diff --git a/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen.cpp b/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen.cpp
@@ -115,7 +115,7 @@ int bar(int n){
 // CHECK: [[ELT_CAST:%.+]] = bitcast double* [[ELT]] to i64*
 // CHECK: [[REMOTE_ELT_CAST:%.+]] = bitcast double* [[REMOTE_ELT]] to i64*
 // CHECK: [[ELT_VAL:%.+]] = load i64, i64* [[ELT_CAST]], align
-// CHECK: [[WS32:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize()
+// CHECK: [[WS32:%.+]] = call i32 @__kmpc_get_warp_size()
 // CHECK: [[WS:%.+]] = trunc i32 [[WS32]] to i16
 // CHECK: [[REMOTE_ELT_VAL64:%.+]] = call i64 @__kmpc_shuffle_int64(i64 [[ELT_VAL]], i16 [[LANEOFFSET]], i16 [[WS]])
 //
@@ -320,7 +320,7 @@ int bar(int n){
 // CHECK: [[ELT_VAL:%.+]] = load i8, i8* [[ELT_VOID]], align
 //
 // CHECK: [[ELT_CAST:%.+]] = sext i8 [[ELT_VAL]] to i32
-// CHECK: [[WS32:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize()
+// CHECK: [[WS32:%.+]] = call i32 @__kmpc_get_warp_size()
 // CHECK: [[WS:%.+]] = trunc i32 [[WS32]] to i16
 // CHECK: [[REMOTE_ELT1_VAL32:%.+]] = call i32 @__kmpc_shuffle_int32(i32 [[ELT_CAST]], i16 [[LANEOFFSET]], i16 [[WS]])
 // CHECK: [[REMOTE_ELT1_VAL:%.+]] = trunc i32 [[REMOTE_ELT1_VAL32]] to i8
@@ -336,7 +336,7 @@ int bar(int n){
 // CHECK: [[ELT_CAST:%.+]] = bitcast float* [[ELT]] to i32*
 // CHECK: [[REMOTE_ELT2_CAST:%.+]] = bitcast float* [[REMOTE_ELT2]] to i32*
 // CHECK: [[ELT_VAL:%.+]] = load i32, i32* [[ELT_CAST]], align
-// CHECK: [[WS32:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize()
+// CHECK: [[WS32:%.+]] = call i32 @__kmpc_get_warp_size()
 // CHECK: [[WS:%.+]] = trunc i32 [[WS32]] to i16
 // CHECK: [[REMOTE_ELT2_VAL32:%.+]] = call i32 @__kmpc_shuffle_int32(i32 [[ELT_VAL]], i16 [[LANEOFFSET]], i16 [[WS]])
 //
@@ -617,7 +617,7 @@ int bar(int n){
 // CHECK: [[ELT:%.+]] = bitcast i8* [[ELT_VOID]] to i32*
 // CHECK: [[ELT_VAL:%.+]] = load i32, i32* [[ELT]], align
 //
-// CHECK: [[WS32:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize()
+// CHECK: [[WS32:%.+]] = call i32 @__kmpc_get_warp_size()
 // CHECK: [[WS:%.+]] = trunc i32 [[WS32]] to i16
 // CHECK: [[REMOTE_ELT1_VAL:%.+]] = call i32 @__kmpc_shuffle_int32(i32 [[ELT_VAL]], i16 [[LANEOFFSET]], i16 [[WS]])
 //
@@ -632,7 +632,7 @@ int bar(int n){
 // CHECK: [[ELT_VAL:%.+]] = load i16, i16* [[ELT]], align
 //
 // CHECK: [[ELT_CAST:%.+]] = sext i16 [[ELT_VAL]] to i32
-// CHECK: [[WS32:%.+]] = call i32 @llvm.nvvm.read.ptx.sreg.warpsize()
+// CHECK: [[WS32:%.+]] = call i32 @__kmpc_get_warp_size()
 // CHECK: [[WS:%.+]] = trunc i32 [[WS32]] to i16
 // CHECK: [[REMOTE_ELT2_VAL32:%.+]] = call i32 @__kmpc_shuffle_int32(i32 [[ELT_CAST]], i16 [[LANEOFFSET]], i16 [[WS]])
 // CHECK: [[REMOTE_ELT2_VAL:%.+]] = trunc i32 [[REMOTE_ELT2_VAL32]] to i16

diff --git a/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp b/clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp
diff --git a/clang/test/OpenMP/nvptx_teams_reduction_codegen.cpp b/clang/test/OpenMP/nvptx_teams_reduction_codegen.cpp
diff --git a/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def b/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
@@ -455,6 +455,8 @@ __OMP_RTL(__kmpc_barrier_simple_generic, false, Void, IdentPtr, Int32)
 __OMP_RTL(__kmpc_warp_active_thread_mask, false, Int64,)
 __OMP_RTL(__kmpc_syncwarp, false, Void, Int64)
 
