; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme2 < %s | FileCheck %s

; lookup table expand one register

define {<vscale x 16 x i8>, <vscale x 16 x i8>} @luti2_i8(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.b, z1.b }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv16i8(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 16 x i8>, <vscale x 16 x i8>} %res
}

define {<vscale x 8 x i16>, <vscale x 8 x i16>} @luti2_i16(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.h, z1.h }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8i16(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 8 x i16>, <vscale x 8 x i16>} %res
}

define {<vscale x 4 x i32>, <vscale x 4 x i32>} @luti2_i32(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.s, z1.s }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv4i32(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 4 x i32>, <vscale x 4 x i32>} %res
}

define {<vscale x 8 x half>, <vscale x 8 x half>} @luti2_f16(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.h, z1.h }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8f16(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 8 x half>, <vscale x 8 x half>} %res
}

define {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @luti2_bf16(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_bf16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.h, z1.h }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8bf16(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} %res
}

define {<vscale x 4 x float>, <vscale x 4 x float>} @luti2_f32(<vscale x 16 x i8> %x) {
; CHECK-LABEL: luti2_f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    luti2 { z0.s, z1.s }, zt0, z0[7]
; CHECK-NEXT:    ret
    %res = call {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv4f32(i32 0, <vscale x 16 x i8> %x, i32 7)
    ret {<vscale x 4 x float>, <vscale x 4 x float>} %res
}

declare {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv16i8(i32, <vscale x 16 x i8>, i32)
declare {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8i16(i32, <vscale x 16 x i8>, i32)
declare {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv4i32(i32, <vscale x 16 x i8>, i32)
declare {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8f16(i32, <vscale x 16 x i8>, i32)
declare {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv8bf16(i32, <vscale x 16 x i8>, i32)
declare {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.luti2.lane.zt.x2.nxv4f32(i32, <vscale x 16 x i8>, i32)