Linux Cuda - failed call to cuInit: CUDA_ERROR_NO_DEVICE

痛定思痛。 2021-11-24 01:50 379阅读 0赞

wu@wu-X555LF:~$ nvidia-smi
    Sat Jul 27 14:07:04 2019       
    +-----------------------------------------------------------------------------+
    | NVIDIA-SMI 384.130                Driver Version: 384.130                   |
    |-------------------------------+----------------------+----------------------+
    | GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
    | Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
    |===============================+======================+======================|
    |   0  GeForce 930M        Off  | 00000000:04:00.0 Off |                  N/A |
    | N/A   45C    P0    N/A /  N/A |    191MiB /  2002MiB |      0%      Default |
    +-------------------------------+----------------------+----------------------+
                                                                                   
    +-----------------------------------------------------------------------------+
    | Processes:                                                       GPU Memory |
    |  GPU       PID   Type   Process name                             Usage      |
    |=============================================================================|
    |    0      1002      G   /usr/lib/xorg/Xorg                           142MiB |
    |    0      1985      G   compiz                                        45MiB |
    |    0      2705      G   /usr/lib/firefox/firefox                       1MiB |
    +-----------------------------------------------------------------------------+

查看电脑的GPU配置状况，只有1块GPU，序号为 0

import tensorflow as tf 
    import os 
    a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
    b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b') 
    c = tf.matmul(a, b) 
    os.environ['CUDA_VISIBLE_DEVICES'] = "1" 
    sess = tf.Session() 
    print (sess.run(c)) 
    sess.close()

wu@wu-X555LF:~$ python os.environ\[\'CUDA_VISIBLE_DEVICES\'\]\ \=\ \"1\".py 
    2019-07-27 14:05:52.372040: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
    2019-07-27 14:05:52.376105: E tensorflow/stream_executor/cuda/cuda_driver.cc:406] failed call to cuInit: CUDA_ERROR_NO_DEVICE
    2019-07-27 14:05:52.376169: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:158] retrieving CUDA diagnostic information for host: wu-X555LF
    2019-07-27 14:05:52.376182: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:165] hostname: wu-X555LF
    2019-07-27 14:05:52.376221: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:189] libcuda reported version is: 384.130.0
    2019-07-27 14:05:52.376323: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:193] kernel reported version is: 384.130.0
    2019-07-27 14:05:52.376334: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:300] kernel version seems to match DSO: 384.130.0
    [[22. 28.]
     [49. 64.]]

会显示出failed call to cuInit: CUDA\_ERROR\_NO\_DEVICE，没用用到GPU设备

需要修改os.environ\['CUDA\_VISIBLE\_DEVICES'\] = "1"

为os.environ\['CUDA\_VISIBLE\_DEVICES'\] = "0"

序号要对应上nvidia-smi命令查找的GPU配置序号

import tensorflow as tf 
    import os 
    a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
    b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b') 
    c = tf.matmul(a, b) 
    os.environ['CUDA_VISIBLE_DEVICES'] = "0" 
    sess = tf.Session() 
    print (sess.run(c)) 
    sess.close()

wu@wu-X555LF:~$ python os.environ\[\'CUDA_VISIBLE_DEVICES\'\]\ \=\ \"1\".py 
    2019-07-27 14:07:07.830300: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
    2019-07-27 14:07:07.931219: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:898] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
    2019-07-27 14:07:07.931666: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1212] Found device 0 with properties: 
    name: GeForce 930M major: 5 minor: 0 memoryClockRate(GHz): 0.941
    pciBusID: 0000:04:00.0
    totalMemory: 1.96GiB freeMemory: 1.74GiB
    2019-07-27 14:07:07.931696: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1312] Adding visible gpu devices: 0
    2019-07-27 14:07:08.557549: I tensorflow/core/common_runtime/gpu/gpu_device.cc:993] Creating TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 1504 MB memory) -> physical GPU (device: 0, name: GeForce 930M, pci bus id: 0000:04:00.0, compute capability: 5.0)
    [[22. 28.]
     [49. 64.]]
    wu@wu-X555LF:~$