RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-02-01 05:01:59 +01:00
Code Issues Projects Releases Wiki Activity

[SVE][NFC] Regenerate a few CodeGen tests

Browse Source 
Regenerated using llvm/utils/update_llc_test_checks.py as part of
D94504, committing separately to reduce the diff for D94504.
This commit is contained in:
Cullen Rhodes 2021-01-12 17:48:52 +00:00
parent ce76c6de45
commit b30ad48824
3 changed files with 311 additions and 219 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

229
test/CodeGen/AArch64/sve-intrinsics-loads-nf.ll
View File

@ -1,3 +1,4 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s 2>%t | FileCheck %s ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s 2>%t | FileCheck %s
; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t ; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t
@ -10,18 +11,20 @@
define <vscale x 16 x i8> @ldnf1b(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b: ; CHECK-LABEL: ldnf1b:
; CHECK: ldnf1b { z0.b }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 16 x i8> @llvm.aarch64.sve.ldnf1.nxv16i8(<vscale x 16 x i1> %pg, i8* %a) %load = call <vscale x 16 x i8> @llvm.aarch64.sve.ldnf1.nxv16i8(<vscale x 16 x i1> %pg, i8* %a)
ret <vscale x 16 x i8> %load ret <vscale x 16 x i8> %load
} }
define <vscale x 16 x i8> @ldnf1b_out_of_lower_bound(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b_out_of_lower_bound(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_out_of_lower_bound: ; CHECK-LABEL: ldnf1b_out_of_lower_bound:
; CHECK: rdvl x[[OFFSET:[0-9]+]], #-9 ; CHECK: // %bb.0:
; CHECK-NEXT: add x[[BASE:[0-9]+]], x0, x[[OFFSET]] ; CHECK-NEXT: rdvl x8, #-9
; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x[[BASE]]] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x8]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 16 x i8>* %base_scalable = bitcast i8* %a to <vscale x 16 x i8>*
%base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 -9 %base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 -9
%base_scalar = bitcast <vscale x 16 x i8>* %base to i8* %base_scalar = bitcast <vscale x 16 x i8>* %base to i8*
@ -31,8 +34,9 @@ define <vscale x 16 x i8> @ldnf1b_out_of_lower_bound(<vscale x 16 x i1> %pg, i8*
define <vscale x 16 x i8> @ldnf1b_lower_bound(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b_lower_bound(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_lower_bound: ; CHECK-LABEL: ldnf1b_lower_bound:
; CHECK: ldnf1b { z0.b }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 16 x i8>* %base_scalable = bitcast i8* %a to <vscale x 16 x i8>*
%base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 -8 %base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 -8
%base_scalar = bitcast <vscale x 16 x i8>* %base to i8* %base_scalar = bitcast <vscale x 16 x i8>* %base to i8*
@ -42,8 +46,9 @@ define <vscale x 16 x i8> @ldnf1b_lower_bound(<vscale x 16 x i1> %pg, i8* %a) {
define <vscale x 16 x i8> @ldnf1b_inbound(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b_inbound(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_inbound: ; CHECK-LABEL: ldnf1b_inbound:
; CHECK: ldnf1b { z0.b }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 16 x i8>* %base_scalable = bitcast i8* %a to <vscale x 16 x i8>*
%base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 1 %base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 16 x i8>* %base to i8* %base_scalar = bitcast <vscale x 16 x i8>* %base to i8*
@ -53,8 +58,9 @@ define <vscale x 16 x i8> @ldnf1b_inbound(<vscale x 16 x i1> %pg, i8* %a) {
define <vscale x 16 x i8> @ldnf1b_upper_bound(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b_upper_bound(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_upper_bound: ; CHECK-LABEL: ldnf1b_upper_bound:
; CHECK: ldnf1b { z0.b }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 16 x i8>* %base_scalable = bitcast i8* %a to <vscale x 16 x i8>*
%base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 16 x i8>* %base to i8* %base_scalar = bitcast <vscale x 16 x i8>* %base to i8*
@ -64,10 +70,11 @@ define <vscale x 16 x i8> @ldnf1b_upper_bound(<vscale x 16 x i1> %pg, i8* %a) {
define <vscale x 16 x i8> @ldnf1b_out_of_upper_bound(<vscale x 16 x i1> %pg, i8* %a) { define <vscale x 16 x i8> @ldnf1b_out_of_upper_bound(<vscale x 16 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_out_of_upper_bound: ; CHECK-LABEL: ldnf1b_out_of_upper_bound:
; CHECK: rdvl x[[OFFSET:[0-9]+]], #8 ; CHECK: // %bb.0:
; CHECK-NEXT: add x[[BASE:[0-9]+]], x0, x[[OFFSET]] ; CHECK-NEXT: rdvl x8, #8
; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x[[BASE]]] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.b }, p0/z, [x8]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 16 x i8>* %base_scalable = bitcast i8* %a to <vscale x 16 x i8>*
%base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 8 %base = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base_scalable, i64 8
%base_scalar = bitcast <vscale x 16 x i8>* %base to i8* %base_scalar = bitcast <vscale x 16 x i8>* %base to i8*
@ -77,8 +84,9 @@ define <vscale x 16 x i8> @ldnf1b_out_of_upper_bound(<vscale x 16 x i1> %pg, i8*
define <vscale x 8 x i16> @ldnf1b_h(<vscale x 8 x i1> %pg, i8* %a) { define <vscale x 8 x i16> @ldnf1b_h(<vscale x 8 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_h: ; CHECK-LABEL: ldnf1b_h:
; CHECK: ldnf1b { z0.h }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 8 x i8> @llvm.aarch64.sve.ldnf1.nxv8i8(<vscale x 8 x i1> %pg, i8* %a) %load = call <vscale x 8 x i8> @llvm.aarch64.sve.ldnf1.nxv8i8(<vscale x 8 x i1> %pg, i8* %a)
%res = zext <vscale x 8 x i8> %load to <vscale x 8 x i16> %res = zext <vscale x 8 x i8> %load to <vscale x 8 x i16>
ret <vscale x 8 x i16> %res ret <vscale x 8 x i16> %res
@ -86,8 +94,9 @@ define <vscale x 8 x i16> @ldnf1b_h(<vscale x 8 x i1> %pg, i8* %a) {
define <vscale x 8 x i16> @ldnf1b_h_inbound(<vscale x 8 x i1> %pg, i8* %a) { define <vscale x 8 x i16> @ldnf1b_h_inbound(<vscale x 8 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_h_inbound: ; CHECK-LABEL: ldnf1b_h_inbound:
; CHECK: ldnf1b { z0.h }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.h }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 8 x i8>* %base_scalable = bitcast i8* %a to <vscale x 8 x i8>*
%base = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 8 x i8>* %base to i8* %base_scalar = bitcast <vscale x 8 x i8>* %base to i8*
@ -98,8 +107,9 @@ define <vscale x 8 x i16> @ldnf1b_h_inbound(<vscale x 8 x i1> %pg, i8* %a) {
define <vscale x 8 x i16> @ldnf1sb_h(<vscale x 8 x i1> %pg, i8* %a) { define <vscale x 8 x i16> @ldnf1sb_h(<vscale x 8 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_h: ; CHECK-LABEL: ldnf1sb_h:
; CHECK: ldnf1sb { z0.h }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 8 x i8> @llvm.aarch64.sve.ldnf1.nxv8i8(<vscale x 8 x i1> %pg, i8* %a) %load = call <vscale x 8 x i8> @llvm.aarch64.sve.ldnf1.nxv8i8(<vscale x 8 x i1> %pg, i8* %a)
%res = sext <vscale x 8 x i8> %load to <vscale x 8 x i16> %res = sext <vscale x 8 x i8> %load to <vscale x 8 x i16>
ret <vscale x 8 x i16> %res ret <vscale x 8 x i16> %res
@ -107,8 +117,9 @@ define <vscale x 8 x i16> @ldnf1sb_h(<vscale x 8 x i1> %pg, i8* %a) {
define <vscale x 8 x i16> @ldnf1sb_h_inbound(<vscale x 8 x i1> %pg, i8* %a) { define <vscale x 8 x i16> @ldnf1sb_h_inbound(<vscale x 8 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_h_inbound: ; CHECK-LABEL: ldnf1sb_h_inbound:
; CHECK: ldnf1sb { z0.h }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.h }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 8 x i8>* %base_scalable = bitcast i8* %a to <vscale x 8 x i8>*
