diff --git a/docs/Trainer/Logging.md b/docs/Trainer/Logging.md
index d6eb39d6d6..27d1e85d24 100644
--- a/docs/Trainer/Logging.md
+++ b/docs/Trainer/Logging.md
@@ -14,6 +14,14 @@ Every k batches lightning will make an entry in the metrics log
 ``` {.python}
 # DEFAULT (ie: save a .csv log file every 10 batches)
 trainer = Trainer(add_log_row_interval=10)
+```   
+
+---
+#### Log metric row every k batches 
+Logs GPU memory when metrics are logged.   
+``` {.python}
+# DEFAULT
+trainer = Trainer(log_gpu_memory=False)
 ```
 
 ---
diff --git a/pytorch_lightning/models/trainer.py b/pytorch_lightning/models/trainer.py
index 4a7ae48fb0..d84e80a710 100644
--- a/pytorch_lightning/models/trainer.py
+++ b/pytorch_lightning/models/trainer.py
@@ -63,6 +63,7 @@ class Trainer(TrainerIO):
                  current_gpu_name=0,
                  nb_gpu_nodes=1,
                  gpus=None,
+                 log_gpu_memory=False,
                  show_progress_bar=True,
                  overfit_pct=0.0,
                  track_grad_norm=-1,
@@ -94,6 +95,9 @@ class Trainer(TrainerIO):
         :param current_gpu_name:
         :param nb_gpu_nodes:
         :param gpus:
+        :param \: Log GPU memory utilization as metric
+            during training. This can lead to lower performance on some
+            servers, in particular when `nvidia-smi` is slow.
         :param show_progress_bar:
         :param overfit_pct:
         :param track_grad_norm:
@@ -118,6 +122,7 @@ class Trainer(TrainerIO):
         """
         # Transfer params
         self.nb_gpu_nodes = nb_gpu_nodes
+        self.log_gpu_memory = log_gpu_memory
         self.gradient_clip = gradient_clip
         self.check_val_every_n_epoch = check_val_every_n_epoch
         self.enable_early_stop = early_stop_callback is not None
@@ -934,7 +939,7 @@ class Trainer(TrainerIO):
                 metrics = self.__tng_tqdm_dic
 
                 # add gpu memory
-                if self.on_gpu:
+                if self.on_gpu and self.log_gpu_memory:
                     mem_map = get_gpu_memory_map()
                     metrics.update(mem_map)