Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR

系统信息

• Ubuntu 18.04
• 安装方式 nvidia-docker
• 源镜像 tensorflow/tensorflow:2.0.0-gpu-py3
• CUDA/cuDNN version: 10.0 / 7.4.1
• GPU model and memory: RTX 2080, driver 410.104

错误信息

[email protected]:/home/ssw/learnTF# python cats_vs_dogs.py
2019-10-15 03:01:46.803820: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2019-10-15 03:01:46.999480: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.8
pciBusID: 0000:01:00.0
2019-10-15 03:01:47.001229: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 1 with properties:
name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.8
pciBusID: 0000:82:00.0
2019-10-15 03:01:47.048776: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2019-10-15 03:01:47.209664: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2019-10-15 03:01:47.270684: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2019-10-15 03:01:47.313241: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2019-10-15 03:01:47.468857: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2019-10-15 03:01:47.587926: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2019-10-15 03:01:47.980507: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2019-10-15 03:01:47.987278: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0, 1
2019-10-15 03:01:47.987668: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-10-15 03:01:48.029860: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2199995000 Hz
2019-10-15 03:01:48.034298: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x495d960 executing computations on platform Host. Devices:
2019-10-15 03:01:48.034345: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Host, Default Version
2019-10-15 03:01:48.525680: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x49bfe80 executing computations on platform CUDA. Devices:
2019-10-15 03:01:48.525733: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): GeForce RTX 2080, Compute Capability 7.5
2019-10-15 03:01:48.525750: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (1): GeForce RTX 2080, Compute Capability 7.5
2019-10-15 03:01:48.527637: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.8
pciBusID: 0000:01:00.0
2019-10-15 03:01:48.528398: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 1 with properties:
name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.8
pciBusID: 0000:82:00.0
2019-10-15 03:01:48.528465: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2019-10-15 03:01:48.528491: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2019-10-15 03:01:48.528505: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2019-10-15 03:01:48.528520: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2019-10-15 03:01:48.528535: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2019-10-15 03:01:48.528548: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2019-10-15 03:01:48.528564: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2019-10-15 03:01:48.531341: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0, 1
2019-10-15 03:01:48.553555: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2019-10-15 03:01:48.557020: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-10-15 03:01:48.557046: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165]      0 1
2019-10-15 03:01:48.557056: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0:   N N
2019-10-15 03:01:48.557063: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 1:   N N
2019-10-15 03:01:48.587799: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1304] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 6467 MB memory) -> physical GPU (device: 0, name: GeForce RTX 2080, pci bus id: 0000:01:00.0, compute capability: 7.5)
2019-10-15 03:01:48.591898: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1304] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:1 with 7414 MB memory) -> physical GPU (device: 1, name: GeForce RTX 2080, pci bus id: 0000:82:00.0, compute capability: 7.5)
Epoch 1/10
2019-10-15 03:01:56.490886: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2019-10-15 03:02:04.519738: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2019-10-15 03:02:06.908614: E tensorflow/stream_executor/cuda/cuda_dnn.cc:329] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR
2019-10-15 03:02:06.916591: E tensorflow/stream_executor/cuda/cuda_dnn.cc:329] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR
2019-10-15 03:02:06.924926: W tensorflow/core/common_runtime/base_collective_executor.cc:216] BaseCollectiveExecutor::StartAbort Unknown: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
     [[{{node sequential/conv2d/Conv2D}}]]
      1/Unknown - 12s 12s/stepTraceback (most recent call last):
  File "cats_vs_dogs.py", line 69, in <module>
    model.fit(train_dataset, epochs=num_epochs)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/training.py", line 728, in fit
    use_multiprocessing=use_multiprocessing)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/training_v2.py", line 324, in fit
    total_epochs=epochs)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/training_v2.py", line 123, in run_one_epoch
    batch_outs = execution_function(iterator)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/training_v2_utils.py", line 86, in execution_function
    distributed_function(input_fn))
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/def_function.py", line 457, in __call__
    result = self._call(*args, **kwds)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/def_function.py", line 520, in _call
    return self._stateless_fn(*args, **kwds)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/function.py", line 1823, in __call__
    return graph_function._filtered_call(args, kwargs)  # pylint: disable=protected-access
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/function.py", line 1141, in _filtered_call
    self.captured_inputs)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/function.py", line 1224, in _call_flat
    ctx, args, cancellation_manager=cancellation_manager)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/function.py", line 511, in call
    ctx=ctx)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/eager/execute.py", line 67, in quick_execute
    six.raise_from(core._status_to_exception(e.code, message), None)
  File "<string>", line 3, in raise_from
tensorflow.python.framework.errors_impl.UnknownError:  Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
     [[node sequential/conv2d/Conv2D (defined at /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/framework/ops.py:1751) ]] [Op:__inference_distributed_function_877]


Function call stack:
distributed_function

解决办法

在代码中加入

physical_devices = tf.config.experimental.list_physical_devices('GPU')
tf.config.experimental.set_memory_growth(physical_devices[0], True)
tf.config.experimental.set_memory_growth(physical_devices[1], True)

 若有三块GPU，则修改为

physical_devices = tf.config.experimental.list_physical_devices('GPU')
tf.config.experimental.set_memory_growth(physical_devices[0], True)
tf.config.experimental.set_memory_growth(physical_devices[1], True)
tf.config.experimental.set_memory_growth(physical_devices[2], True)