%base = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 8 x i8>* %base to i8* %base_scalar = bitcast <vscale x 8 x i8>* %base to i8*
@ -119,16 +130,18 @@ define <vscale x 8 x i16> @ldnf1sb_h_inbound(<vscale x 8 x i1> %pg, i8* %a) {
define <vscale x 8 x i16> @ldnf1h(<vscale x 8 x i1> %pg, i16* %a) { define <vscale x 8 x i16> @ldnf1h(<vscale x 8 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h: ; CHECK-LABEL: ldnf1h:
; CHECK: ldnf1h { z0.h }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 8 x i16> @llvm.aarch64.sve.ldnf1.nxv8i16(<vscale x 8 x i1> %pg, i16* %a) %load = call <vscale x 8 x i16> @llvm.aarch64.sve.ldnf1.nxv8i16(<vscale x 8 x i1> %pg, i16* %a)
ret <vscale x 8 x i16> %load ret <vscale x 8 x i16> %load
} }
define <vscale x 8 x i16> @ldnf1h_inbound(<vscale x 8 x i1> %pg, i16* %a) { define <vscale x 8 x i16> @ldnf1h_inbound(<vscale x 8 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h_inbound: ; CHECK-LABEL: ldnf1h_inbound:
; CHECK: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i16* %a to <vscale x 8 x i16>* %base_scalable = bitcast i16* %a to <vscale x 8 x i16>*
%base = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base_scalable, i64 1 %base = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 8 x i16>* %base to i16* %base_scalar = bitcast <vscale x 8 x i16>* %base to i16*
@ -138,24 +151,27 @@ define <vscale x 8 x i16> @ldnf1h_inbound(<vscale x 8 x i1> %pg, i16* %a) {
define <vscale x 8 x half> @ldnf1h_f16(<vscale x 8 x i1> %pg, half* %a) { define <vscale x 8 x half> @ldnf1h_f16(<vscale x 8 x i1> %pg, half* %a) {
; CHECK-LABEL: ldnf1h_f16: ; CHECK-LABEL: ldnf1h_f16:
; CHECK: ldnf1h { z0.h }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 8 x half> @llvm.aarch64.sve.ldnf1.nxv8f16(<vscale x 8 x i1> %pg, half* %a) %load = call <vscale x 8 x half> @llvm.aarch64.sve.ldnf1.nxv8f16(<vscale x 8 x i1> %pg, half* %a)
ret <vscale x 8 x half> %load ret <vscale x 8 x half> %load
} }
define <vscale x 8 x bfloat> @ldnf1h_bf16(<vscale x 8 x i1> %pg, bfloat* %a) #0 { define <vscale x 8 x bfloat> @ldnf1h_bf16(<vscale x 8 x i1> %pg, bfloat* %a) #0 {
; CHECK-LABEL: ldnf1h_bf16: ; CHECK-LABEL: ldnf1h_bf16:
; CHECK: ldnf1h { z0.h }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnf1.nxv8bf16(<vscale x 8 x i1> %pg, bfloat* %a) %load = call <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnf1.nxv8bf16(<vscale x 8 x i1> %pg, bfloat* %a)
ret <vscale x 8 x bfloat> %load ret <vscale x 8 x bfloat> %load
} }
define <vscale x 8 x half> @ldnf1h_f16_inbound(<vscale x 8 x i1> %pg, half* %a) { define <vscale x 8 x half> @ldnf1h_f16_inbound(<vscale x 8 x i1> %pg, half* %a) {
; CHECK-LABEL: ldnf1h_f16_inbound: ; CHECK-LABEL: ldnf1h_f16_inbound:
; CHECK: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast half* %a to <vscale x 8 x half>* %base_scalable = bitcast half* %a to <vscale x 8 x half>*
%base = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base_scalable, i64 1 %base = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 8 x half>* %base to half* %base_scalar = bitcast <vscale x 8 x half>* %base to half*
@ -165,8 +181,9 @@ define <vscale x 8 x half> @ldnf1h_f16_inbound(<vscale x 8 x i1> %pg, half* %a)
define <vscale x 8 x bfloat> @ldnf1h_bf16_inbound(<vscale x 8 x i1> %pg, bfloat* %a) #0 { define <vscale x 8 x bfloat> @ldnf1h_bf16_inbound(<vscale x 8 x i1> %pg, bfloat* %a) #0 {
; CHECK-LABEL: ldnf1h_bf16_inbound: ; CHECK-LABEL: ldnf1h_bf16_inbound:
; CHECK: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.h }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast bfloat* %a to <vscale x 8 x bfloat>* %base_scalable = bitcast bfloat* %a to <vscale x 8 x bfloat>*
%base = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base_scalable, i64 1 %base = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 8 x bfloat>* %base to bfloat* %base_scalar = bitcast <vscale x 8 x bfloat>* %base to bfloat*
@ -176,8 +193,9 @@ define <vscale x 8 x bfloat> @ldnf1h_bf16_inbound(<vscale x 8 x i1> %pg, bfloat*
define <vscale x 4 x i32> @ldnf1b_s(<vscale x 4 x i1> %pg, i8* %a) { define <vscale x 4 x i32> @ldnf1b_s(<vscale x 4 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_s: ; CHECK-LABEL: ldnf1b_s:
; CHECK: ldnf1b { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x i8> @llvm.aarch64.sve.ldnf1.nxv4i8(<vscale x 4 x i1> %pg, i8* %a) %load = call <vscale x 4 x i8> @llvm.aarch64.sve.ldnf1.nxv4i8(<vscale x 4 x i1> %pg, i8* %a)
%res = zext <vscale x 4 x i8> %load to <vscale x 4 x i32> %res = zext <vscale x 4 x i8> %load to <vscale x 4 x i32>
ret <vscale x 4 x i32> %res ret <vscale x 4 x i32> %res
@ -185,8 +203,9 @@ define <vscale x 4 x i32> @ldnf1b_s(<vscale x 4 x i1> %pg, i8* %a) {
define <vscale x 4 x i32> @ldnf1b_s_inbound(<vscale x 4 x i1> %pg, i8* %a) { define <vscale x 4 x i32> @ldnf1b_s_inbound(<vscale x 4 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_s_inbound: ; CHECK-LABEL: ldnf1b_s_inbound:
; CHECK: ldnf1b { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 4 x i8>* %base_scalable = bitcast i8* %a to <vscale x 4 x i8>*
%base = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x i8>* %base to i8* %base_scalar = bitcast <vscale x 4 x i8>* %base to i8*
@ -197,8 +216,9 @@ define <vscale x 4 x i32> @ldnf1b_s_inbound(<vscale x 4 x i1> %pg, i8* %a) {
define <vscale x 4 x i32> @ldnf1sb_s(<vscale x 4 x i1> %pg, i8* %a) { define <vscale x 4 x i32> @ldnf1sb_s(<vscale x 4 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_s: ; CHECK-LABEL: ldnf1sb_s:
; CHECK: ldnf1sb { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x i8> @llvm.aarch64.sve.ldnf1.nxv4i8(<vscale x 4 x i1> %pg, i8* %a) %load = call <vscale x 4 x i8> @llvm.aarch64.sve.ldnf1.nxv4i8(<vscale x 4 x i1> %pg, i8* %a)
%res = sext <vscale x 4 x i8> %load to <vscale x 4 x i32> %res = sext <vscale x 4 x i8> %load to <vscale x 4 x i32>
ret <vscale x 4 x i32> %res ret <vscale x 4 x i32> %res
@ -206,8 +226,9 @@ define <vscale x 4 x i32> @ldnf1sb_s(<vscale x 4 x i1> %pg, i8* %a) {
define <vscale x 4 x i32> @ldnf1sb_s_inbound(<vscale x 4 x i1> %pg, i8* %a) { define <vscale x 4 x i32> @ldnf1sb_s_inbound(<vscale x 4 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_s_inbound: ; CHECK-LABEL: ldnf1sb_s_inbound:
; CHECK: ldnf1sb { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 4 x i8>* %base_scalable = bitcast i8* %a to <vscale x 4 x i8>*
%base = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x i8>* %base to i8* %base_scalar = bitcast <vscale x 4 x i8>* %base to i8*
@ -218,8 +239,9 @@ define <vscale x 4 x i32> @ldnf1sb_s_inbound(<vscale x 4 x i1> %pg, i8* %a) {
define <vscale x 4 x i32> @ldnf1h_s(<vscale x 4 x i1> %pg, i16* %a) { define <vscale x 4 x i32> @ldnf1h_s(<vscale x 4 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h_s: ; CHECK-LABEL: ldnf1h_s:
; CHECK: ldnf1h { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> %pg, i16* %a) %load = call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> %pg, i16* %a)
%res = zext <vscale x 4 x i16> %load to <vscale x 4 x i32> %res = zext <vscale x 4 x i16> %load to <vscale x 4 x i32>
ret <vscale x 4 x i32> %res ret <vscale x 4 x i32> %res
@ -227,8 +249,9 @@ define <vscale x 4 x i32> @ldnf1h_s(<vscale x 4 x i1> %pg, i16* %a) {
define <vscale x 4 x i32> @ldnf1h_s_inbound(<vscale x 4 x i1> %pg, i16* %a) { define <vscale x 4 x i32> @ldnf1h_s_inbound(<vscale x 4 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h_s_inbound: ; CHECK-LABEL: ldnf1h_s_inbound:
; CHECK: ldnf1h { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i16* %a to <vscale x 4 x i16>* %base_scalable = bitcast i16* %a to <vscale x 4 x i16>*
%base = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x i16>* %base to i16* %base_scalar = bitcast <vscale x 4 x i16>* %base to i16*
@ -239,8 +262,9 @@ define <vscale x 4 x i32> @ldnf1h_s_inbound(<vscale x 4 x i1> %pg, i16* %a) {
