diff --git a/src/common/common.h b/src/common/common.h
index 9772b9bd01094a371a86d47aa0cfe1004ff66868..8d067cc980d20de596c4a517225f98f2e88ad58d 100644
--- a/src/common/common.h
+++ b/src/common/common.h
@@ -37,8 +37,12 @@
     #define TIME_STOP 
 #else 
     #define TIME_DEF struct timespec __start,__end;
-    #define TIME_START clock_gettime(CLOCK_MONOTONIC,&__start);
+    #define TIME_START { \
+        MPI_Barrier(ctx -> mpi_communicator); \
+        clock_gettime(CLOCK_MONOTONIC,&__start); \
+    }
     #define TIME_STOP { \
+        MPI_Barrier(ctx -> mpi_communicator); \
         clock_gettime(CLOCK_MONOTONIC,&__end); \
         MPI_DB_PRINT("[TIME] elapsed time %.2lfs\n", (double)(__end.tv_sec - __start.tv_sec) \
                 + (__end.tv_nsec - __start.tv_nsec)/1e9); \
diff --git a/src/tree/tree.c b/src/tree/tree.c
index 771b40a8e8c203116f130c529c01f8ca1e9bd641..4af8fa8d2d89cf1bf4ffd6687c7db06b90475760 100644
--- a/src/tree/tree.c
+++ b/src/tree/tree.c
@@ -1116,6 +1116,7 @@ void exchange_points(global_context_t* ctx, top_kdtree_t* tree)
 
 	ctx -> local_n_points = tot_count / ctx -> dims; 
     int* ppp = (int*)malloc(ctx -> world_size * sizeof(int));
+
     MPI_Allgather(&(ctx -> local_n_points), 1, MPI_INT, ppp, 1, MPI_INT, ctx -> mpi_communicator);
 	ctx -> idx_start = 0;
 	for(int i = 0; i < ctx -> mpi_rank; ++i)
@@ -1129,23 +1130,14 @@ void exchange_points(global_context_t* ctx, top_kdtree_t* tree)
 	free(ctx -> local_data);
 	ctx -> local_data = rcvbuffer;
 
-    for(int i = 0; i < ctx -> local_n_points; ++i)
-    {
-        float_t d1 = ctx -> local_data[i * ctx -> dims] + 8.33416939;
-        float_t d2 = ctx -> local_data[i * ctx -> dims + 1] + 8.22858047;
-        if (sqrt(d1 * d1 + d2 * d2) < 1e-5)
-        {
-            DB_PRINT("Rank %d found it!!! idx %d\n", ctx -> mpi_rank, i);
-        }
-    }
-	
 	/* check exchange */
+    /*
 	for(size_t i = 0; i < ctx -> local_n_points; ++i)
 	{
-		/* tree walk */
 		int o = compute_point_owner(ctx, tree, ctx -> local_data + (i * ctx -> dims));
 		if(o != ctx -> mpi_rank) DB_PRINT("rank %d got an error\n",ctx -> mpi_rank);
 	}
+    */
 	free(points_owners);
 	free(points_per_proc);
 	free(partition_offset);
@@ -1497,12 +1489,12 @@ void mpi_ngbh_search(global_context_t* ctx, datapoint_info_t* dp_info, top_kdtre
     {
         heap_batches_per_node[p] = __heap_batches_to_snd + (uint64_t)rcv_displ[p] * (uint64_t)k;
     }
-    TIME_STOP
 
     /* compute everything */
-    MPI_DB_PRINT("[MASTER] Working on recieved points\n");
     MPI_Barrier(ctx -> mpi_communicator);
+    TIME_STOP
 
+    MPI_DB_PRINT("[MASTER] Working on recieved points\n");
 
     TIME_START
     for(int p = 0; p < ctx -> world_size; ++p)
@@ -1676,6 +1668,7 @@ void mpi_ngbh_search(global_context_t* ctx, datapoint_info_t* dp_info, top_kdtre
 
 
     
+    TIME_START
 	for(int i = 0; i < ctx -> world_size; ++i)
 	{
 		if(data_to_send_per_proc[i])  free(data_to_send_per_proc[i]);
@@ -1697,6 +1690,7 @@ void mpi_ngbh_search(global_context_t* ctx, datapoint_info_t* dp_info, top_kdtre
     free(__heap_batches_to_snd);
     free(__rcv_points);
     free(__snd_points);
+    TIME_STOP
 }
 
 void test_the_idea(global_context_t* ctx)
@@ -1779,7 +1773,10 @@ void ordered_data_to_file(global_context_t* ctx)
 void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx) 
 {
 	float_t *data;
+    TIME_DEF
 
+    TIME_START
+    MPI_DB_PRINT("[MASTER] Reading file and scattering\n");
 	if (ctx->mpi_rank == 0) 
 	{
 		//data = read_data_file(ctx, "../norm_data/50_blobs_more_var.npy", MY_TRUE);
@@ -1789,7 +1786,7 @@ void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx)
     
         // 190M points
         // std_g2980844_091_0000
-		 data = read_data_file(ctx,"../norm_data/std_g2980844_091_0000",MY_TRUE);
+		data = read_data_file(ctx,"../norm_data/std_g2980844_091_0000",MY_TRUE);
         
         /* 1M points ca.*/
 		//data = read_data_file(ctx,"../norm_data/std_LR_091_0001",MY_TRUE);
@@ -1808,7 +1805,7 @@ void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx)
 
 		// ctx -> n_points = 48*5*2000;
 		ctx->n_points = ctx->n_points / ctx->dims;
-		//ctx->n_points = ctx->n_points / 2;
+		ctx->n_points = (ctx->n_points * 10) / 10;
         //ctx -> n_points = ctx -> world_size * 1000;
 		mpi_printf(ctx, "Read %lu points in %u dims\n", ctx->n_points, ctx->dims);
 	}
@@ -1856,9 +1853,9 @@ void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx)
 	original_ps.ub_box = (float_t*)malloc(ctx -> dims * sizeof(float_t));
 
 	float_t tol = 0.002;
+    TIME_STOP
 
 	top_kdtree_t tree;
-    TIME_DEF
     TIME_START
 	top_tree_init(ctx, &tree);
     TIME_STOP
@@ -1866,6 +1863,7 @@ void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx)
     TIME_START
 	build_top_kdtree(ctx, &original_ps, &tree, k_global, tol);
 	exchange_points(ctx, &tree);
+    MPI_Barrier(ctx -> mpi_communicator);
     TIME_STOP
 	//test_the_idea(ctx);
 
@@ -1879,7 +1877,15 @@ void simulate_master_read_and_scatter(int dims, size_t n, global_context_t *ctx)
 	build_local_tree(ctx, &local_tree);
     TIME_STOP
 
+    TIME_START
+    MPI_DB_PRINT("[MASTER] Performing ngbh search\n");
+    MPI_Barrier(ctx -> mpi_communicator);
+
 	mpi_ngbh_search(ctx, dp_info, &tree, &local_tree, ctx -> local_data, k);
+
+    MPI_Barrier(ctx -> mpi_communicator);
+    MPI_DB_PRINT("Tot time ---->");
+    TIME_STOP
 		
 
     #if defined (WRITE_NGBH)