Error deepspeech-gpu

Deep Speech
#1

DeepSpeech works well.
But deepspeech-gpu gives below mistake

deepspeech -h
Traceback (most recent call last):
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/site-packages/deepspeech/impl.py”, line 14, in swig_import_helper
return importlib.import_module(mname)
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/importlib/init.py”, line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File “”, line 994, in _gcd_import
File “”, line 971, in _find_and_load
File “”, line 955, in _find_and_load_unlocked
File “”, line 658, in _load_unlocked
File “”, line 571, in module_from_spec
File “”, line 922, in create_module
File “”, line 219, in _call_with_frames_removed
ImportError: libcublas.so.10.0: cannot open shared object file: No such file or directory

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File “/home/farruh/tmp/deepspeech-gpu-venv/bin/deepspeech”, line 5, in
from deepspeech.client import main
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/site-packages/deepspeech/init.py”, line 23, in
from deepspeech.impl import PrintVersions as printVersions
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/site-packages/deepspeech/impl.py”, line 17, in
_impl = swig_import_helper()
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/site-packages/deepspeech/impl.py”, line 16, in swig_import_helper
return importlib.import_module(’_impl’)
File “/home/farruh/tmp/deepspeech-gpu-venv/lib/python3.6/importlib/init.py”, line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
ModuleNotFoundError: No module named ‘_impl’

Version on deep speech is:
pip list | grep deepspeech
deepspeech-gpu 0.6.0

Cuda version:
nvcc --version
nvcc: NVIDIA ® Cuda compiler driver
Copyright © 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:24:38_PDT_2019
Cuda compilation tools, release 10.2, V10.2.89

Tensorflow version:
pip3 list | grep tensorflow
tensorflow-estimator 1.14.0
tensorflow-gpu 1.14.0

#2

As documented, you need CUDA 10.0 accessible by the GPU runtime. You can just download and install in your home directory, and then use LD_LIBRARY_PATH to make it accessible.

#3

Got you, I will try with CUDA 10. When update will come that can works with cuda 10.2.89?

#4

Never, this depends on TensorFlow, and even r1.15 is CUDA 10.0.

#5

@farruh As I said, you can always do a local installation of CUDA.

#6

@lissyx Do you have a tutorial of how to do a local installation of CUDA?

#7

No, but it’s really simple, and i think it’s documented by NVIDIA. Just download the tarball, extract it to e.g., $HOME/CUDA/ and then set LD_LIBRARY_PATH accordingly.

You might also need to install cudnn in this place as well.

1 Like