define <vscale x 4 x i32> @ldnf1sh_s(<vscale x 4 x i1> %pg, i16* %a) { define <vscale x 4 x i32> @ldnf1sh_s(<vscale x 4 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1sh_s: ; CHECK-LABEL: ldnf1sh_s:
; CHECK: ldnf1sh { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sh { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> %pg, i16* %a) %load = call <vscale x 4 x i16> @llvm.aarch64.sve.ldnf1.nxv4i16(<vscale x 4 x i1> %pg, i16* %a)
%res = sext <vscale x 4 x i16> %load to <vscale x 4 x i32> %res = sext <vscale x 4 x i16> %load to <vscale x 4 x i32>
ret <vscale x 4 x i32> %res ret <vscale x 4 x i32> %res
@ -248,8 +272,9 @@ define <vscale x 4 x i32> @ldnf1sh_s(<vscale x 4 x i1> %pg, i16* %a) {
define <vscale x 4 x i32> @ldnf1sh_s_inbound(<vscale x 4 x i1> %pg, i16* %a) { define <vscale x 4 x i32> @ldnf1sh_s_inbound(<vscale x 4 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1sh_s_inbound: ; CHECK-LABEL: ldnf1sh_s_inbound:
; CHECK: ldnf1sh { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sh { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i16* %a to <vscale x 4 x i16>* %base_scalable = bitcast i16* %a to <vscale x 4 x i16>*
%base = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x i16>* %base to i16* %base_scalar = bitcast <vscale x 4 x i16>* %base to i16*
@ -260,16 +285,18 @@ define <vscale x 4 x i32> @ldnf1sh_s_inbound(<vscale x 4 x i1> %pg, i16* %a) {
define <vscale x 4 x i32> @ldnf1w(<vscale x 4 x i1> %pg, i32* %a) { define <vscale x 4 x i32> @ldnf1w(<vscale x 4 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1w: ; CHECK-LABEL: ldnf1w:
; CHECK: ldnf1w { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x i32> @llvm.aarch64.sve.ldnf1.nxv4i32(<vscale x 4 x i1> %pg, i32* %a) %load = call <vscale x 4 x i32> @llvm.aarch64.sve.ldnf1.nxv4i32(<vscale x 4 x i1> %pg, i32* %a)
ret <vscale x 4 x i32> %load ret <vscale x 4 x i32> %load
} }
define <vscale x 4 x i32> @ldnf1w_inbound(<vscale x 4 x i1> %pg, i32* %a) { define <vscale x 4 x i32> @ldnf1w_inbound(<vscale x 4 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1w_inbound: ; CHECK-LABEL: ldnf1w_inbound:
; CHECK: ldnf1w { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i32* %a to <vscale x 4 x i32>* %base_scalable = bitcast i32* %a to <vscale x 4 x i32>*
%base = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x i32>* %base to i32* %base_scalar = bitcast <vscale x 4 x i32>* %base to i32*
@ -279,16 +306,18 @@ define <vscale x 4 x i32> @ldnf1w_inbound(<vscale x 4 x i1> %pg, i32* %a) {
define <vscale x 4 x float> @ldnf1w_f32(<vscale x 4 x i1> %pg, float* %a) { define <vscale x 4 x float> @ldnf1w_f32(<vscale x 4 x i1> %pg, float* %a) {
; CHECK-LABEL: ldnf1w_f32: ; CHECK-LABEL: ldnf1w_f32:
; CHECK: ldnf1w { z0.s }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.s }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 4 x float> @llvm.aarch64.sve.ldnf1.nxv4f32(<vscale x 4 x i1> %pg, float* %a) %load = call <vscale x 4 x float> @llvm.aarch64.sve.ldnf1.nxv4f32(<vscale x 4 x i1> %pg, float* %a)
ret <vscale x 4 x float> %load ret <vscale x 4 x float> %load
} }
define <vscale x 4 x float> @ldnf1w_f32_inbound(<vscale x 4 x i1> %pg, float* %a) { define <vscale x 4 x float> @ldnf1w_f32_inbound(<vscale x 4 x i1> %pg, float* %a) {
; CHECK-LABEL: ldnf1w_f32_inbound: ; CHECK-LABEL: ldnf1w_f32_inbound:
; CHECK: ldnf1w { z0.s }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.s }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast float* %a to <vscale x 4 x float>* %base_scalable = bitcast float* %a to <vscale x 4 x float>*
%base = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base_scalable, i64 7 %base = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 4 x float>* %base to float* %base_scalar = bitcast <vscale x 4 x float>* %base to float*
@ -298,8 +327,9 @@ define <vscale x 4 x float> @ldnf1w_f32_inbound(<vscale x 4 x i1> %pg, float* %a
define <vscale x 2 x i64> @ldnf1b_d(<vscale x 2 x i1> %pg, i8* %a) { define <vscale x 2 x i64> @ldnf1b_d(<vscale x 2 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_d: ; CHECK-LABEL: ldnf1b_d:
; CHECK: ldnf1b { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i8> @llvm.aarch64.sve.ldnf1.nxv2i8(<vscale x 2 x i1> %pg, i8* %a) %load = call <vscale x 2 x i8> @llvm.aarch64.sve.ldnf1.nxv2i8(<vscale x 2 x i1> %pg, i8* %a)
%res = zext <vscale x 2 x i8> %load to <vscale x 2 x i64> %res = zext <vscale x 2 x i8> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -307,8 +337,9 @@ define <vscale x 2 x i64> @ldnf1b_d(<vscale x 2 x i1> %pg, i8* %a) {
define <vscale x 2 x i64> @ldnf1b_d_inbound(<vscale x 2 x i1> %pg, i8* %a) { define <vscale x 2 x i64> @ldnf1b_d_inbound(<vscale x 2 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1b_d_inbound: ; CHECK-LABEL: ldnf1b_d_inbound:
; CHECK: ldnf1b { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1b { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 2 x i8>* %base_scalable = bitcast i8* %a to <vscale x 2 x i8>*
%base = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i8>* %base to i8* %base_scalar = bitcast <vscale x 2 x i8>* %base to i8*
@ -319,8 +350,9 @@ define <vscale x 2 x i64> @ldnf1b_d_inbound(<vscale x 2 x i1> %pg, i8* %a) {
define <vscale x 2 x i64> @ldnf1sb_d(<vscale x 2 x i1> %pg, i8* %a) { define <vscale x 2 x i64> @ldnf1sb_d(<vscale x 2 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_d: ; CHECK-LABEL: ldnf1sb_d:
; CHECK: ldnf1sb { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i8> @llvm.aarch64.sve.ldnf1.nxv2i8(<vscale x 2 x i1> %pg, i8* %a) %load = call <vscale x 2 x i8> @llvm.aarch64.sve.ldnf1.nxv2i8(<vscale x 2 x i1> %pg, i8* %a)
%res = sext <vscale x 2 x i8> %load to <vscale x 2 x i64> %res = sext <vscale x 2 x i8> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -328,8 +360,9 @@ define <vscale x 2 x i64> @ldnf1sb_d(<vscale x 2 x i1> %pg, i8* %a) {
define <vscale x 2 x i64> @ldnf1sb_d_inbound(<vscale x 2 x i1> %pg, i8* %a) { define <vscale x 2 x i64> @ldnf1sb_d_inbound(<vscale x 2 x i1> %pg, i8* %a) {
; CHECK-LABEL: ldnf1sb_d_inbound: ; CHECK-LABEL: ldnf1sb_d_inbound:
; CHECK: ldnf1sb { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sb { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i8* %a to <vscale x 2 x i8>* %base_scalable = bitcast i8* %a to <vscale x 2 x i8>*
%base = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i8>* %base to i8* %base_scalar = bitcast <vscale x 2 x i8>* %base to i8*
@ -340,8 +373,9 @@ define <vscale x 2 x i64> @ldnf1sb_d_inbound(<vscale x 2 x i1> %pg, i8* %a) {
define <vscale x 2 x i64> @ldnf1h_d(<vscale x 2 x i1> %pg, i16* %a) { define <vscale x 2 x i64> @ldnf1h_d(<vscale x 2 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h_d: ; CHECK-LABEL: ldnf1h_d:
; CHECK: ldnf1h { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> %pg, i16* %a) %load = call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> %pg, i16* %a)
%res = zext <vscale x 2 x i16> %load to <vscale x 2 x i64> %res = zext <vscale x 2 x i16> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -349,8 +383,9 @@ define <vscale x 2 x i64> @ldnf1h_d(<vscale x 2 x i1> %pg, i16* %a) {
define <vscale x 2 x i64> @ldnf1h_d_inbound(<vscale x 2 x i1> %pg, i16* %a) { define <vscale x 2 x i64> @ldnf1h_d_inbound(<vscale x 2 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1h_d_inbound: ; CHECK-LABEL: ldnf1h_d_inbound:
; CHECK: ldnf1h { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1h { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i16* %a to <vscale x 2 x i16>* %base_scalable = bitcast i16* %a to <vscale x 2 x i16>*
%base = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i16>* %base to i16* %base_scalar = bitcast <vscale x 2 x i16>* %base to i16*
