iff --git a/benchmarks/bfs/src/hip/kernel.cc b/benchmarks/bfs/src/hip/kernel.cc
index 985436d..89f8c2e 100644
--- a/benchmarks/bfs/src/hip/kernel.cc
+++ b/benchmarks/bfs/src/hip/kernel.cc
@@ -225,16 +225,17 @@ BFS_in_GPU_kernel(grid_launch_parm lp,
     int tid = hipBlockIdx_x*MAX_THREADS_PER_BLOCK + hipThreadIdx_x;
     if( tid<no_of_nodes)
     {
-      int pid;
-      if(tot_sum == 0)//this is the first BFS level of current kernel call
-        pid = q1[tid];
-      else
-        pid = next_wf[tid];//read the current frontier info from last level's propagation
-
       // Visit a node from the current frontier; update costs, colors, and
       // output queue
-      visit_node(pid, hipThreadIdx_x & MOD_OP, g_graph_nodes, g_graph_edges,
-                 local_q, overflow, g_color, g_cost, gray_shade);
+      if (tot_sum == 0) {^M
+        //this is the first BFS level of current kernel call^M
+        visit_node(q1[tid], hipThreadIdx_x & MOD_OP, g_graph_nodes, g_graph_edges,^M
+                   local_q, overflow, g_color, g_cost, gray_shade);^M
+      } else {^M
+        //read the current frontier info from last level's propagation^M
+        visit_node(next_wf[tid], hipThreadIdx_x & MOD_OP, g_graph_nodes, g_graph_edges,^M
+                   local_q, overflow, g_color, g_cost, gray_shade);^M
+      }^M
     }
     __syncthreads();
     if(hipThreadIdx_x == 0){