Getting error while installing JetPack on Jetson TK1

I am trying to download the JetPack 2.0 on my Jetson TK1 but while installing it an error accures in the post installation step, and shows an error in the matrixMul Sample.
It saya"no matrixMul samples and no such file and directory, please press enter to continue"
I pressed enter but nothing happened and the installation stays uncompleted.

How to solve the error? Since i need to se the kit urgently for a project that im working on.

I’ll appreciate your help, thank you.


It seems you also posted in another thread, and been answered, see:

Does it help?


yes it did, thanks.
but I need also to install the CUDA, so I tried to download it separately (following this tutorial: )but I faced a this problem


root@tegra-ubuntu:/home/ubuntu# sudo apt-get install cuda-toolkit-6-5 -y
Reading package lists… Done
Building dependency tree
Reading state information… Done
cuda-toolkit-6-5 is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 496 not upgraded.
root@tegra-ubuntu:/home/ubuntu# sudo usermod -a -G video $USER
root@tegra-ubuntu:/home/ubuntu# #Add the 32-bit CUDA paths to your .bashrc login script, and start using it in your current console:
root@tegra-ubuntu:/home/ubuntu# echo “# Add CUDA bin & library paths:” >> ~/.bashrc
root@tegra-ubuntu:/home/ubuntu# echo “export PATH=/usr/local/cuda-6.5/bin:$PATH” >> ~/.bashrc
root@tegra-ubuntu:/home/ubuntu# echo “export LD_LIBRARY_PATH=/usr/local/cuda-6.5/lib:$LD_LIBRARY_PATH” >> ~/.bashrc
root@tegra-ubuntu:/home/ubuntu# source ~/.bashrc
root@tegra-ubuntu:/home/ubuntu# # cd /usr/local/cuda-6.5/bin
root@tegra-ubuntu:/home/ubuntu# ls
cuda-l4t OpenCV4Tegra
cuda-repo-l4t-r21.1-6-5-prod_6.5-14_armhf.deb PerfKit
cudnn Pictures
Desktop Public
GameWorksOpenGLSamples tegrastats
libvisionworks-repo_1.0.25-919_armhf_l4t_r21.deb Templates
libvisionworks-sfm-repo_0.80.8_armhf_l4t_r21.deb Videos
Music VisionWorks-SFM-0.80-Samples


I’ll appreciate your help, thanks.

When I have installed JTK1 manually, and have a clean install, here is the order of doing a few things manually (I have the .deb packages in the same directory, running sudo -s):

dpkg -i cuda-repo-l4t-r21.3-6-5-prod_6.5-42_armhf.deb
apt update

# (apt-get install cuda-toolkit-6-5...this is all one long line)
apt-get install cuda-toolkit-6-5 debian-keyring g++-multilib g++-4.8-multilib gcc-4.8-doc libstdc++6-4.8-dbg libice-doc libsm-doc libstdc++-4.8-doc libxcb-doc libxext-doc libxt-doc libsfstdc++6-4.8-dbg libsfstdc++6-4.8-dbg

# username typically is "ubuntu".
usermod -a -G video <username>

dpkg -i libopencv4tegra-repo_l4t-r21_2.4.10.1_armhf.deb
apt update

apt-get install libopencv4tegra libopencv4tegra-dev
# (apt-get autoremove ...unused...)

# For camera, apt-get install luvcview

so I do this after installing the JTK1? and after entering this code the CUDA will be downloaded?

The steps I listed are for manual install of other packages on JTK1 which has been freshly flashed with L4T (freshly flashed isn’t mandatory, just needing those other packages usually occurs right after flash). The CUDA repo package adds the repository which has CUDA (note “repo” in that .deb file name). The “apt update” line essentially gives apt a list of what packages are relevant. The “apt-get” line mentioning the actual CUDA package installs the actual CUDA downloaded from the CUDA repo installed at the first step. You could stop after the usermod step and CUDA would be installed on the JTK1.