@ -361,8 +396,9 @@ define <vscale x 2 x i64> @ldnf1h_d_inbound(<vscale x 2 x i1> %pg, i16* %a) {
define <vscale x 2 x i64> @ldnf1sh_d(<vscale x 2 x i1> %pg, i16* %a) { define <vscale x 2 x i64> @ldnf1sh_d(<vscale x 2 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1sh_d: ; CHECK-LABEL: ldnf1sh_d:
; CHECK: ldnf1sh { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sh { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> %pg, i16* %a) %load = call <vscale x 2 x i16> @llvm.aarch64.sve.ldnf1.nxv2i16(<vscale x 2 x i1> %pg, i16* %a)
%res = sext <vscale x 2 x i16> %load to <vscale x 2 x i64> %res = sext <vscale x 2 x i16> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -370,8 +406,9 @@ define <vscale x 2 x i64> @ldnf1sh_d(<vscale x 2 x i1> %pg, i16* %a) {
define <vscale x 2 x i64> @ldnf1sh_d_inbound(<vscale x 2 x i1> %pg, i16* %a) { define <vscale x 2 x i64> @ldnf1sh_d_inbound(<vscale x 2 x i1> %pg, i16* %a) {
; CHECK-LABEL: ldnf1sh_d_inbound: ; CHECK-LABEL: ldnf1sh_d_inbound:
; CHECK: ldnf1sh { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sh { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i16* %a to <vscale x 2 x i16>* %base_scalable = bitcast i16* %a to <vscale x 2 x i16>*
%base = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i16>* %base to i16* %base_scalar = bitcast <vscale x 2 x i16>* %base to i16*
@ -382,8 +419,9 @@ define <vscale x 2 x i64> @ldnf1sh_d_inbound(<vscale x 2 x i1> %pg, i16* %a) {
define <vscale x 2 x i64> @ldnf1w_d(<vscale x 2 x i1> %pg, i32* %a) { define <vscale x 2 x i64> @ldnf1w_d(<vscale x 2 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1w_d: ; CHECK-LABEL: ldnf1w_d:
; CHECK: ldnf1w { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i32> @llvm.aarch64.sve.ldnf1.nxv2i32(<vscale x 2 x i1> %pg, i32* %a) %load = call <vscale x 2 x i32> @llvm.aarch64.sve.ldnf1.nxv2i32(<vscale x 2 x i1> %pg, i32* %a)
%res = zext <vscale x 2 x i32> %load to <vscale x 2 x i64> %res = zext <vscale x 2 x i32> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -391,8 +429,9 @@ define <vscale x 2 x i64> @ldnf1w_d(<vscale x 2 x i1> %pg, i32* %a) {
define <vscale x 2 x i64> @ldnf1w_d_inbound(<vscale x 2 x i1> %pg, i32* %a) { define <vscale x 2 x i64> @ldnf1w_d_inbound(<vscale x 2 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1w_d_inbound: ; CHECK-LABEL: ldnf1w_d_inbound:
; CHECK: ldnf1w { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1w { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i32* %a to <vscale x 2 x i32>* %base_scalable = bitcast i32* %a to <vscale x 2 x i32>*
%base = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i32>* %base to i32* %base_scalar = bitcast <vscale x 2 x i32>* %base to i32*
@ -403,8 +442,9 @@ define <vscale x 2 x i64> @ldnf1w_d_inbound(<vscale x 2 x i1> %pg, i32* %a) {
define <vscale x 2 x i64> @ldnf1sw_d(<vscale x 2 x i1> %pg, i32* %a) { define <vscale x 2 x i64> @ldnf1sw_d(<vscale x 2 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1sw_d: ; CHECK-LABEL: ldnf1sw_d:
; CHECK: ldnf1sw { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sw { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i32> @llvm.aarch64.sve.ldnf1.nxv2i32(<vscale x 2 x i1> %pg, i32* %a) %load = call <vscale x 2 x i32> @llvm.aarch64.sve.ldnf1.nxv2i32(<vscale x 2 x i1> %pg, i32* %a)
%res = sext <vscale x 2 x i32> %load to <vscale x 2 x i64> %res = sext <vscale x 2 x i32> %load to <vscale x 2 x i64>
ret <vscale x 2 x i64> %res ret <vscale x 2 x i64> %res
@ -412,8 +452,9 @@ define <vscale x 2 x i64> @ldnf1sw_d(<vscale x 2 x i1> %pg, i32* %a) {
define <vscale x 2 x i64> @ldnf1sw_d_inbound(<vscale x 2 x i1> %pg, i32* %a) { define <vscale x 2 x i64> @ldnf1sw_d_inbound(<vscale x 2 x i1> %pg, i32* %a) {
; CHECK-LABEL: ldnf1sw_d_inbound: ; CHECK-LABEL: ldnf1sw_d_inbound:
; CHECK: ldnf1sw { z0.d }, p0/z, [x0, #7, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1sw { z0.d }, p0/z, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i32* %a to <vscale x 2 x i32>* %base_scalable = bitcast i32* %a to <vscale x 2 x i32>*
%base = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base_scalable, i64 7 %base = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base_scalable, i64 7
%base_scalar = bitcast <vscale x 2 x i32>* %base to i32* %base_scalar = bitcast <vscale x 2 x i32>* %base to i32*
@ -424,16 +465,18 @@ define <vscale x 2 x i64> @ldnf1sw_d_inbound(<vscale x 2 x i1> %pg, i32* %a) {
define <vscale x 2 x i64> @ldnf1d(<vscale x 2 x i1> %pg, i64* %a) { define <vscale x 2 x i64> @ldnf1d(<vscale x 2 x i1> %pg, i64* %a) {
; CHECK-LABEL: ldnf1d: ; CHECK-LABEL: ldnf1d:
; CHECK: ldnf1d { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1d { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnf1.nxv2i64(<vscale x 2 x i1> %pg, i64* %a) %load = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnf1.nxv2i64(<vscale x 2 x i1> %pg, i64* %a)
ret <vscale x 2 x i64> %load ret <vscale x 2 x i64> %load
} }
define <vscale x 2 x i64> @ldnf1d_inbound(<vscale x 2 x i1> %pg, i64* %a) { define <vscale x 2 x i64> @ldnf1d_inbound(<vscale x 2 x i1> %pg, i64* %a) {
; CHECK-LABEL: ldnf1d_inbound: ; CHECK-LABEL: ldnf1d_inbound:
; CHECK: ldnf1d { z0.d }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1d { z0.d }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast i64* %a to <vscale x 2 x i64>* %base_scalable = bitcast i64* %a to <vscale x 2 x i64>*
%base = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base_scalable, i64 1 %base = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 2 x i64>* %base to i64* %base_scalar = bitcast <vscale x 2 x i64>* %base to i64*
@ -443,16 +486,18 @@ define <vscale x 2 x i64> @ldnf1d_inbound(<vscale x 2 x i1> %pg, i64* %a) {
define <vscale x 2 x double> @ldnf1d_f64(<vscale x 2 x i1> %pg, double* %a) { define <vscale x 2 x double> @ldnf1d_f64(<vscale x 2 x i1> %pg, double* %a) {
; CHECK-LABEL: ldnf1d_f64: ; CHECK-LABEL: ldnf1d_f64:
; CHECK: ldnf1d { z0.d }, p0/z, [x0] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1d { z0.d }, p0/z, [x0]
; CHECK-NEXT: ret
%load = call <vscale x 2 x double> @llvm.aarch64.sve.ldnf1.nxv2f64(<vscale x 2 x i1> %pg, double* %a) %load = call <vscale x 2 x double> @llvm.aarch64.sve.ldnf1.nxv2f64(<vscale x 2 x i1> %pg, double* %a)
ret <vscale x 2 x double> %load ret <vscale x 2 x double> %load
} }
define <vscale x 2 x double> @ldnf1d_f64_inbound(<vscale x 2 x i1> %pg, double* %a) { define <vscale x 2 x double> @ldnf1d_f64_inbound(<vscale x 2 x i1> %pg, double* %a) {
; CHECK-LABEL: ldnf1d_f64_inbound: ; CHECK-LABEL: ldnf1d_f64_inbound:
; CHECK: ldnf1d { z0.d }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ldnf1d { z0.d }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_scalable = bitcast double* %a to <vscale x 2 x double>* %base_scalable = bitcast double* %a to <vscale x 2 x double>*
%base = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base_scalable, i64 1 %base = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base_scalable, i64 1
%base_scalar = bitcast <vscale x 2 x double>* %base to double* %base_scalar = bitcast <vscale x 2 x double>* %base to double*

227
test/CodeGen/AArch64/sve-pred-contiguous-ldst-addressing-mode-reg-imm.ll
View File

@ -1,4 +1,5 @@
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve --asm-verbose=false < %s 2>%t | FileCheck %s ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s 2>%t | FileCheck %s
; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t ; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t
; If this check fails please read test/CodeGen/AArch64/README for instructions on how to resolve it. ; If this check fails please read test/CodeGen/AArch64/README for instructions on how to resolve it.
