1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
2 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme2 -force-streaming < %s | FileCheck %s
4 ; lookup table expand one register
6 define <vscale x 16 x i8> @luti2_i8(<vscale x 16 x i8> %x) {
7 ; CHECK-LABEL: luti2_i8:
9 ; CHECK-NEXT: luti2 z0.b, zt0, z0[15]
11 %res = call <vscale x 16 x i8> @llvm.aarch64.sme.luti2.lane.zt.nxv16i8(i32 0, <vscale x 16 x i8> %x, i32 15)
12 ret <vscale x 16 x i8> %res
15 define <vscale x 8 x i16> @luti2_i16(<vscale x 16 x i8> %x) {
16 ; CHECK-LABEL: luti2_i16:
18 ; CHECK-NEXT: luti2 z0.h, zt0, z0[15]
20 %res = call <vscale x 8 x i16> @llvm.aarch64.sme.luti2.lane.zt.nxv8i16(i32 0, <vscale x 16 x i8> %x, i32 15)
21 ret <vscale x 8 x i16> %res
24 define <vscale x 4 x i32> @luti2_i32(<vscale x 16 x i8> %x) {
25 ; CHECK-LABEL: luti2_i32:
27 ; CHECK-NEXT: luti2 z0.s, zt0, z0[15]
29 %res = call <vscale x 4 x i32> @llvm.aarch64.sme.luti2.lane.zt.nxv4i32(i32 0, <vscale x 16 x i8> %x, i32 15)
30 ret <vscale x 4 x i32> %res
33 define <vscale x 8 x half> @luti2_f16(<vscale x 16 x i8> %x) {
34 ; CHECK-LABEL: luti2_f16:
36 ; CHECK-NEXT: luti2 z0.h, zt0, z0[15]
38 %res = call <vscale x 8 x half> @llvm.aarch64.sme.luti2.lane.zt.nxv8f16(i32 0, <vscale x 16 x i8> %x, i32 15)
39 ret <vscale x 8 x half> %res
42 define <vscale x 8 x bfloat> @luti2_bf16(<vscale x 16 x i8> %x) {
43 ; CHECK-LABEL: luti2_bf16:
45 ; CHECK-NEXT: luti2 z0.h, zt0, z0[15]
47 %res = call <vscale x 8 x bfloat> @llvm.aarch64.sme.luti2.lane.zt.nxv8bf16(i32 0, <vscale x 16 x i8> %x, i32 15)
48 ret <vscale x 8 x bfloat> %res
51 define <vscale x 4 x float> @luti2_f32(<vscale x 16 x i8> %x) {
52 ; CHECK-LABEL: luti2_f32:
54 ; CHECK-NEXT: luti2 z0.s, zt0, z0[15]
56 %res = call <vscale x 4 x float> @llvm.aarch64.sme.luti2.lane.zt.nxv4f32(i32 0, <vscale x 16 x i8> %x, i32 15)
57 ret <vscale x 4 x float> %res
60 declare <vscale x 16 x i8> @llvm.aarch64.sme.luti2.lane.zt.nxv16i8(i32, <vscale x 16 x i8>, i32)
61 declare <vscale x 8 x i16> @llvm.aarch64.sme.luti2.lane.zt.nxv8i16(i32, <vscale x 16 x i8>, i32)
62 declare <vscale x 4 x i32> @llvm.aarch64.sme.luti2.lane.zt.nxv4i32(i32, <vscale x 16 x i8>, i32)
63 declare <vscale x 8 x half> @llvm.aarch64.sme.luti2.lane.zt.nxv8f16(i32, <vscale x 16 x i8>, i32)
64 declare <vscale x 8 x bfloat> @llvm.aarch64.sme.luti2.lane.zt.nxv8bf16(i32, <vscale x 16 x i8>, i32)
65 declare <vscale x 4 x float> @llvm.aarch64.sme.luti2.lane.zt.nxv4f32(i32, <vscale x 16 x i8>, i32)