1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -aarch64-sve-vector-bits-min=256 < %s | FileCheck %s
3 ; RUN: llc -aarch64-sve-vector-bits-min=512 < %s | FileCheck %s
5 target triple = "aarch64-unknown-linux-gnu"
7 define void @masked_gather_base_plus_stride_v8f32(ptr %dst, ptr %src) #0 {
8 ; CHECK-LABEL: masked_gather_base_plus_stride_v8f32:
10 ; CHECK-NEXT: index z0.s, #0, #7
11 ; CHECK-NEXT: ptrue p0.s, vl8
12 ; CHECK-NEXT: ld1w { z0.s }, p0/z, [x1, z0.s, sxtw #2]
13 ; CHECK-NEXT: st1w { z0.s }, p0, [x0]
15 %ptrs = getelementptr float, ptr %src, <8 x i64> <i64 0, i64 7, i64 14, i64 21, i64 28, i64 35, i64 42, i64 49>
16 %data = tail call <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr> %ptrs, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x float> undef)
17 store <8 x float> %data, ptr %dst, align 4
21 define void @masked_gather_base_plus_stride_v4f64(ptr %dst, ptr %src) #0 {
22 ; CHECK-LABEL: masked_gather_base_plus_stride_v4f64:
24 ; CHECK-NEXT: mov x8, #-32 // =0xffffffffffffffe0
25 ; CHECK-NEXT: ptrue p0.d, vl4
26 ; CHECK-NEXT: index z0.d, #-2, x8
27 ; CHECK-NEXT: ld1d { z0.d }, p0/z, [x1, z0.d, lsl #3]
28 ; CHECK-NEXT: st1d { z0.d }, p0, [x0]
30 %ptrs = getelementptr double, ptr %src, <4 x i64> <i64 -2, i64 -34, i64 -66, i64 -98>
31 %data = tail call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> %ptrs, i32 8, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x double> undef)
32 store <4 x double> %data, ptr %dst, align 8
36 define void @masked_scatter_base_plus_stride_v8f32(ptr %dst, ptr %src) #0 {
37 ; CHECK-LABEL: masked_scatter_base_plus_stride_v8f32:
39 ; CHECK-NEXT: ptrue p0.s, vl8
40 ; CHECK-NEXT: index z0.s, #0, #-7
41 ; CHECK-NEXT: ld1w { z1.s }, p0/z, [x1]
42 ; CHECK-NEXT: st1w { z1.s }, p0, [x0, z0.s, sxtw #2]
44 %data = load <8 x float>, ptr %src, align 4
45 %ptrs = getelementptr float, ptr %dst, <8 x i64> <i64 0, i64 -7, i64 -14, i64 -21, i64 -28, i64 -35, i64 -42, i64 -49>
46 tail call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> %data, <8 x ptr> %ptrs, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
50 define void @masked_scatter_base_plus_stride_v4f64(ptr %dst, ptr %src) #0 {
51 ; CHECK-LABEL: masked_scatter_base_plus_stride_v4f64:
53 ; CHECK-NEXT: ptrue p0.d, vl4
54 ; CHECK-NEXT: index z0.d, #-2, #3
55 ; CHECK-NEXT: ld1d { z1.d }, p0/z, [x1]
56 ; CHECK-NEXT: st1d { z1.d }, p0, [x0, z0.d, lsl #3]
58 %data = load <4 x double>, ptr %src, align 8
59 %ptrs = getelementptr double, ptr %dst, <4 x i64> <i64 -2, i64 1, i64 4, i64 7>
60 tail call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> %data, <4 x ptr> %ptrs, i32 8, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
64 declare <8 x float> @llvm.masked.gather.v8f32.v8p0(<8 x ptr>, i32 immarg, <8 x i1>, <8 x float>)
65 declare <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr>, i32 immarg, <4 x i1>, <4 x double>)
67 declare void @llvm.masked.scatter.v8f32.v8p0(<8 x float>, <8 x ptr>, i32 immarg, <8 x i1>)
68 declare void @llvm.masked.scatter.v4f64.v4p0(<4 x double>, <4 x ptr>, i32 immarg, <4 x i1>)
70 attributes #0 = { "target-features"="+sve" }