mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 11:13:28 +01:00
1050fc5c47
Summary: Implements the following intrinsics: - @llvm.aarch64.sve.cdot - @llvm.aarch64.sve.cdot.lane Reviewers: sdesmalen, efriedma, dancgr, c-rhodes, rengolin Reviewed By: efriedma Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, cfe-commits, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D73687
62 lines
2.8 KiB
LLVM
62 lines
2.8 KiB
LLVM
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve2 < %s | FileCheck %s
|
|
|
|
|
|
;
|
|
; CDOT
|
|
;
|
|
|
|
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) {
|
|
; CHECK-LABEL: cdot_s:
|
|
; CHECK: cdot z0.s, z1.b, z2.b, #0
|
|
; CHECK-NEXT: ret
|
|
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.cdot.nxv4i32(<vscale x 4 x i32> %a,
|
|
<vscale x 16 x i8> %b,
|
|
<vscale x 16 x i8> %c,
|
|
i32 0)
|
|
ret <vscale x 4 x i32> %out
|
|
}
|
|
|
|
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) {
|
|
; CHECK-LABEL: cdot_d:
|
|
; CHECK: cdot z0.d, z1.h, z2.h, #90
|
|
; CHECK-NEXT: ret
|
|
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.cdot.nxv2i64(<vscale x 2 x i64> %a,
|
|
<vscale x 8 x i16> %b,
|
|
<vscale x 8 x i16> %c,
|
|
i32 90)
|
|
ret <vscale x 2 x i64> %out
|
|
}
|
|
|
|
;
|
|
; CDOT(indexed)
|
|
;
|
|
|
|
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) {
|
|
; CHECK-LABEL: cdot_s_idx:
|
|
; CHECK: cdot z0.s, z1.b, z2.b[0], #180
|
|
; CHECK-NEXT: ret
|
|
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.cdot.lane.nxv4i32(<vscale x 4 x i32> %a,
|
|
<vscale x 16 x i8> %b,
|
|
<vscale x 16 x i8> %c,
|
|
i32 0, i32 180)
|
|
ret <vscale x 4 x i32> %out
|
|
}
|
|
|
|
|
|
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) {
|
|
; CHECK-LABEL: cdot_d_idx:
|
|
; CHECK: cdot z0.d, z1.h, z2.h[1], #270
|
|
; CHECK-NEXT: ret
|
|
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.cdot.lane.nxv2i64(<vscale x 2 x i64> %a,
|
|
<vscale x 8 x i16> %b,
|
|
<vscale x 8 x i16> %c,
|
|
i32 1, i32 270)
|
|
ret <vscale x 2 x i64> %out
|
|
}
|
|
|
|
|
|
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)
|
|
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)
|
|
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)
|
|
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)
|