@ -11,13 +12,14 @@
define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind { define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: imm_out_of_range: ; CHECK-LABEL: imm_out_of_range:
; CHECK-NEXT: rdvl x8, #8 ; CHECK: // %bb.0:
; CHECK-NEXT: add x8, x0, x8 ; CHECK-NEXT: rdvl x8, #8
; CHECK-NEXT: ld1d { z[[DATA:[0-9]+]].d }, p0/z, [x{{[0-9]+}}] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: rdvl x8, #-9 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x8]
; CHECK-NEXT: add x8, x0, x8 ; CHECK-NEXT: rdvl x8, #-9
; CHECK-NEXT: st1d { z[[DATA]].d }, p0, [x{{[0-9]+}}] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: ret ; CHECK-NEXT: st1d { z0.d }, p0, [x8]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 8 %base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 8
%data = call <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64>* %base_load, %data = call <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64>* %base_load,
i32 1, i32 1,
@ -35,9 +37,10 @@ define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mas
define void @test_masked_ldst_sv2i8(<vscale x 2 x i8> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2i8(<vscale x 2 x i8> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i8: ; CHECK-LABEL: test_masked_ldst_sv2i8:
; CHECK-NEXT: ld1sb { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1b { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1sb { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -8
%data = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load, %data = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load,
i32 1, i32 1,
@ -53,9 +56,10 @@ define void @test_masked_ldst_sv2i8(<vscale x 2 x i8> * %base, <vscale x 2 x i1>
define void @test_masked_ldst_sv2i16(<vscale x 2 x i16> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2i16(<vscale x 2 x i16> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i16: ; CHECK-LABEL: test_masked_ldst_sv2i16:
; CHECK-NEXT: ld1sh { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1sh { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 -8
%data = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load, %data = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load,
i32 1, i32 1,
@ -72,9 +76,10 @@ define void @test_masked_ldst_sv2i16(<vscale x 2 x i16> * %base, <vscale x 2 x i
define void @test_masked_ldst_sv2i32(<vscale x 2 x i32> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2i32(<vscale x 2 x i32> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i32: ; CHECK-LABEL: test_masked_ldst_sv2i32:
; CHECK-NEXT: ld1sw { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1w { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1sw { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1w { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -8
%data = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load, %data = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load,
i32 1, i32 1,
@ -90,9 +95,10 @@ define void @test_masked_ldst_sv2i32(<vscale x 2 x i32> * %base, <vscale x 2 x i
define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i64: ; CHECK-LABEL: test_masked_ldst_sv2i64:
; CHECK-NEXT: ld1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1d { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1d { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 -8
%data = call <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64>* %base_load, %data = call <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64>* %base_load,
i32 1, i32 1,
@ -108,9 +114,10 @@ define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i
define void @test_masked_ldst_sv2f16(<vscale x 2 x half> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2f16(<vscale x 2 x half> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2f16: ; CHECK-LABEL: test_masked_ldst_sv2f16:
; CHECK-NEXT: ld1h { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1h { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x half>, <vscale x 2 x half>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x half>, <vscale x 2 x half>* %base, i64 -8
%data = call <vscale x 2 x half> @llvm.masked.load.nxv2f16(<vscale x 2 x half>* %base_load, %data = call <vscale x 2 x half> @llvm.masked.load.nxv2f16(<vscale x 2 x half>* %base_load,
i32 1, i32 1,
@ -127,9 +134,10 @@ define void @test_masked_ldst_sv2f16(<vscale x 2 x half> * %base, <vscale x 2 x
define void @test_masked_ldst_sv2f32(<vscale x 2 x float> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2f32(<vscale x 2 x float> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2f32: ; CHECK-LABEL: test_masked_ldst_sv2f32:
; CHECK-NEXT: ld1w { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1w { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ld1w { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1w { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x float>, <vscale x 2 x float>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x float>, <vscale x 2 x float>* %base, i64 -8
%data = call <vscale x 2 x float> @llvm.masked.load.nxv2f32(<vscale x 2 x float>* %base_load, %data = call <vscale x 2 x float> @llvm.masked.load.nxv2f32(<vscale x 2 x float>* %base_load,
i32 1, i32 1,
@ -145,9 +153,10 @@ define void @test_masked_ldst_sv2f32(<vscale x 2 x float> * %base, <vscale x 2 x
define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2f64: ; CHECK-LABEL: test_masked_ldst_sv2f64:
; CHECK-NEXT: ld1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1d { z[[DATA]].d }, p0, [x0, #-5, mul vl] ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0, #-6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1d { z0.d }, p0, [x0, #-5, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base, i64 -6 %base_load = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base, i64 -6
%data = call <vscale x 2 x double> @llvm.masked.load.nxv2f64(<vscale x 2 x double>* %base_load, %data = call <vscale x 2 x double> @llvm.masked.load.nxv2f64(<vscale x 2 x double>* %base_load,
i32 1, i32 1,
@ -165,8 +174,9 @@ define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2
define <vscale x 2 x i64> @masked_zload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_zload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv2i8_to_sv2i64: ; CHECK-LABEL: masked_zload_sv2i8_to_sv2i64:
; CHECK-NEXT: ld1b { z0.d }, p0/z, [x0, #-4, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1b { z0.d }, p0/z, [x0, #-4, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -4 %base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -4
%load = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load, %load = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load,
i32 1, i32 1,
@ -178,8 +188,9 @@ define <vscale x 2 x i64> @masked_zload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base
define <vscale x 2 x i64> @masked_sload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_sload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv2i8_to_sv2i64: ; CHECK-LABEL: masked_sload_sv2i8_to_sv2i64:
; CHECK-NEXT: ld1sb { z0.d }, p0/z, [x0, #-3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sb { z0.d }, p0/z, [x0, #-3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -3 %base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 -3
%load = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load, %load = call <vscale x 2 x i8> @llvm.masked.load.nxv2i8(<vscale x 2 x i8>* %base_load,
i32 1, i32 1,
@ -191,8 +202,9 @@ define <vscale x 2 x i64> @masked_sload_sv2i8_to_sv2i64(<vscale x 2 x i8>* %base
define <vscale x 2 x i64> @masked_zload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_zload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv2i16_to_sv2i64: ; CHECK-LABEL: masked_zload_sv2i16_to_sv2i64:
; CHECK-NEXT: ld1h { z0.d }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1h { z0.d }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 1 %base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 1
%load = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load, %load = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load,
i32 1, i32 1,
@ -204,8 +216,9 @@ define <vscale x 2 x i64> @masked_zload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %ba
define <vscale x 2 x i64> @masked_sload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_sload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv2i16_to_sv2i64: ; CHECK-LABEL: masked_sload_sv2i16_to_sv2i64:
; CHECK-NEXT: ld1sh { z0.d }, p0/z, [x0, #2, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sh { z0.d }, p0/z, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 2 %base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 2
%load = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load, %load = call <vscale x 2 x i16> @llvm.masked.load.nxv2i16(<vscale x 2 x i16>* %base_load,
i32 1, i32 1,
@ -217,8 +230,9 @@ define <vscale x 2 x i64> @masked_sload_sv2i16_to_sv2i64(<vscale x 2 x i16>* %ba
define <vscale x 2 x i64> @masked_zload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_zload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv2i32_to_sv2i64: ; CHECK-LABEL: masked_zload_sv2i32_to_sv2i64:
; CHECK-NEXT: ld1w { z0.d }, p0/z, [x0, #-2, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1w { z0.d }, p0/z, [x0, #-2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -2 %base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -2
%load = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load, %load = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load,
i32 1, i32 1,
@ -230,8 +244,9 @@ define <vscale x 2 x i64> @masked_zload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %ba
