diff --git a/tensorflow/compiler/jit/mark_for_compilation_pass.cc b/tensorflow/compiler/jit/mark_for_compilation_pass.cc
index 96cf169b914285..37a2aa72d9feb8 100644
--- a/tensorflow/compiler/jit/mark_for_compilation_pass.cc
+++ b/tensorflow/compiler/jit/mark_for_compilation_pass.cc
@@ -990,8 +990,7 @@ Status MarkForCompilationPassImpl::CreateClusters() {
     // trouble.
 
     if (cluster->effective_cluster_size() >= debug_options_.min_cluster_size ||
-        cluster->has_functional_control_flow() ||
-        cluster->is_xla_compile_attr_true()) {
+        cluster->has_functional_control_flow()) {
       string& name = cluster_names[cluster->cycles_graph_node_id()];
 
       if (name.empty()) {
diff --git a/tensorflow/core/kernels/segment_reduction_ops_gpu.cu.h b/tensorflow/core/kernels/segment_reduction_ops_gpu.cu.h
index a9bf175f205f4e..8e2efd84549000 100644
--- a/tensorflow/core/kernels/segment_reduction_ops_gpu.cu.h
+++ b/tensorflow/core/kernels/segment_reduction_ops_gpu.cu.h
@@ -902,7 +902,7 @@ struct UnsortedSegmentFunctor<GPUDevice, T, Index, InitialValueF, ReductionF> {
       config = GetGpuLaunchConfig(data_size, d);
       TF_CHECK_OK(GpuLaunchKernel(
           UnsortedSegmentCustomKernel<
-              T, Index, typename ReduceUpdateOpFor<ReductionF>::atomic_op>,
+              T, Index, typename ReduceUpdateOpFor<ReductionF>::nonatomic_op>,
           config.block_count, config.thread_per_block, 0, d.stream(),
           input_outer_dim_size, input_inner_dim_size, output_outer_dim_size,
           unsorted_segment_ids.data(), data.data(), output.data()));