PyTorch Model Training: RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR

Athena Wisdom picture Athena Wisdom · May 28, 2020 · Viewed 14.6k times · Source

After training a PyTorch model on a GPU for several hours, the program fails with the error

RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR

Training Conditions

  • Neural Network: PyTorch 4-layer nn.LSTM with nn.Linear output
  • Deep Q Network Agent (Vanilla DQN with Replay Memory)
  • state passed into forward() has the shape (32, 20, 15), where 32 is the batch size
  • 50 seconds per episode
  • Error occurs after about 583 episodes (8 hours) or 1,150,000 steps, where each step involves a forward pass through the LSTM model.

My code also has the following values set before the training began

torch.manual_seed(0)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
np.random.seed(0)

How can we troubleshoot this problem? Since this occurred 8 hours into the training, some educated guess will be very helpful here!

Thanks!


Update:

Commenting out the 2 torch.backends.cudnn... lines did not work. CUDNN_STATUS_INTERNAL_ERROR still occurs, but much earlier at around Episode 300 (585,000 steps).

torch.manual_seed(0)
#torch.backends.cudnn.deterministic = True
#torch.backends.cudnn.benchmark = False
np.random.seed(0)

System

  • PyTorch 1.6.0.dev20200525
  • CUDA 10.2
  • cuDNN 7604
  • Python 3.8
  • Windows 10
  • nVidia 1080 GPU

Error Traceback

RuntimeError                              Traceback (most recent call last)
<ipython-input-18-f5bbb4fdfda5> in <module>
     57 
     58     while not done:
---> 59         action = agent.choose_action(state)
     60         state_, reward, done, info = env.step(action)
     61         score += reward

<ipython-input-11-5ad4dd57b5ad> in choose_action(self, state)
     58         if np.random.random() > self.epsilon:
     59             state = T.tensor([state], dtype=T.float).to(self.q_eval.device)
---> 60             actions = self.q_eval.forward(state)
     61             action = T.argmax(actions).item()
     62         else:

<ipython-input-10-94271a92f66e> in forward(self, state)
     20 
     21     def forward(self, state):
---> 22         lstm, hidden = self.lstm(state)
     23         actions = self.fc1(lstm[:,-1:].squeeze(1))
     24         return actions

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\nn\modules\module.py in __call__(self, *input, **kwargs)
    575             result = self._slow_forward(*input, **kwargs)
    576         else:
--> 577             result = self.forward(*input, **kwargs)
    578         for hook in self._forward_hooks.values():
    579             hook_result = hook(self, input, result)

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\nn\modules\rnn.py in forward(self, input, hx)
    571         self.check_forward_args(input, hx, batch_sizes)
    572         if batch_sizes is None:
--> 573             result = _VF.lstm(input, hx, self._flat_weights, self.bias, self.num_layers,
    574                               self.dropout, self.training, self.bidirectional, self.batch_first)
    575         else:

RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR

Update: Tried try... except on my code where this error occurs at, and in addition to RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR, we also get a second traceback for the error RuntimeError: CUDA error: unspecified launch failure

During handling of the above exception, another exception occurred:

RuntimeError                              Traceback (most recent call last)
<ipython-input-4-e8f15cc8cf4f> in <module>
     61 
     62     while not done:
---> 63         action = agent.choose_action(state)
     64         state_, reward, done, info = env.step(action)
     65         score += reward

<ipython-input-3-1aae79080e99> in choose_action(self, state)
     58         if np.random.random() > self.epsilon:
     59             state = T.tensor([state], dtype=T.float).to(self.q_eval.device)
---> 60             actions = self.q_eval.forward(state)
     61             action = T.argmax(actions).item()
     62         else:

<ipython-input-2-6d22bb632c4c> in forward(self, state)
     25         except Exception as e:
     26             print('error in forward() with state:', state.shape, 'exception:', e)
---> 27             print('state:', state)
     28         actions = self.fc1(lstm[:,-1:].squeeze(1))
     29         return actions

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\tensor.py in __repr__(self)
    152     def __repr__(self):
    153         # All strings are unicode in Python 3.
--> 154         return torch._tensor_str._str(self)
    155 
    156     def backward(self, gradient=None, retain_graph=None, create_graph=False):

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\_tensor_str.py in _str(self)
    331                 tensor_str = _tensor_str(self.to_dense(), indent)
    332             else:
--> 333                 tensor_str = _tensor_str(self, indent)
    334 
    335     if self.layout != torch.strided:

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\_tensor_str.py in _tensor_str(self, indent)
    227     if self.dtype is torch.float16 or self.dtype is torch.bfloat16:
    228         self = self.float()
--> 229     formatter = _Formatter(get_summarized_data(self) if summarize else self)
    230     return _tensor_str_with_formatter(self, indent, formatter, summarize)
    231 

~\AppData\Local\Continuum\anaconda3\envs\rl\lib\site-packages\torch\_tensor_str.py in __init__(self, tensor)
     99 
    100         else:
--> 101             nonzero_finite_vals = torch.masked_select(tensor_view, torch.isfinite(tensor_view) & tensor_view.ne(0))
    102 
    103             if nonzero_finite_vals.numel() == 0:

RuntimeError: CUDA error: unspecified launch failure

Answer

Michael Jungo picture Michael Jungo · May 28, 2020

The error RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR is notoriously difficult to debug, but surprisingly often it's an out of memory problem. Usually, you would get the out of memory error, but depending on where it occurs, PyTorch cannot intercept the error and therefore not provide a meaningful error message.

A memory issue seems to be likely in your case, because you are using a while loop until the agent is done, which might take long enough that you run out of memory, it's just a matter of time. That can also possibly occur rather late, once the model's parameters in combination with a certain input is unable to finish in time.

You can avoid that scenario by limiting the number of allowed actions instead of hoping that the actor will be done in a reasonable time.

What you also need to be careful about, is that you don't occupy unnecessary memory. A common mistake is to keep computing gradients of the past states in future iterations. The state from the last iteration should be considered constant, since the current action should not affect past actions, therefore no gradients are required. This is usually achieved by detaching the state from the computational graph for the next iteration, e.g. state = state_.detach(). Maybe you are already doing that, but without the code it's impossible to tell.

Similarly, if you keep a history of the states, you should detach them and even more importantly put them on the CPU, i.e. history.append(state.detach().cpu()).