define <vscale x 2 x i64> @masked_sload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %base, <vscale x 2 x i1> %mask) nounwind { define <vscale x 2 x i64> @masked_sload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv2i32_to_sv2i64: ; CHECK-LABEL: masked_sload_sv2i32_to_sv2i64:
; CHECK-NEXT: ld1sw { z0.d }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sw { z0.d }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -1 %base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 -1
%load = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load, %load = call <vscale x 2 x i32> @llvm.masked.load.nxv2i32(<vscale x 2 x i32>* %base_load,
i32 1, i32 1,
@ -245,8 +260,9 @@ define <vscale x 2 x i64> @masked_sload_sv2i32_to_sv2i64(<vscale x 2 x i32>* %ba
define void @masked_trunc_store_sv2i64_to_sv2i8(<vscale x 2 x i64> %val, <vscale x 2 x i8> *%base, <vscale x 2 x i1> %mask) nounwind { define void @masked_trunc_store_sv2i64_to_sv2i8(<vscale x 2 x i64> %val, <vscale x 2 x i8> *%base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i8: ; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i8:
; CHECK-NEXT: st1b { z0.d }, p0, [x0, #3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.d }, p0, [x0, #3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 3 %base_load = getelementptr <vscale x 2 x i8>, <vscale x 2 x i8>* %base, i64 3
%trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i8> %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i8>
call void @llvm.masked.store.nxv2i8(<vscale x 2 x i8> %trunc, call void @llvm.masked.store.nxv2i8(<vscale x 2 x i8> %trunc,
@ -259,8 +275,9 @@ define void @masked_trunc_store_sv2i64_to_sv2i8(<vscale x 2 x i64> %val, <vscale
define void @masked_trunc_store_sv2i64_to_sv2i16(<vscale x 2 x i64> %val, <vscale x 2 x i16> *%base, <vscale x 2 x i1> %mask) nounwind { define void @masked_trunc_store_sv2i64_to_sv2i16(<vscale x 2 x i64> %val, <vscale x 2 x i16> *%base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i16: ; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i16:
; CHECK-NEXT: st1h { z0.d }, p0, [x0, #4, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.d }, p0, [x0, #4, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 4 %base_load = getelementptr <vscale x 2 x i16>, <vscale x 2 x i16>* %base, i64 4
%trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i16> %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i16>
call void @llvm.masked.store.nxv2i16(<vscale x 2 x i16> %trunc, call void @llvm.masked.store.nxv2i16(<vscale x 2 x i16> %trunc,
@ -272,8 +289,9 @@ define void @masked_trunc_store_sv2i64_to_sv2i16(<vscale x 2 x i64> %val, <vscal
define void @masked_trunc_store_sv2i64_to_sv2i32(<vscale x 2 x i64> %val, <vscale x 2 x i32> *%base, <vscale x 2 x i1> %mask) nounwind { define void @masked_trunc_store_sv2i64_to_sv2i32(<vscale x 2 x i64> %val, <vscale x 2 x i32> *%base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i32: ; CHECK-LABEL: masked_trunc_store_sv2i64_to_sv2i32:
; CHECK-NEXT: st1w { z0.d }, p0, [x0, #5, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1w { z0.d }, p0, [x0, #5, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 5 %base_load = getelementptr <vscale x 2 x i32>, <vscale x 2 x i32>* %base, i64 5
%trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i32> %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i32>
call void @llvm.masked.store.nxv2i32(<vscale x 2 x i32> %trunc, call void @llvm.masked.store.nxv2i32(<vscale x 2 x i32> %trunc,
@ -287,9 +305,10 @@ define void @masked_trunc_store_sv2i64_to_sv2i32(<vscale x 2 x i64> %val, <vscal
define void @test_masked_ldst_sv4i8(<vscale x 4 x i8> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4i8(<vscale x 4 x i8> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4i8: ; CHECK-LABEL: test_masked_ldst_sv4i8:
; CHECK-NEXT: ld1sb { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1b { z[[DATA]].s }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1sb { z0.s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -1 %base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -1
%data = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load, %data = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load,
i32 1, i32 1,
@ -305,9 +324,10 @@ define void @test_masked_ldst_sv4i8(<vscale x 4 x i8> * %base, <vscale x 4 x i1>
define void @test_masked_ldst_sv4i16(<vscale x 4 x i16> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4i16(<vscale x 4 x i16> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4i16: ; CHECK-LABEL: test_masked_ldst_sv4i16:
; CHECK-NEXT: ld1sh { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].s }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1sh { z0.s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 -1 %base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 -1
%data = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load, %data = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load,
i32 1, i32 1,
@ -323,9 +343,10 @@ define void @test_masked_ldst_sv4i16(<vscale x 4 x i16> * %base, <vscale x 4 x i
define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4i32: ; CHECK-LABEL: test_masked_ldst_sv4i32:
; CHECK-NEXT: ld1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1w { z[[DATA]].s }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1w { z0.s }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base, i64 6 %base_load = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base, i64 6
%data = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(<vscale x 4 x i32>* %base_load, %data = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(<vscale x 4 x i32>* %base_load,
i32 1, i32 1,
@ -341,9 +362,10 @@ define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i
define void @test_masked_ldst_sv4f16(<vscale x 4 x half> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4f16(<vscale x 4 x half> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4f16: ; CHECK-LABEL: test_masked_ldst_sv4f16:
; CHECK-NEXT: ld1h { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].s }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1h { z0.s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x half>, <vscale x 4 x half>* %base, i64 -1 %base_load = getelementptr <vscale x 4 x half>, <vscale x 4 x half>* %base, i64 -1
%data = call <vscale x 4 x half> @llvm.masked.load.nxv4f16(<vscale x 4 x half>* %base_load, %data = call <vscale x 4 x half> @llvm.masked.load.nxv4f16(<vscale x 4 x half>* %base_load,
i32 1, i32 1,
@ -359,9 +381,10 @@ define void @test_masked_ldst_sv4f16(<vscale x 4 x half> * %base, <vscale x 4 x
define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4f32: ; CHECK-LABEL: test_masked_ldst_sv4f32:
; CHECK-NEXT: ld1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1w { z[[DATA]].s }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1w { z0.s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base, i64 -1 %base_load = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base, i64 -1
%data = call <vscale x 4 x float> @llvm.masked.load.nxv4f32(<vscale x 4 x float>* %base_load, %data = call <vscale x 4 x float> @llvm.masked.load.nxv4f32(<vscale x 4 x float>* %base_load,
i32 1, i32 1,
@ -379,8 +402,9 @@ define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x
define <vscale x 4 x i32> @masked_zload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base, <vscale x 4 x i1> %mask) nounwind { define <vscale x 4 x i32> @masked_zload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv4i8_to_sv4i32: ; CHECK-LABEL: masked_zload_sv4i8_to_sv4i32:
; CHECK-NEXT: ld1b { z0.s }, p0/z, [x0, #-4, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1b { z0.s }, p0/z, [x0, #-4, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -4 %base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -4
%load = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load, %load = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load,
i32 1, i32 1,
@ -392,8 +416,9 @@ define <vscale x 4 x i32> @masked_zload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base
define <vscale x 4 x i32> @masked_sload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base, <vscale x 4 x i1> %mask) nounwind { define <vscale x 4 x i32> @masked_sload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv4i8_to_sv4i32: ; CHECK-LABEL: masked_sload_sv4i8_to_sv4i32:
; CHECK-NEXT: ld1sb { z0.s }, p0/z, [x0, #-3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sb { z0.s }, p0/z, [x0, #-3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -3 %base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 -3
%load = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load, %load = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8(<vscale x 4 x i8>* %base_load,
i32 1, i32 1,
@ -405,8 +430,9 @@ define <vscale x 4 x i32> @masked_sload_sv4i8_to_sv4i32(<vscale x 4 x i8>* %base
define <vscale x 4 x i32> @masked_zload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %base, <vscale x 4 x i1> %mask) nounwind { define <vscale x 4 x i32> @masked_zload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv4i16_to_sv4i32: ; CHECK-LABEL: masked_zload_sv4i16_to_sv4i32:
; CHECK-NEXT: ld1h { z0.s }, p0/z, [x0, #1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1h { z0.s }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 1 %base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 1
%load = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load, %load = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load,
i32 1, i32 1,
@ -418,8 +444,9 @@ define <vscale x 4 x i32> @masked_zload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %ba
