brintos

brintos / llvm-project-archived public Read only

0
0
Text · 2.0 KiB · 89771d2 Raw
46 lines · plain
1// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 52// RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn-amd-amdhsa -target-cpu gfx942 \3// RUN:   %s -emit-llvm -o - -disable-llvm-passes | FileCheck %s4 5// REQUIRES: amdgpu-registered-target6 7// `Ptr.getElementType()` in `CheckAtomicAlignment` returns8//      %struct.__half2 = type { %union.anon }9// Check we do not crash when handling that.10 11typedef half  __attribute__((ext_vector_type(2))) half2;12typedef short  __attribute__((ext_vector_type(2))) short2;13 14struct __half2 {15    union {16        struct {17            half x;18            half y;19        };20        half2 data;21    };22};23 24// CHECK-LABEL: define dso_local <2 x half> @test_flat_add_2f16(25// CHECK-SAME: ptr noundef [[ADDR:%.*]], <2 x half> noundef [[VAL:%.*]]) #[[ATTR0:[0-9]+]] {26// CHECK-NEXT:  [[ENTRY:.*:]]27// CHECK-NEXT:    [[RETVAL:%.*]] = alloca <2 x half>, align 4, addrspace(5)28// CHECK-NEXT:    [[ADDR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)29// CHECK-NEXT:    [[VAL_ADDR:%.*]] = alloca <2 x half>, align 4, addrspace(5)30// CHECK-NEXT:    [[RETVAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RETVAL]] to ptr31// CHECK-NEXT:    [[ADDR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[ADDR_ADDR]] to ptr32// CHECK-NEXT:    [[VAL_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VAL_ADDR]] to ptr33// CHECK-NEXT:    store ptr [[ADDR]], ptr [[ADDR_ADDR_ASCAST]], align 834// CHECK-NEXT:    store <2 x half> [[VAL]], ptr [[VAL_ADDR_ASCAST]], align 435// CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[ADDR_ADDR_ASCAST]], align 836// CHECK-NEXT:    [[TMP1:%.*]] = load <2 x half>, ptr [[VAL_ADDR_ASCAST]], align 437// CHECK-NEXT:    [[TMP2:%.*]] = atomicrmw fadd ptr [[TMP0]], <2 x half> [[TMP1]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4:![0-9]+]]38// CHECK-NEXT:    ret <2 x half> [[TMP2]]39//40half2 test_flat_add_2f16(short2 *addr, half2 val) {41  return __builtin_amdgcn_flat_atomic_fadd_v2f16((struct __half2*)addr, val);42}43//.44// CHECK: [[META4]] = !{}45//.46