1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve < %s | FileCheck %s
4 define <vscale x 2 x i64> @masked_gather_nxv2i8(<vscale x 2 x i8*> %ptrs, <vscale x 2 x i1> %mask) {
5 ; CHECK-LABEL: masked_gather_nxv2i8:
7 ; CHECK-NEXT: ld1b { z0.d }, p0/z, [z0.d]
9 %vals = call <vscale x 2 x i8> @llvm.masked.gather.nxv2i8(<vscale x 2 x i8*> %ptrs, i32 1, <vscale x 2 x i1> %mask, <vscale x 2 x i8> undef)
10 %vals.zext = zext <vscale x 2 x i8> %vals to <vscale x 2 x i64>
11 ret <vscale x 2 x i64> %vals.zext
14 define <vscale x 2 x i64> @masked_gather_nxv2i16(<vscale x 2 x i16*> %ptrs, <vscale x 2 x i1> %mask) {
15 ; CHECK-LABEL: masked_gather_nxv2i16:
17 ; CHECK-NEXT: ld1h { z0.d }, p0/z, [z0.d]
19 %vals = call <vscale x 2 x i16> @llvm.masked.gather.nxv2i16(<vscale x 2 x i16*> %ptrs, i32 2, <vscale x 2 x i1> %mask, <vscale x 2 x i16> undef)
20 %vals.zext = zext <vscale x 2 x i16> %vals to <vscale x 2 x i64>
21 ret <vscale x 2 x i64> %vals.zext
24 define <vscale x 2 x i64> @masked_gather_nxv2i32(<vscale x 2 x i32*> %ptrs, <vscale x 2 x i1> %mask) {
25 ; CHECK-LABEL: masked_gather_nxv2i32:
27 ; CHECK-NEXT: ld1w { z0.d }, p0/z, [z0.d]
29 %vals = call <vscale x 2 x i32> @llvm.masked.gather.nxv2i32(<vscale x 2 x i32*> %ptrs, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x i32> undef)
30 %vals.zext = zext <vscale x 2 x i32> %vals to <vscale x 2 x i64>
31 ret <vscale x 2 x i64> %vals.zext
34 define <vscale x 2 x i64> @masked_gather_nxv2i64(<vscale x 2 x i64*> %ptrs, <vscale x 2 x i1> %mask) {
35 ; CHECK-LABEL: masked_gather_nxv2i64:
37 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [z0.d]
39 %vals = call <vscale x 2 x i64> @llvm.masked.gather.nxv2i64(<vscale x 2 x i64*> %ptrs, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x i64> undef)
40 ret <vscale x 2 x i64> %vals
43 define <vscale x 2 x half> @masked_gather_nxv2f16(<vscale x 2 x half*> %ptrs, <vscale x 2 x i1> %mask) {
44 ; CHECK-LABEL: masked_gather_nxv2f16:
46 ; CHECK-NEXT: ld1h { z0.d }, p0/z, [z0.d]
48 %vals = call <vscale x 2 x half> @llvm.masked.gather.nxv2f16(<vscale x 2 x half*> %ptrs, i32 2, <vscale x 2 x i1> %mask, <vscale x 2 x half> undef)
49 ret <vscale x 2 x half> %vals
52 define <vscale x 2 x bfloat> @masked_gather_nxv2bf16(<vscale x 2 x bfloat*> %ptrs, <vscale x 2 x i1> %mask) #0 {
53 ; CHECK-LABEL: masked_gather_nxv2bf16:
55 ; CHECK-NEXT: ld1h { z0.d }, p0/z, [z0.d]
57 %vals = call <vscale x 2 x bfloat> @llvm.masked.gather.nxv2bf16(<vscale x 2 x bfloat*> %ptrs, i32 2, <vscale x 2 x i1> %mask, <vscale x 2 x bfloat> undef)
58 ret <vscale x 2 x bfloat> %vals
61 define <vscale x 2 x float> @masked_gather_nxv2f32(<vscale x 2 x float*> %ptrs, <vscale x 2 x i1> %mask) {
62 ; CHECK-LABEL: masked_gather_nxv2f32:
64 ; CHECK-NEXT: ld1w { z0.d }, p0/z, [z0.d]
66 %vals = call <vscale x 2 x float> @llvm.masked.gather.nxv2f32(<vscale x 2 x float*> %ptrs, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x float> undef)
67 ret <vscale x 2 x float> %vals
70 define <vscale x 2 x double> @masked_gather_nxv2f64(<vscale x 2 x double*> %ptrs, <vscale x 2 x i1> %mask) {
71 ; CHECK-LABEL: masked_gather_nxv2f64:
73 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [z0.d]
75 %vals = call <vscale x 2 x double> @llvm.masked.gather.nxv2f64(<vscale x 2 x double*> %ptrs, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x double> undef)
76 ret <vscale x 2 x double> %vals
79 define <vscale x 2 x i64> @masked_sgather_nxv2i8(<vscale x 2 x i8*> %ptrs, <vscale x 2 x i1> %mask) {
80 ; CHECK-LABEL: masked_sgather_nxv2i8:
82 ; CHECK-NEXT: ld1sb { z0.d }, p0/z, [z0.d]
84 %vals = call <vscale x 2 x i8> @llvm.masked.gather.nxv2i8(<vscale x 2 x i8*> %ptrs, i32 1, <vscale x 2 x i1> %mask, <vscale x 2 x i8> undef)
85 %vals.sext = sext <vscale x 2 x i8> %vals to <vscale x 2 x i64>
86 ret <vscale x 2 x i64> %vals.sext
89 define <vscale x 2 x i64> @masked_sgather_nxv2i16(<vscale x 2 x i16*> %ptrs, <vscale x 2 x i1> %mask) {
90 ; CHECK-LABEL: masked_sgather_nxv2i16:
92 ; CHECK-NEXT: ld1sh { z0.d }, p0/z, [z0.d]