define <vscale x 4 x i32> @masked_sload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %base, <vscale x 4 x i1> %mask) nounwind { define <vscale x 4 x i32> @masked_sload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv4i16_to_sv4i32: ; CHECK-LABEL: masked_sload_sv4i16_to_sv4i32:
; CHECK-NEXT: ld1sh { z0.s }, p0/z, [x0, #2, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sh { z0.s }, p0/z, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 2 %base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 2
%load = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load, %load = call <vscale x 4 x i16> @llvm.masked.load.nxv4i16(<vscale x 4 x i16>* %base_load,
i32 1, i32 1,
@ -433,8 +460,9 @@ define <vscale x 4 x i32> @masked_sload_sv4i16_to_sv4i32(<vscale x 4 x i16>* %ba
define void @masked_trunc_store_sv4i32_to_sv4i8(<vscale x 4 x i32> %val, <vscale x 4 x i8> *%base, <vscale x 4 x i1> %mask) nounwind { define void @masked_trunc_store_sv4i32_to_sv4i8(<vscale x 4 x i32> %val, <vscale x 4 x i8> *%base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv4i32_to_sv4i8: ; CHECK-LABEL: masked_trunc_store_sv4i32_to_sv4i8:
; CHECK-NEXT: st1b { z0.s }, p0, [x0, #3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.s }, p0, [x0, #3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 3 %base_load = getelementptr <vscale x 4 x i8>, <vscale x 4 x i8>* %base, i64 3
%trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i8> %trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i8>
call void @llvm.masked.store.nxv4i8(<vscale x 4 x i8> %trunc, call void @llvm.masked.store.nxv4i8(<vscale x 4 x i8> %trunc,
@ -447,8 +475,9 @@ define void @masked_trunc_store_sv4i32_to_sv4i8(<vscale x 4 x i32> %val, <vscale
define void @masked_trunc_store_sv4i32_to_sv4i16(<vscale x 4 x i32> %val, <vscale x 4 x i16> *%base, <vscale x 4 x i1> %mask) nounwind { define void @masked_trunc_store_sv4i32_to_sv4i16(<vscale x 4 x i32> %val, <vscale x 4 x i16> *%base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv4i32_to_sv4i16: ; CHECK-LABEL: masked_trunc_store_sv4i32_to_sv4i16:
; CHECK-NEXT: st1h { z0.s }, p0, [x0, #4, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.s }, p0, [x0, #4, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 4 %base_load = getelementptr <vscale x 4 x i16>, <vscale x 4 x i16>* %base, i64 4
%trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i16> %trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i16>
call void @llvm.masked.store.nxv4i16(<vscale x 4 x i16> %trunc, call void @llvm.masked.store.nxv4i16(<vscale x 4 x i16> %trunc,
@ -462,9 +491,10 @@ define void @masked_trunc_store_sv4i32_to_sv4i16(<vscale x 4 x i32> %val, <vscal
define void @test_masked_ldst_sv8i8(<vscale x 8 x i8> * %base, <vscale x 8 x i1> %mask) nounwind { define void @test_masked_ldst_sv8i8(<vscale x 8 x i8> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8i8: ; CHECK-LABEL: test_masked_ldst_sv8i8:
; CHECK-NEXT: ld1sb { z[[DATA:[0-9]+]].h }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1b { z[[DATA]].h }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ld1sb { z0.h }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.h }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 6 %base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 6
%data = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load, %data = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load,
i32 1, i32 1,
@ -480,9 +510,10 @@ define void @test_masked_ldst_sv8i8(<vscale x 8 x i8> * %base, <vscale x 8 x i1>
define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i1> %mask) nounwind { define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8i16: ; CHECK-LABEL: test_masked_ldst_sv8i16:
; CHECK-NEXT: ld1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].h }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.h }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base, i64 6 %base_load = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base, i64 6
%data = call <vscale x 8 x i16> @llvm.masked.load.nxv8i16(<vscale x 8 x i16>* %base_load, %data = call <vscale x 8 x i16> @llvm.masked.load.nxv8i16(<vscale x 8 x i16>* %base_load,
i32 1, i32 1,
@ -498,9 +529,10 @@ define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i
define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x i1> %mask) nounwind { define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8f16: ; CHECK-LABEL: test_masked_ldst_sv8f16:
; CHECK-NEXT: ld1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].h }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base, i64 -1 %base_load = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base, i64 -1
%data = call <vscale x 8 x half> @llvm.masked.load.nxv8f16(<vscale x 8 x half>* %base_load, %data = call <vscale x 8 x half> @llvm.masked.load.nxv8f16(<vscale x 8 x half>* %base_load,
i32 1, i32 1,
@ -516,9 +548,10 @@ define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x
define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8 x i1> %mask) nounwind #0 { define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8 x i1> %mask) nounwind #0 {
; CHECK-LABEL: test_masked_ldst_sv8bf16: ; CHECK-LABEL: test_masked_ldst_sv8bf16:
; CHECK-NEXT: ld1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1h { z[[DATA]].h }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1h { z0.h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base, i64 -1 %base_load = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base, i64 -1
%data = call <vscale x 8 x bfloat> @llvm.masked.load.nxv8bf16(<vscale x 8 x bfloat>* %base_load, %data = call <vscale x 8 x bfloat> @llvm.masked.load.nxv8bf16(<vscale x 8 x bfloat>* %base_load,
i32 1, i32 1,
@ -536,8 +569,9 @@ define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8
define <vscale x 8 x i16> @masked_zload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base, <vscale x 8 x i1> %mask) nounwind { define <vscale x 8 x i16> @masked_zload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: masked_zload_sv8i8_to_sv8i16: ; CHECK-LABEL: masked_zload_sv8i8_to_sv8i16:
; CHECK-NEXT: ld1b { z0.h }, p0/z, [x0, #-4, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1b { z0.h }, p0/z, [x0, #-4, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 -4 %base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 -4
%load = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load, %load = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load,
i32 1, i32 1,
@ -549,8 +583,9 @@ define <vscale x 8 x i16> @masked_zload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base
define <vscale x 8 x i16> @masked_sload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base, <vscale x 8 x i1> %mask) nounwind { define <vscale x 8 x i16> @masked_sload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: masked_sload_sv8i8_to_sv8i16: ; CHECK-LABEL: masked_sload_sv8i8_to_sv8i16:
; CHECK-NEXT: ld1sb { z0.h }, p0/z, [x0, #-3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: ld1sb { z0.h }, p0/z, [x0, #-3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 -3 %base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 -3
%load = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load, %load = call <vscale x 8 x i8> @llvm.masked.load.nxv8i8(<vscale x 8 x i8>* %base_load,
i32 1, i32 1,
@ -564,8 +599,9 @@ define <vscale x 8 x i16> @masked_sload_sv8i8_to_sv8i16(<vscale x 8 x i8>* %base
define void @masked_trunc_store_sv8i16_to_sv8i8(<vscale x 8 x i16> %val, <vscale x 8 x i8> *%base, <vscale x 8 x i1> %mask) nounwind { define void @masked_trunc_store_sv8i16_to_sv8i8(<vscale x 8 x i16> %val, <vscale x 8 x i8> *%base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: masked_trunc_store_sv8i16_to_sv8i8: ; CHECK-LABEL: masked_trunc_store_sv8i16_to_sv8i8:
; CHECK-NEXT: st1b { z0.h }, p0, [x0, #3, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.h }, p0, [x0, #3, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 3 %base_load = getelementptr <vscale x 8 x i8>, <vscale x 8 x i8>* %base, i64 3
%trunc = trunc <vscale x 8 x i16> %val to <vscale x 8 x i8> %trunc = trunc <vscale x 8 x i16> %val to <vscale x 8 x i8>
call void @llvm.masked.store.nxv8i8(<vscale x 8 x i8> %trunc, call void @llvm.masked.store.nxv8i8(<vscale x 8 x i8> %trunc,
@ -579,9 +615,10 @@ define void @masked_trunc_store_sv8i16_to_sv8i8(<vscale x 8 x i16> %val, <vscale
define void @test_masked_ldst_sv16i8(<vscale x 16 x i8> * %base, <vscale x 16 x i1> %mask) nounwind { define void @test_masked_ldst_sv16i8(<vscale x 16 x i8> * %base, <vscale x 16 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv16i8: ; CHECK-LABEL: test_masked_ldst_sv16i8:
; CHECK-NEXT: ld1b { z[[DATA:[0-9]+]].b }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: st1b { z[[DATA]].b }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ld1b { z0.b }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: st1b { z0.b }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base, i64 6 %base_load = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base, i64 6
%data = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8(<vscale x 16 x i8>* %base_load, %data = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8(<vscale x 16 x i8>* %base_load,
i32 1, i32 1,

74
test/CodeGen/AArch64/sve-pred-non-temporal-ldst-addressing-mode-reg-imm.ll
View File

@ -1,4 +1,5 @@
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve --asm-verbose=false < %s 2>%t | FileCheck %s ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s 2>%t | FileCheck %s
; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t ; RUN: FileCheck --check-prefix=WARN --allow-empty %s <%t
; If this check fails please read test/CodeGen/AArch64/README for instructions on how to resolve it. ; If this check fails please read test/CodeGen/AArch64/README for instructions on how to resolve it.
