v1.8h
; CHECK-GI-NEXT:    bsl v1.16b, v2.16b, v3.16b
; CHECK-GI-NEXT:    cmgt v2.8h, v0.8h, #0
; CHECK-GI-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-GI-NEXT:    ret
entry:
  %0 = icmp sgt <8 x i16> %A, zeroinitializer
  %1 = icmp eq <8 x i16> %A, <i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768, i16 -32768>
  %2 = sub nsw <8 x i16> zeroinitializer, %A
  %3 = select <8 x i1> %1, <8 x i16> <i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767, i16 32767>, <8 x i16> %2
  %4 = select <8 x i1> %0, <8 x i16> %A, <8 x i16> %3
  ret <8 x i16> %4
}

define <4 x i32> @vqabs_test4(<4 x i32> %A) nounwind {
; CHECK-SD-LABEL: vqabs_test4:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    movi v1.4s, #128, lsl #24
; CHECK-SD-NEXT:    neg v2.4s, v0.4s
; CHECK-SD-NEXT:    cmgt v3.4s, v0.4s, #0
; CHECK-SD-NEXT:    cmeq v1.4s, v0.4s, v1.4s
; CHECK-SD-NEXT:    bic v2.16b, v2.16b, v1.16b
; CHECK-SD-NEXT:    bic v1.4s, #128, lsl #24
; CHECK-SD-NEXT:    orr v1.16b, v1.16b, v2.16b
; CHECK-SD-NEXT:    bif v0.16b, v1.16b, v3.16b
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: vqabs_test4:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    movi v1.4s, #128, lsl #24
; CHECK-GI-NEXT:    mvni v2.4s, #128, lsl #24
; CHECK-GI-NEXT:    neg v3.4s, v0.4s
; CHECK-GI-NEXT:    cmeq v1.4s, v0.4s, v1.4s
; CHECK-GI-NEXT:    bsl v1.16b, v2.16b, v3.16b
; CHECK-GI-NEXT:    cmgt v2.4s, v0.4s, #0
; CHECK-GI-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-GI-NEXT:    ret
entry:
  %0 = icmp sgt <4 x i32> %A, zeroinitializer
  %1 = icmp eq <4 x i32> %A, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
  %2 = sub nsw <4 x i32> zeroinitializer, %A
  %3 = select <4 x i1> %1, <4 x i32> <i32 2147483647, i32 2147483647, i32 2147483647, i32 2147483647>, <4 x i32> %2
  %4 = select <4 x i1> %0, <4 x i32> %A, <4 x i32> %3
  ret <4 x i32> %4
}

define <16 x i8> @vqabs_sat_v16i8(<16 x i8> %A) {
; CHECK-LABEL: vqabs_sat_v16i8:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.16b, v0.16b
; CHECK-NEXT:    cmgt v2.16b, v0.16b, #0
; CHECK-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <16 x i8> %A, zeroinitializer
  %1 = tail call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> zeroinitializer, <16 x i8> %A)
  %2 = select <16 x i1> %0, <16 x i8> %A, <16 x i8> %1
  ret <16 x i8> %2
}

define <8 x i16> @vqabs_sat_v8i16(<8 x i16> %A) {
; CHECK-LABEL: vqabs_sat_v8i16:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.8h, v0.8h
; CHECK-NEXT:    cmgt v2.8h, v0.8h, #0
; CHECK-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <8 x i16> %A, zeroinitializer
  %1 = tail call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> zeroinitializer, <8 x i16> %A)
  %2 = select <8 x i1> %0, <8 x i16> %A, <8 x i16> %1
  ret <8 x i16> %2
}

define <4 x i32> @vqabs_sat_v4i32(<4 x i32> %A) {
; CHECK-LABEL: vqabs_sat_v4i32:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.4s, v0.4s
; CHECK-NEXT:    cmgt v2.4s, v0.4s, #0
; CHECK-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <4 x i32> %A, zeroinitializer
  %1 = tail call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> zeroinitializer, <4 x i32> %A)
  %2 = select <4 x i1> %0, <4 x i32> %A, <4 x i32> %1
  ret <4 x i32> %2
}