+__OMP_RTL(__kmpc_get_warp_size, false, Int32, )
+
 __OMP_RTL(__kmpc_is_generic_main_thread_id, false, Int8, Int32)
 
 __OMP_RTL(__last, false, Void, )

diff --git a/openmp/libomptarget/DeviceRTL/src/Mapping.cpp b/openmp/libomptarget/DeviceRTL/src/Mapping.cpp
@@ -277,5 +277,10 @@ __attribute__((noinline)) uint32_t __kmpc_get_hardware_num_threads_in_block() {
   FunctionTracingRAII();
   return impl::getNumHardwareThreadsInBlock();
 }
+
+__attribute__((noinline)) uint32_t __kmpc_get_warp_size() {
+  FunctionTracingRAII();
+  return impl::getWarpSize();
+}
 }
 #pragma omp end declare target
diff --git a/openmp/libomptarget/DeviceRTL/src/Utils.cpp b/openmp/libomptarget/DeviceRTL/src/Utils.cpp
@@ -24,6 +24,7 @@ namespace _OMP {
 __attribute__((used, weak, optnone)) void keepAlive() {
   __kmpc_get_hardware_thread_id_in_block();
   __kmpc_get_hardware_num_threads_in_block();
+  __kmpc_get_warp_size();
   __kmpc_barrier_simple_spmd(nullptr, 0);
   __kmpc_barrier_simple_generic(nullptr, 0);
 }

diff --git a/openmp/libomptarget/deviceRTLs/amdgcn/src/target_impl.hip b/openmp/libomptarget/deviceRTLs/amdgcn/src/target_impl.hip
@@ -133,8 +133,11 @@ EXTERN int __kmpc_get_hardware_num_threads_in_block() {
                            __builtin_amdgcn_workgroup_size_x());
 }
 
+EXTERN unsigned __kmpc_get_warp_size() {
+  return WARPSIZE;
+}
+
 EXTERN unsigned GetWarpId() { return __kmpc_get_hardware_thread_id_in_block() / WARPSIZE; }
-EXTERN unsigned GetWarpSize() { return WARPSIZE; }
 EXTERN unsigned GetLaneId() {
   return __builtin_amdgcn_mbcnt_hi(~0u, __builtin_amdgcn_mbcnt_lo(~0u, 0u));
 }

diff --git a/openmp/libomptarget/deviceRTLs/common/include/target/shuffle.h b/openmp/libomptarget/deviceRTLs/common/include/target/shuffle.h
@@ -35,7 +35,7 @@ int64_t __kmpc_shuffle_int64(int64_t val, int16_t delta, int16_t size);
 ///{
 extern "C" {
 unsigned GetLaneId();
-unsigned GetWarpSize();
+unsigned __kmpc_get_warp_size();
 void __kmpc_impl_unpack(uint64_t val, uint32_t &lo, uint32_t &hi);
 uint64_t __kmpc_impl_pack(uint32_t lo, uint32_t hi);
 }
@@ -60,7 +60,7 @@ int32_t __kmpc_impl_shfl_down_sync(uint64_t Mask, int32_t Var, uint32_t Delta,
 
 inline int32_t __kmpc_impl_shfl_sync(uint64_t Mask, int32_t Var,
                                      int32_t SrcLane) {
-  int Width = GetWarpSize();
+  int Width = __kmpc_get_warp_size();
   int Self = GetLaneId();
   int Index = SrcLane + (Self & ~(Width - 1));
   return __builtin_amdgcn_ds_bpermute(Index << 2, Var);
@@ -90,7 +90,7 @@ inline int32_t __kmpc_impl_shfl_sync(uint64_t Mask, int32_t Var,
 
 inline int32_t __kmpc_impl_shfl_down_sync(uint64_t Mask, int32_t Var,
                                           uint32_t Delta, int32_t Width) {
-  int32_t T = ((GetWarpSize() - Width) << 8) | 0x1f;
+  int32_t T = ((__kmpc_get_warp_size() - Width) << 8) | 0x1f;
   return __nvvm_shfl_sync_down_i32(Mask, Var, Delta, T);
 }
 

diff --git a/openmp/libomptarget/deviceRTLs/nvptx/src/target_impl.cu b/openmp/libomptarget/deviceRTLs/nvptx/src/target_impl.cu
@@ -102,10 +102,10 @@ EXTERN int __kmpc_get_hardware_num_blocks() {
 EXTERN int __kmpc_get_hardware_num_threads_in_block() {
   return __nvvm_read_ptx_sreg_ntid_x();
 }
+EXTERN unsigned __kmpc_get_warp_size() { return WARPSIZE; }
 EXTERN unsigned GetWarpId() {
   return __kmpc_get_hardware_thread_id_in_block() / WARPSIZE;
 }
-EXTERN unsigned GetWarpSize() { return WARPSIZE; }
 EXTERN unsigned GetLaneId() {
   return __kmpc_get_hardware_thread_id_in_block() & (WARPSIZE - 1);
 }