@ -11,13 +12,14 @@
define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind { define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: imm_out_of_range: ; CHECK-LABEL: imm_out_of_range:
; CHECK-NEXT: rdvl x8, #8 ; CHECK: // %bb.0:
; CHECK-NEXT: add x8, x0, x8 ; CHECK-NEXT: rdvl x8, #8
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x{{[0-9]+}}] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: rdvl x8, #-9 ; CHECK-NEXT: ldnt1d { z0.d }, p0/z, [x8]
; CHECK-NEXT: add x8, x0, x8 ; CHECK-NEXT: rdvl x8, #-9
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x{{[0-9]+}}] ; CHECK-NEXT: add x8, x0, x8
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1d { z0.d }, p0, [x8]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 8 %base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 8
%base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64* %base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64*
%data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask, %data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask,
@ -35,9 +37,10 @@ define void @imm_out_of_range(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mas
define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2i64: ; CHECK-LABEL: test_masked_ldst_sv2i64:
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-8, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x0, #-7, mul vl] ; CHECK-NEXT: ldnt1d { z0.d }, p0/z, [x0, #-8, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1d { z0.d }, p0, [x0, #-7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 -8 %base_load = getelementptr <vscale x 2 x i64>, <vscale x 2 x i64>* %base, i64 -8
%base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64* %base_load_bc = bitcast <vscale x 2 x i64>* %base_load to i64*
%data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask, %data = call <vscale x 2 x i64> @llvm.aarch64.sve.ldnt1.nxv2i64(<vscale x 2 x i1> %mask,
@ -52,9 +55,10 @@ define void @test_masked_ldst_sv2i64(<vscale x 2 x i64> * %base, <vscale x 2 x i
define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2 x i1> %mask) nounwind { define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv2f64: ; CHECK-LABEL: test_masked_ldst_sv2f64:
; CHECK-NEXT: ldnt1d { z[[DATA:[0-9]+]].d }, p0/z, [x0, #-6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1d { z[[DATA]].d }, p0, [x0, #-5, mul vl] ; CHECK-NEXT: ldnt1d { z0.d }, p0/z, [x0, #-6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1d { z0.d }, p0, [x0, #-5, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base, i64 -6 %base_load = getelementptr <vscale x 2 x double>, <vscale x 2 x double>* %base, i64 -6
%base_load_bc = bitcast <vscale x 2 x double>* %base_load to double* %base_load_bc = bitcast <vscale x 2 x double>* %base_load to double*
%data = call <vscale x 2 x double> @llvm.aarch64.sve.ldnt1.nxv2f64(<vscale x 2 x i1> %mask, %data = call <vscale x 2 x double> @llvm.aarch64.sve.ldnt1.nxv2f64(<vscale x 2 x i1> %mask,
@ -71,9 +75,10 @@ define void @test_masked_ldst_sv2f64(<vscale x 2 x double> * %base, <vscale x 2
define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4i32: ; CHECK-LABEL: test_masked_ldst_sv4i32:
; CHECK-NEXT: ldnt1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1w { z[[DATA]].s }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ldnt1w { z0.s }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1w { z0.s }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base, i64 6 %base_load = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* %base, i64 6
%base_load_bc = bitcast <vscale x 4 x i32>* %base_load to i32* %base_load_bc = bitcast <vscale x 4 x i32>* %base_load to i32*
%data = call <vscale x 4 x i32> @llvm.aarch64.sve.ldnt1.nxv4i32(<vscale x 4 x i1> %mask, %data = call <vscale x 4 x i32> @llvm.aarch64.sve.ldnt1.nxv4i32(<vscale x 4 x i1> %mask,
@ -88,9 +93,10 @@ define void @test_masked_ldst_sv4i32(<vscale x 4 x i32> * %base, <vscale x 4 x i
define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x i1> %mask) nounwind { define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv4f32: ; CHECK-LABEL: test_masked_ldst_sv4f32:
; CHECK-NEXT: ldnt1w { z[[DATA:[0-9]+]].s }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1w { z[[DATA]].s }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ldnt1w { z0.s }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1w { z0.s }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base, i64 -1 %base_load = getelementptr <vscale x 4 x float>, <vscale x 4 x float>* %base, i64 -1
%base_load_bc = bitcast <vscale x 4 x float>* %base_load to float* %base_load_bc = bitcast <vscale x 4 x float>* %base_load to float*
%data = call <vscale x 4 x float> @llvm.aarch64.sve.ldnt1.nxv4f32(<vscale x 4 x i1> %mask, %data = call <vscale x 4 x float> @llvm.aarch64.sve.ldnt1.nxv4f32(<vscale x 4 x i1> %mask,
@ -108,9 +114,10 @@ define void @test_masked_ldst_sv4f32(<vscale x 4 x float> * %base, <vscale x 4 x
define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i1> %mask) nounwind { define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8i16: ; CHECK-LABEL: test_masked_ldst_sv8i16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ldnt1h { z0.h }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1h { z0.h }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base, i64 6 %base_load = getelementptr <vscale x 8 x i16>, <vscale x 8 x i16>* %base, i64 6
%base_load_bc = bitcast <vscale x 8 x i16>* %base_load to i16* %base_load_bc = bitcast <vscale x 8 x i16>* %base_load to i16*
%data = call <vscale x 8 x i16> @llvm.aarch64.sve.ldnt1.nxv8i16(<vscale x 8 x i1> %mask, %data = call <vscale x 8 x i16> @llvm.aarch64.sve.ldnt1.nxv8i16(<vscale x 8 x i1> %mask,
@ -125,9 +132,10 @@ define void @test_masked_ldst_sv8i16(<vscale x 8 x i16> * %base, <vscale x 8 x i
define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x i1> %mask) nounwind { define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv8f16: ; CHECK-LABEL: test_masked_ldst_sv8f16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ldnt1h { z0.h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1h { z0.h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base, i64 -1 %base_load = getelementptr <vscale x 8 x half>, <vscale x 8 x half>* %base, i64 -1
%base_load_bc = bitcast <vscale x 8 x half>* %base_load to half* %base_load_bc = bitcast <vscale x 8 x half>* %base_load to half*
%data = call <vscale x 8 x half> @llvm.aarch64.sve.ldnt1.nxv8f16(<vscale x 8 x i1> %mask, %data = call <vscale x 8 x half> @llvm.aarch64.sve.ldnt1.nxv8f16(<vscale x 8 x i1> %mask,
@ -142,9 +150,10 @@ define void @test_masked_ldst_sv8f16(<vscale x 8 x half> * %base, <vscale x 8 x
define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8 x i1> %mask) nounwind #0 { define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8 x i1> %mask) nounwind #0 {
; CHECK-LABEL: test_masked_ldst_sv8bf16: ; CHECK-LABEL: test_masked_ldst_sv8bf16:
; CHECK-NEXT: ldnt1h { z[[DATA:[0-9]+]].h }, p0/z, [x0, #-1, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1h { z[[DATA]].h }, p0, [x0, #2, mul vl] ; CHECK-NEXT: ldnt1h { z0.h }, p0/z, [x0, #-1, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1h { z0.h }, p0, [x0, #2, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base, i64 -1 %base_load = getelementptr <vscale x 8 x bfloat>, <vscale x 8 x bfloat>* %base, i64 -1
%base_load_bc = bitcast <vscale x 8 x bfloat>* %base_load to bfloat* %base_load_bc = bitcast <vscale x 8 x bfloat>* %base_load to bfloat*
%data = call <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnt1.nxv8bf16(<vscale x 8 x i1> %mask, %data = call <vscale x 8 x bfloat> @llvm.aarch64.sve.ldnt1.nxv8bf16(<vscale x 8 x i1> %mask,
@ -161,9 +170,10 @@ define void @test_masked_ldst_sv8bf16(<vscale x 8 x bfloat> * %base, <vscale x 8
define void @test_masked_ldst_sv16i8(<vscale x 16 x i8> * %base, <vscale x 16 x i1> %mask) nounwind { define void @test_masked_ldst_sv16i8(<vscale x 16 x i8> * %base, <vscale x 16 x i1> %mask) nounwind {
; CHECK-LABEL: test_masked_ldst_sv16i8: ; CHECK-LABEL: test_masked_ldst_sv16i8:
; CHECK-NEXT: ldnt1b { z[[DATA:[0-9]+]].b }, p0/z, [x0, #6, mul vl] ; CHECK: // %bb.0:
; CHECK-NEXT: stnt1b { z[[DATA]].b }, p0, [x0, #7, mul vl] ; CHECK-NEXT: ldnt1b { z0.b }, p0/z, [x0, #6, mul vl]
; CHECK-NEXT: ret ; CHECK-NEXT: stnt1b { z0.b }, p0, [x0, #7, mul vl]
; CHECK-NEXT: ret
%base_load = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base, i64 6 %base_load = getelementptr <vscale x 16 x i8>, <vscale x 16 x i8>* %base, i64 6
%base_load_bc = bitcast <vscale x 16 x i8>* %base_load to i8* %base_load_bc = bitcast <vscale x 16 x i8>* %base_load to i8*
%data = call <vscale x 16 x i8> @llvm.aarch64.sve.ldnt1.nxv16i8(<vscale x 16 x i1> %mask, %data = call <vscale x 16 x i8> @llvm.aarch64.sve.ldnt1.nxv16i8(<vscale x 16 x i1> %mask,