define <2 x i64> @vqabs_sat_v2i64(<2 x i64> %A) {
; CHECK-LABEL: vqabs_sat_v2i64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.2d, v0.2d
; CHECK-NEXT:    cmgt v2.2d, v0.2d, #0
; CHECK-NEXT:    bif v0.16b, v1.16b, v2.16b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <2 x i64> %A, zeroinitializer
  %1 = tail call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> zeroinitializer, <2 x i64> %A)
  %2 = select <2 x i1> %0, <2 x i64> %A, <2 x i64> %1
  ret <2 x i64> %2
}

define <8 x i8> @vqabs_sat_v8i8(<8 x i8> %A) {
; CHECK-LABEL: vqabs_sat_v8i8:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.8b, v0.8b
; CHECK-NEXT:    cmgt v2.8b, v0.8b, #0
; CHECK-NEXT:    bif v0.8b, v1.8b, v2.8b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <8 x i8> %A, zeroinitializer
  %1 = tail call <8 x i8> @llvm.ssub.sat.v8i8(<8 x i8> zeroinitializer, <8 x i8> %A)
  %2 = select <8 x i1> %0, <8 x i8> %A, <8 x i8> %1
  ret <8 x i8> %2
}

define <4 x i16> @vqabs_sat_v4i16(<4 x i16> %A) {
; CHECK-LABEL: vqabs_sat_v4i16:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.4h, v0.4h
; CHECK-NEXT:    cmgt v2.4h, v0.4h, #0
; CHECK-NEXT:    bif v0.8b, v1.8b, v2.8b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <4 x i16> %A, zeroinitializer
  %1 = tail call <4 x i16> @llvm.ssub.sat.v4i16(<4 x i16> zeroinitializer, <4 x i16> %A)
  %2 = select <4 x i1> %0, <4 x i16> %A, <4 x i16> %1
  ret <4 x i16> %2
}

define <2 x i32> @vqabs_sat_v2i32(<2 x i32> %A) {
; CHECK-LABEL: vqabs_sat_v2i32:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    sqneg v1.2s, v0.2s
; CHECK-NEXT:    cmgt v2.2s, v0.2s, #0
; CHECK-NEXT:    bif v0.8b, v1.8b, v2.8b
; CHECK-NEXT:    ret
entry:
  %0 = icmp sgt <2 x i32> %A, zeroinitializer
  %1 = tail call <2 x i32> @llvm.ssub.sat.v2i32(<2 x i32> zeroinitializer, <2 x i32> %A)
  %2 = select <2 x i1> %0, <2 x i32> %A, <2 x i32> %1
  ret <2 x i32> %2
}

define <1 x i64> @vqabs_sat_v1i64(<1 x i64> %A) {
; CHECK-SD-LABEL: vqabs_sat_v1i64:
; CHECK-SD:       // %bb.0: // %entry
; CHECK-SD-NEXT:    sqneg d1, d0
; CHECK-SD-NEXT:    cmgt d2, d0, #0
; CHECK-SD-NEXT:    bif v0.8b, v1.8b, v2.8b
; CHECK-SD-NEXT:    ret
;
; CHECK-GI-LABEL: vqabs_sat_v1i64:
; CHECK-GI:       // %bb.0: // %entry
; CHECK-GI-NEXT:    fmov x9, d0
; CHECK-GI-NEXT:    mov x8, #-9223372036854775808 // =0x8000000000000000
; CHECK-GI-NEXT:    negs x9, x9
; CHECK-GI-NEXT:    cset w10, vs
; CHECK-GI-NEXT:    add x8, x8, x9, asr #63
; CHECK-GI-NEXT:    tst w10, #0x1
; CHECK-GI-NEXT:    fmov x10, d0
; CHECK-GI-NEXT:    csel x8, x8, x9, ne
; CHECK-GI-NEXT:    fmov d1, x8
; CHECK-GI-NEXT:    cmp x10, #0
; CHECK-GI-NEXT:    fcsel d0, d0, d1, gt
; CHECK-GI-NEXT:    ret
entry:
  %0 = icmp sgt <1 x i64> %A, zeroinitializer
  %1 = tail call <1 x i64> @llvm.ssub.sat.v1i64(<1 x i64> zeroinitializer, <1 x i64> %A)
  %2 = select <1 x i1> %0, <1 x i64> %A, <1 x i64> %1
  ret <1 x i64> %2
}