1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mattr=+sme-fa64 -force-streaming-compatible-sve < %s | FileCheck %s -check-prefix=FA64
3 ; RUN: llc -mattr=+sve -force-streaming-compatible-sve < %s | FileCheck %s -check-prefix=NO-FA64
6 target triple = "aarch64-unknown-linux-gnu"
8 define half @fadda_v4f16(half %start, <4 x half> %a) {
9 ; FA64-LABEL: fadda_v4f16:
11 ; FA64-NEXT: ptrue p0.h, vl4
12 ; FA64-NEXT: // kill: def $h0 killed $h0 def $z0
13 ; FA64-NEXT: // kill: def $d1 killed $d1 def $z1
14 ; FA64-NEXT: fadda h0, p0, h0, z1.h
15 ; FA64-NEXT: // kill: def $h0 killed $h0 killed $z0
18 ; NO-FA64-LABEL: fadda_v4f16:
20 ; NO-FA64-NEXT: // kill: def $d1 killed $d1 def $z1
21 ; NO-FA64-NEXT: fadd h0, h0, h1
22 ; NO-FA64-NEXT: mov z2.h, z1.h[1]
23 ; NO-FA64-NEXT: fadd h0, h0, h2
24 ; NO-FA64-NEXT: mov z2.h, z1.h[2]
25 ; NO-FA64-NEXT: mov z1.h, z1.h[3]
26 ; NO-FA64-NEXT: fadd h0, h0, h2
27 ; NO-FA64-NEXT: fadd h0, h0, h1
29 %res = call half @llvm.vector.reduce.fadd.v4f16(half %start, <4 x half> %a)
33 declare half @llvm.vector.reduce.fadd.v4f16(half, <4 x half>)