1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve2 < %s | FileCheck %s
3 ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme -force-streaming < %s | FileCheck %s
9 define <vscale x 4 x i32> @cdot_s(<vscale x 4 x i32> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) {
10 ; CHECK-LABEL: cdot_s:
12 ; CHECK-NEXT: cdot z0.s, z1.b, z2.b, #0
14 %out = call <vscale x 4 x i32> @llvm.aarch64.sve.cdot.nxv4i32(<vscale x 4 x i32> %a,
15 <vscale x 16 x i8> %b,
16 <vscale x 16 x i8> %c,
18 ret <vscale x 4 x i32> %out
21 define <vscale x 2 x i64> @cdot_d(<vscale x 2 x i64> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c) {
22 ; CHECK-LABEL: cdot_d:
24 ; CHECK-NEXT: cdot z0.d, z1.h, z2.h, #90
26 %out = call <vscale x 2 x i64> @llvm.aarch64.sve.cdot.nxv2i64(<vscale x 2 x i64> %a,
27 <vscale x 8 x i16> %b,
28 <vscale x 8 x i16> %c,
30 ret <vscale x 2 x i64> %out
37 define <vscale x 4 x i32> @cdot_s_idx(<vscale x 4 x i32> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) {
38 ; CHECK-LABEL: cdot_s_idx:
40 ; CHECK-NEXT: cdot z0.s, z1.b, z2.b[0], #180
42 %out = call <vscale x 4 x i32> @llvm.aarch64.sve.cdot.lane.nxv4i32(<vscale x 4 x i32> %a,
43 <vscale x 16 x i8> %b,
44 <vscale x 16 x i8> %c,
46 ret <vscale x 4 x i32> %out
49 define <vscale x 2 x i64> @cdot_d_idx(<vscale x 2 x i64> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c) {
50 ; CHECK-LABEL: cdot_d_idx:
52 ; CHECK-NEXT: cdot z0.d, z1.h, z2.h[1], #270
54 %out = call <vscale x 2 x i64> @llvm.aarch64.sve.cdot.lane.nxv2i64(<vscale x 2 x i64> %a,
55 <vscale x 8 x i16> %b,
56 <vscale x 8 x i16> %c,
58 ret <vscale x 2 x i64> %out
61 declare <vscale x 4 x i32> @llvm.aarch64.sve.cdot.nxv4i32(<vscale x 4 x i32>, <vscale x 16 x i8>, <vscale x 16 x i8>, i32)
62 declare <vscale x 2 x i64> @llvm.aarch64.sve.cdot.nxv2i64(<vscale x 2 x i64>, <vscale x 8 x i16>, <vscale x 8 x i16>, i32)
63 declare <vscale x 4 x i32> @llvm.aarch64.sve.cdot.lane.nxv4i32(<vscale x 4 x i32>, <vscale x 16 x i8>, <vscale x 16 x i8>, i32, i32)
64 declare <vscale x 2 x i64> @llvm.aarch64.sve.cdot.lane.nxv2i64(<vscale x 2 x i64>, <vscale x 8 x i16>, <vscale x 8 x i16>, i32, i32)