Has anyone tryed CUDA on Ubuntu 12.10 with Blender or another programm? In the last beta of Quantal CUDA wasn't working. I would like to update from 12.04 to 12.10 final but only if CUDA still works. Is there someone who already has experience with this? Thanks
Add a comment
|
3 Answers
The installation of CUDA 5.0 fails on ubuntu 12.10 :
$ uname -a
Linux yoyo-laptop 3.5.0-17-generic #28-Ubuntu SMP Tue Oct 9 19:31:23 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 12.10
Release: 12.10
Codename: quantal
$ cd /opt
$ sudo wget -c http://developer.download.nvidia.com/compute/cuda/5_0/rel/installers/cuda_5.0.35_linux_64_ubuntu11.10.run
$ md5sum cuda_5.0.35_linux_64_ubuntu11.10.run
dd82e9d1c5739806690c311e2f56c928 cuda_5.0.35_linux_64_ubuntu11.10.run
$ sha1sum cuda_5.0.35_linux_64_ubuntu11.10.run
bae5870868217bb0bc07d53a7e824301f9752e6a cuda_5.0.35_linux_64_ubuntu11.10.run
$ sudo chmod +x cuda_5.0.35_linux_64_ubuntu11.10.run
$ sudo update-alternatives --install /usr/bin/gcc cc /usr/bin/gcc-4.5 50
$ gcc --version
gcc (Ubuntu/Linaro 4.5.4-1ubuntu2) 4.5.4
$ sudo ln -s /usr/lib/x86_64-linux-gnu/libglut.so /usr/lib/libglut.so
$ sudo ./cuda_5.0.35_linux_64_ubuntu11.10.run -verbose -optimus
Do you accept the previously read EULA? (accept/decline/quit): accept
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 304.54? ((y)es/(n)o/(q)uit): y
Install the CUDA 5.0 Toolkit? ((y)es/(n)o/(q)uit): y
Enter Toolkit Location [ default is /usr/local/cuda-5.0 ]:
Install the CUDA 5.0 Samples? ((y)es/(n)o/(q)uit): y
Enter CUDA Samples Location [ default is /usr/local/cuda-5.0/samples ]:
Installing the NVIDIA display driver...
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 304.54...............................................................................................................................................................
Installing the CUDA Toolkit in /usr/local/cuda-5.0.
Verifying archive integrity... All good.
Uncompressing NVIDIA CUDA..........................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
===========
= Summary =
===========
Driver: Installation Failed
Toolkit: Installation Failed
Samples: Cannot find Toolkit in /usr/local/cuda-5.0
Logfile is /tmp/cuda_install_30203.log
-
For Reference, it works following those instructions: devtalk.nvidia.com/default/topic/522699/… (including changing gcc/g++ to use 4.6 [or older] and the soft link for the glut library otherwise you cannot install the samples; it's all documented on that page.) Dec 27, 2012 at 6:54
First install gcc-4.4 and g++-4.4 then make a symlink of sudo ln -s /usr/bin/gcc /usr/bin/gcc-4.4, it is not too clean but probably you should delete first the symbolic link of /usr/bin/gcc.
-
It worked for me with gcc-4.6. Any reason why you would choose 4.4? Dec 27, 2012 at 6:59
Simplest solution is:
- install blender from repo for dependency purposes
- go to: http://www.blender.org/download/get-blender/ and download zip archive
- unpack to /usr/local/blender-2.64
- start blender typing /usr/local/blender-2.64/blender
It detects then cuda gpu without any problems. (Of course assuming that you have propersly installed latest propietrary nvidia drivers)
Yep, i know that is not that elegant as making it work when installed from repo or ppa, but for me most important is that works, so i can use blender and cycles.
If you dont need the cuda toolkit (for develping etc) i think it is acceptable workaround for you too.
