cuDNN/CUDA/TensorFlow setup prroblem

Accelerated Computing CUDA CUDA Setup and Installation
1

Hello,
Here is my setup keypoints:

  • CUDA: version 10.2

  • cuDNN: version 7.6

  • NVIDIA driver: version 440

  • TensorFlow: version 2.0

When I check my GPU from TensorFlow, I get this:

import tensorflow as tf
tf.test.is_gpu_available()
2020-03-16 13:50:11.216347: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2020-03-16 13:50:11.279329: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2299965000 Hz
2020-03-16 13:50:11.280490: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x4a0dea0 executing computations on platform Host. Devices:
2020-03-16 13:50:11.280547: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): Host, Default Version
2020-03-16 13:50:11.286390: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2020-03-16 13:50:11.381738: 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
2020-03-16 13:50:11.382114: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x4a9d180 executing computations on platform CUDA. Devices:
2020-03-16 13:50:11.382136: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): GeForce GTX 1050, Compute Capability 6.1
2020-03-16 13:50:11.382311: 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
2020-03-16 13:50:11.382557: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce GTX 1050 major: 6 minor: 1 memoryClockRate(GHz): 1.493
pciBusID: 0000:01:00.0
2020-03-16 13:50:11.382686: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcudart.so.10.0’; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.382791: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcublas.so.10.0’; dlerror: libcublas.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.382877: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcufft.so.10.0’; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.382958: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcurand.so.10.0’; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.383037: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcusolver.so.10.0’; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.383115: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcusparse.so.10.0’; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory
2020-03-16 13:50:11.439296: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2020-03-16 13:50:11.439360: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1641] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at Install TensorFlow with pip for how to download and setup the required libraries for your platform.
Skipping registering GPU devices…
2020-03-16 13:50:11.439396: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-03-16 13:50:11.439415: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165] 0
2020-03-16 13:50:11.439430: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0: N
False

Thanks in advance for any suggestion.

2

It appears to be looking for the CUDA 10.0 libraries. You have CUDA 10.2 installed.

3

I am in complete confusion regarding NVIDIA setup. Someone on another forum advised me to use Anaconda, which apparently is able to manage all of these interdependencies between different packages. But probably before proceeding with Anaconda it will be necessary to uninstall everything - cuDNN, NVIDIA drivers, CUDA, TensirFlow. For example here is how the system reacted when I try to upgrade:

pavel@ALABAMA:~$ sudo apt-get upgrade
Reading package lists... Done
Building dependency tree       
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
 nvidia-cuda-toolkit : Depends: nvidia-cuda-dev (= 9.1.85-3ubuntu1) but it is not installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or specify a solution).
pavel@ALABAMA:~$ apt --fix-broken install
E: Could not open lock file /var/lib/dpkg/lock-frontend - open (13: Permission denied)
E: Unable to acquire the dpkg frontend lock (/var/lib/dpkg/lock-frontend), are you root?
pavel@ALABAMA:~$ sudo apt --fix-broken install
Reading package lists... Done
Building dependency tree       
Reading state information... Done
Correcting dependencies... Done
The following additional packages will be installed:
  nvidia-cuda-dev
Recommended packages:
  libnvcuvid1
The following NEW packages will be installed:
  nvidia-cuda-dev
0 upgraded, 1 newly installed, 0 to remove and 13 not upgraded.
36 not fully installed or removed.
Need to get 0 B/263 MB of archives.
After this operation, 734 MB of additional disk space will be used.
Do you want to continue? [Y/n] y
(Reading database ... 310808 files and directories currently installed.)
Preparing to unpack .../nvidia-cuda-dev_9.1.85-3ubuntu1_amd64.deb ...
Unpacking nvidia-cuda-dev (9.1.85-3ubuntu1) ...
dpkg: error processing archive /var/cache/apt/archives/nvidia-cuda-dev_9.1.85-3ubuntu1_amd64.deb (--unpack):
 trying to overwrite '/usr/include/cublas.h', which is also in package libcublas-dev 10.2.2.89-1
dpkg-deb: error: paste subprocess was killed by signal (Broken pipe)
Errors were encountered while processing:
 /var/cache/apt/archives/nvidia-cuda-dev_9.1.85-3ubuntu1_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)
pavel@ALABAMA:~$

So, how to proceed to remove whole this staff (i.e. drivers, CUDA, etc.)
Thanks.