3

Ubuntu 18.04 TLS has moved to CUDA 9.1 now, which deprecates the Nvidia Fermi Architecture.

I have tried installing the cuda 8 toolkit from nvidia with an --override option (to discard compiler verification) butthe installation failed as follows:

Command: sudo ./cuda_8.0.61_375.26_linux.run --override --toolkitpath /usr/local/cuda-8.0/

Output:

You are attempting to install on an unsupported configuration. Do you wish to continue?
(y)es/(n)o [ default is no ]: y

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 375.26?
(y)es/(n)o/(q)uit: n

Install the CUDA 8.0 Toolkit?
(y)es/(n)o/(q)uit: y

Enter Toolkit Location
 [ default is /usr/local/cuda-8.0 ]: /usr/local/cuda-8.0

Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y

Install the CUDA 8.0 Samples?
(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location
 [ default is /home/user ]: 

Installing the CUDA Toolkit in /usr/local/cuda-8.0 ...
Warning: cannot find Toolkit in /usr/local/cuda-8.0. Use --toolkitpath to specify the toolkit location.
Installing the CUDA Samples in /home/user ...
sh: 1: /usr/local/cuda-8.0/bin/cuda-install-samples-8.0.sh: not found
chown: cannot access '/home/user/NVIDIA_CUDA-8.0_Samples': No such file or directory

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installation Failed
Samples:  Installation Failed
Blue Giant
  • 81
  • 1
    Apparently, you need to specify the toolkit location using option --toolkitpath. – francescop21 Oct 22 '18 at 16:25
  • I have already tried. Added my command in the post. – Blue Giant Oct 22 '18 at 16:34
  • Actually, you don't need that option, since you're installing in the default directory. Have you uninstalled any previous version of Cuda Toolkit? – francescop21 Oct 22 '18 at 17:16
  • I have tried without the option --toolkitpath, and I was getting the same error. Indeed, I have installed the default nvidia-cuda-toolkit 9.1 from the ubuntu 18.04 repos. But I removed it after realizing my architecture was deprecated. – Blue Giant Oct 22 '18 at 18:17
  • Make sure that there are no files remaining from the cuda 9.1 installation. – francescop21 Oct 23 '18 at 08:39

2 Answers2

5

I have solved the issue using the following procedure:

  1. unpack .run file using the command: ./cuda*.run --tar mxvf (or sh ./cuda*.run --tar mxvf)
  2. copy unpacked file InstallUtils.pm to /usr/lib/x86_64-linux-gnu/perl-base
  3. enter command in terminal: export $PERL5LIB
Community
  • 1
Blue Giant
  • 81
1

There are two cleaner ways to solve this issue:

  1. Set current directory as perl library search path by: PERL5LIB=. sh cuda_8.0.61_375.26_linux.run your-other-options. Related link.
  2. Install the missing perl module by: sudo apt install module-install-perl. Related link
John Smith
  • 11