| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2 |
| ; RUN: opt -S -passes=infer-address-spaces -assume-default-is-flat-addrspace %s | FileCheck %s |
| |
| define <4 x i32> @masked_gather_inferas(ptr addrspace(1) %out, <4 x i64> %index) { |
| ; CHECK-LABEL: define <4 x i32> @masked_gather_inferas |
| ; CHECK-SAME: (ptr addrspace(1) [[OUT:%.*]], <4 x i64> [[INDEX:%.*]]) { |
| ; CHECK-NEXT: entry: |
| ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT]], <4 x i64> [[INDEX]] |
| ; CHECK-NEXT: [[VALUE:%.*]] = tail call <4 x i32> @llvm.masked.gather.v4i32.v4p1(<4 x ptr addrspace(1)> [[PTRS]], i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> poison) |
| ; CHECK-NEXT: ret <4 x i32> [[VALUE]] |
| ; |
| entry: |
| %out.1 = addrspacecast ptr addrspace(1) %out to ptr |
| %ptrs = getelementptr inbounds i32, ptr %out.1, <4 x i64> %index |
| %value = tail call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> poison) |
| ret <4 x i32> %value |
| } |
| |
| define void @masked_scatter_inferas(ptr addrspace(1) %out, <4 x i64> %index, <4 x i32> %value) { |
| ; CHECK-LABEL: define void @masked_scatter_inferas |
| ; CHECK-SAME: (ptr addrspace(1) [[OUT:%.*]], <4 x i64> [[INDEX:%.*]], <4 x i32> [[VALUE:%.*]]) { |
| ; CHECK-NEXT: entry: |
| ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT]], <4 x i64> [[INDEX]] |
| ; CHECK-NEXT: tail call void @llvm.masked.scatter.v4i32.v4p1(<4 x i32> [[VALUE]], <4 x ptr addrspace(1)> [[PTRS]], i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>) |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| %out.1 = addrspacecast ptr addrspace(1) %out to ptr |
| %ptrs = getelementptr inbounds i32, ptr %out.1, <4 x i64> %index |
| tail call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> %value, <4 x ptr> %ptrs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>) |
| ret void |
| } |
| |
| declare <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr>, i32 immarg, <4 x i1>, <4 x i32>) |
| |
| declare void @llvm.masked.scatter.v4i32.v4p0(<4 x i32>, <4 x ptr>, i32 immarg, <4 x i1>) |