Я пытаюсь настроить правильную среду с тензорным потоком Anaconda. У меня есть: Ubuntu 18.04.3 LTS, Nvidia GeForce GTX 950M / PCIe / SSE2, python 3.7.6, тензор потока 2.0.0, cudatoolkit-10.0.130 и cudnn-7.6.5. Я скачал Tensorflow, cudatoolkit и cudnn, используя anaconda. После настройки gpuenv tenorflow я активировал его и попробовал следующий код, чтобы проверить, все ли работает правильно:
from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())
и получил следующий вывод:
2020-03-08 17:09:17.921816: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2020-03-08 17:09:17.959404: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2592000000 Hz
2020-03-08 17:09:17.960777: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x5568d22e8240 executing computations on platform Host. Devices:
2020-03-08 17:09:17.960903: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): Host, Default Version
2020-03-08 17:09:17.963622: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-03-08 17:09:17.991113: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:09:17.991937: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce GTX 950M major: 5 minor: 0 memoryClockRate(GHz): 1.124
pciBusID: 0000:01:00.0
2020-03-08 17:09:17.992200: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2020-03-08 17:09:17.994216: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2020-03-08 17:09:17.995631: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2020-03-08 17:09:17.995968: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2020-03-08 17:09:17.997913: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2020-03-08 17:09:17.999453: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2020-03-08 17:09:18.003490: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-03-08 17:09:18.003645: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:09:18.005516: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:09:18.006266: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
2020-03-08 17:09:18.006348: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/tobia/anaconda3/envs/tensorflow_gpuenv/lib/python3.7/site-packages/tensorflow_core/python/client/device_lib.py", line 41, in list_local_devices
for s in pywrap_tensorflow.list_devices(session_config=session_config)
File "/home/tobia/anaconda3/envs/tensorflow_gpuenv/lib/python3.7/site-packages/tensorflow_core/python/pywrap_tensorflow_internal.py", line 2249, in list_devices
return ListDevices()
tensorflow.python.framework.errors_impl.InternalError: cudaGetDevice() failed. Status: CUDA driver version is insufficient for CUDA runtime version
Я также попытался:
import tensorflow as tf
sess = tf.compat.v1.Session()
но получил похожий вывод:
2020-03-08 17:06:40.164183: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-03-08 17:06:40.190642: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.192888: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce GTX 950M major: 5 minor: 0 memoryClockRate(GHz): 1.124
pciBusID: 0000:01:00.0
2020-03-08 17:06:40.193286: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2020-03-08 17:06:40.196146: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2020-03-08 17:06:40.198010: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2020-03-08 17:06:40.198443: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2020-03-08 17:06:40.200486: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2020-03-08 17:06:40.202031: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2020-03-08 17:06:40.206287: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-03-08 17:06:40.206425: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.208358: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.209928: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
2020-03-08 17:06:40.210261: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2020-03-08 17:06:40.235145: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2592000000 Hz
2020-03-08 17:06:40.235690: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x558d9cbb41d0 executing computations on platform Host. Devices:
2020-03-08 17:06:40.235718: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): Host, Default Version
2020-03-08 17:06:40.235879: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.236477: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce GTX 950M major: 5 minor: 0 memoryClockRate(GHz): 1.124
pciBusID: 0000:01:00.0
2020-03-08 17:06:40.236518: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2020-03-08 17:06:40.236530: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2020-03-08 17:06:40.236542: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2020-03-08 17:06:40.236553: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2020-03-08 17:06:40.236565: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2020-03-08 17:06:40.236576: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2020-03-08 17:06:40.236587: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-03-08 17:06:40.236641: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.237220: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-03-08 17:06:40.237863: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
2020-03-08 17:06:40.237903: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/tobia/anaconda3/envs/tensorflow_gpuenv/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1585, in __init__
super(Session, self).__init__(target, graph, config=config)
File "/home/tobia/anaconda3/envs/tensorflow_gpuenv/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 699, in __init__
self._session = tf_session.TF_NewSessionRef(self._graph._c_graph, opts)
tensorflow.python.framework.errors_impl.InternalError: cudaGetDevice() failed. Status: CUDA driver version is insufficient for CUDA runtime version
Так что я делаю не так? Есть что сделать, чтобы решить проблему? Заранее спасибо за помощь!