Download cuDNN issues

On Ubuntu 20.4, I am trying to download cuDNN.

$ nvidia-smi
Sat Oct 24 10:16:44 2020       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 455.23.05    Driver Version: 455.23.05    **CUDA Version: 11.1**     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce RTX 208...  On   | 00000000:0B:00.0  On |                  N/A |
| 18%   36C    P8     1W / 250W |    241MiB / 11011MiB |      1%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      3607      G   /usr/lib/xorg/Xorg                 35MiB |
|    0   N/A  N/A      7468      G   /usr/lib/xorg/Xorg                129MiB |
|    0   N/A  N/A      7670      G   /usr/bin/gnome-shell               41MiB |
|    0   N/A  N/A      9445      G   /usr/lib/firefox/firefox            3MiB |
|    0   N/A  N/A      9651      G   /usr/lib/firefox/firefox            3MiB |
|    0   N/A  N/A      9733      G   /usr/lib/firefox/firefox            3MiB |
|    0   N/A  N/A     12954      G   /usr/lib/firefox/firefox            3MiB |
|    0   N/A  N/A     13171      G   /usr/lib/firefox/firefox            3MiB |
|    0   N/A  N/A     21844      G   /usr/lib/firefox/firefox            3MiB |
+-----------------------------------------------------------------------------+

First, CUDA is the newest version 11.1, which does not have a corresponding cuDNN yet according to https://developer.nvidia.com/rdp/cudnn-archive#a-collapse803-110. So, I downloaded cudnn-11.0-linux-x64-v8.0.3.33.tgz. Would this be OK, or I need to install CUDA 11.0 instead? Second, after I extracted cudnn-11.0-linux-x64-v8.0.3.33.tgz,

I did:

 
$ sudo cp cuda/include/cudnn*.h /usr/local/cuda/include
 
$ sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
 
$ sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*

Adding this line at the end of ~/.bashrc:

export CUDA_HOME=/usr/local/cuda

 
$ source ~/.bashrc

/etc/ld.so.conf.d/999_cuda-11-1.conf was edited from:

/usr/local/cuda-11.1/targets/x86_64-linux/lib

to

/usr/local/cuda/lib64
/usr/local/cuda/extras/CUPTI/lib64

 
$ sudo ldconfig

However,

 $ cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
has no output. I do have:
/usr/local/cuda/include$ ls -l
total 81
-rw-r--r-- 1 root root 29025 Oct 24 20:39 cudnn_adv_infer.h
-rw-r--r-- 1 root root 27700 Oct 24 20:39 cudnn_adv_train.h
-rw-r--r-- 1 root root 12527 Oct 24 20:39 cudnn_backend.h
-rw-r--r-- 1 root root 29011 Oct 24 20:39 cudnn_cnn_infer.h
-rw-r--r-- 1 root root 10177 Oct 24 20:39 cudnn_cnn_train.h
-rw-r--r-- 1 root root  2968 Oct 24 20:39 cudnn.h
-rw-r--r-- 1 root root 48825 Oct 24 20:39 cudnn_ops_infer.h
-rw-r--r-- 1 root root 25733 Oct 24 20:39 cudnn_ops_train.h
-rw-r--r-- 1 root root  2785 Oct 24 20:39 cudnn_version.h

Did I install cuDNN cuccessfully? Thank you!

In addition, insider the Docker container:

λ f0cfbaac32eb / cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
cat: /usr/local/cuda/include/cudnn.h: No such file or directory

I plan to run DeepSpeech2 in a Docker container. Any comments? Thanks.

$ nvidia-smi
tells that
 CUDA Version: 11.1 

However,

$ cat /usr/local/cuda/version.txt
CUDA Version 10.1.243

Why it tells two versions? And

λ f0cfbaac32eb / cat /usr/local/cuda/version.txt
CUDA Version 9.0.176

When I run Deepspeech2 via a Docker container, which CUDA will be used, if there are more than one version have been installed? Thanks.

There is a forum section dedicated to cuDNN, my suggestion would be to post your questions there:

https://forums.developer.nvidia.com/c/ai-deep-learning/libraries-sdk/cudnn/90

Thank Robert_Crovella for the helpful comment! Any comments regarding it shows different versions of CUDA? Thanks.

This covers it, I believe.