94 %vals = call <vscale x 2 x i16> @llvm.masked.gather.nxv2i16(<vscale x 2 x i16*> %ptrs, i32 2, <vscale x 2 x i1> %mask, <vscale x 2 x i16> undef)
95 %vals.sext = sext <vscale x 2 x i16> %vals to <vscale x 2 x i64>
96 ret <vscale x 2 x i64> %vals.sext
99 define <vscale x 2 x i64> @masked_sgather_nxv2i32(<vscale x 2 x i32*> %ptrs, <vscale x 2 x i1> %mask) {
100 ; CHECK-LABEL: masked_sgather_nxv2i32:
102 ; CHECK-NEXT: ld1sw { z0.d }, p0/z, [z0.d]
104 %vals = call <vscale x 2 x i32> @llvm.masked.gather.nxv2i32(<vscale x 2 x i32*> %ptrs, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x i32> undef)
105 %vals.sext = sext <vscale x 2 x i32> %vals to <vscale x 2 x i64>
106 ret <vscale x 2 x i64> %vals.sext
109 define <vscale x 2 x i64> @masked_gather_passthru(<vscale x 2 x i32*> %ptrs, <vscale x 2 x i1> %mask, <vscale x 2 x i32> %passthru) {
110 ; CHECK-LABEL: masked_gather_passthru:
112 ; CHECK-NEXT: ld1sw { z0.d }, p0/z, [z0.d]
113 ; CHECK-NEXT: sel z0.d, p0, z0.d, z1.d
115 %vals = call <vscale x 2 x i32> @llvm.masked.gather.nxv2i32(<vscale x 2 x i32*> %ptrs, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x i32> %passthru)
116 %vals.sext = sext <vscale x 2 x i32> %vals to <vscale x 2 x i64>
117 ret <vscale x 2 x i64> %vals.sext
120 define <vscale x 2 x i64> @masked_gather_passthru_0(<vscale x 2 x i32*> %ptrs, <vscale x 2 x i1> %mask) {
121 ; CHECK-LABEL: masked_gather_passthru_0:
123 ; CHECK-NEXT: ld1sw { z0.d }, p0/z, [z0.d]
125 %vals = call <vscale x 2 x i32> @llvm.masked.gather.nxv2i32(<vscale x 2 x i32*> %ptrs, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x i32> zeroinitializer)
126 %vals.sext = sext <vscale x 2 x i32> %vals to <vscale x 2 x i64>
127 ret <vscale x 2 x i64> %vals.sext
130 %i64_x3 = type { i64, i64, i64}
131 define <vscale x 2 x i64> @masked_gather_non_power_of_two_based_scaling(ptr %base, <vscale x 2 x i64> %offsets, <vscale x 2 x i1> %mask) {
132 ; CHECK-LABEL: masked_gather_non_power_of_two_based_scaling:
134 ; CHECK-NEXT: mul z0.d, z0.d, #24
135 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0, z0.d]
137 %ptrs = getelementptr inbounds %i64_x3, ptr %base, <vscale x 2 x i64> %offsets
138 %vals = call <vscale x 2 x i64> @llvm.masked.gather.nxv2i64(<vscale x 2 x ptr> %ptrs, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x i64> undef)
139 ret <vscale x 2 x i64> %vals
142 %i64_x4 = type { i64, i64, i64, i64}
143 define <vscale x 2 x i64> @masked_gather_non_element_type_based_scaling(ptr %base, <vscale x 2 x i64> %offsets, <vscale x 2 x i1> %mask) {
144 ; CHECK-LABEL: masked_gather_non_element_type_based_scaling:
146 ; CHECK-NEXT: lsl z0.d, z0.d, #5
147 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0, z0.d]
149 %ptrs = getelementptr inbounds %i64_x4, ptr %base, <vscale x 2 x i64> %offsets
150 %vals = call <vscale x 2 x i64> @llvm.masked.gather.nxv2i64(<vscale x 2 x ptr> %ptrs, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x i64> undef)
151 ret <vscale x 2 x i64> %vals
154 declare <vscale x 2 x i8> @llvm.masked.gather.nxv2i8(<vscale x 2 x i8*>, i32, <vscale x 2 x i1>, <vscale x 2 x i8>)
155 declare <vscale x 2 x i16> @llvm.masked.gather.nxv2i16(<vscale x 2 x i16*>, i32, <vscale x 2 x i1>, <vscale x 2 x i16>)
156 declare <vscale x 2 x i32> @llvm.masked.gather.nxv2i32(<vscale x 2 x i32*>, i32, <vscale x 2 x i1>, <vscale x 2 x i32>)
157 declare <vscale x 2 x i64> @llvm.masked.gather.nxv2i64(<vscale x 2 x i64*>, i32, <vscale x 2 x i1>, <vscale x 2 x i64>)
158 declare <vscale x 2 x half> @llvm.masked.gather.nxv2f16(<vscale x 2 x half*>, i32, <vscale x 2 x i1>, <vscale x 2 x half>)
159 declare <vscale x 2 x bfloat> @llvm.masked.gather.nxv2bf16(<vscale x 2 x bfloat*>, i32, <vscale x 2 x i1>, <vscale x 2 x bfloat>)
160 declare <vscale x 2 x float> @llvm.masked.gather.nxv2f32(<vscale x 2 x float*>, i32, <vscale x 2 x i1>, <vscale x 2 x float>)
161 declare <vscale x 2 x double> @llvm.masked.gather.nxv2f64(<vscale x 2 x double*>, i32, <vscale x 2 x i1>, <vscale x 2 x double>)
162 attributes #0 = { "target-features"="+sve,+bf16" }