Difference between revisions of "Jetson/Installing CUDA"

From eLinux.org
Jump to: navigation, search
(Installing the CUDA Toolkit onto your device for native CUDA development: Removed version dependency)
(Installing the CUDA Toolkit onto your device for native CUDA development: Shown new method to install CUDA samples directly on the board)
 
(7 intermediate revisions by the same user not shown)
Line 3: Line 3:
 
* '''cross-compilation''' (compiling code on an x86 desktop in a special way so it can execute on the [[Jetson TK1]] target device).
 
* '''cross-compilation''' (compiling code on an x86 desktop in a special way so it can execute on the [[Jetson TK1]] target device).
  
Native compilation is generally the easiest option, but takes longer to compile, whereas cross-compilation is typically more complex to configure and debug, but for large projects it will be noticeably faster at compiling. The CUDA Toolkit currently only supports cross-compilation from an Ubuntu 12.04 Linux desktop. In comparison, native compilation happens onboard the Jetson device and thus is the same no matter which OS or desktop you have.
+
Native compilation is generally the easiest option, but takes longer to compile, whereas cross-compilation is typically more complex to configure and debug, but for large projects it will be noticeably faster at compiling. The CUDA Toolkit currently only supports cross-compilation from an Ubuntu 12.04 or 14.04 Linux desktop. In comparison, native compilation happens onboard the Jetson device and thus is the same no matter which OS or desktop you have.
  
 
== Installing the CUDA Toolkit onto your device for native CUDA development ==
 
== Installing the CUDA Toolkit onto your device for native CUDA development ==
Download the .deb file for the [https://developer.nvidia.com/jetson-tk1-support CUDA Toolkit for L4T]. (Make sure you download the ''Toolkit for L4T'' and not the ''Toolkit for Ubuntu'' since that is for cross-compilation instead of native compilation). You will need to register & log in first before downloading, so the easiest way is perhaps to download the file on your PC. Then if you want to copy the file to your device you can copy it onto a USB flash stick then plug it into the device, or transfer it through your local network such as by running this on a Linux PC:
+
Download the .deb file for the [https://developer.nvidia.com/jetson-tk1-support CUDA Toolkit for L4T] either using a web browser on the device, or download on your PC then copy the file to your device using a USB flash stick or across the network. (Make sure you download the ''Toolkit for L4T'' and not the ''Toolkit for Ubuntu'' since that is for cross-compilation instead of native compilation). Some more direct links for Jetson TK1: [https://developer.nvidia.com/linux-tegra-rel-21 CUDA 6.5 Toolkit for L4T Rel 21.2], and [https://developer.nvidia.com/cuda-toolkit-65#linux-arm CUDA 6.5 Toolkit] including the [https://developer.download.nvidia.com/compute/cuda/6_5/rel/docs/CUDA_Getting_Started_Linux.pdf CUDA 6.5 Getting Started Linux Guide].
scp ~/Downloads/cuda-repo-l4t-r19.2_6.0-42_armhf.deb ubuntu@tegra-ubuntu:Downloads/.
+
 
 
On the device, install the .deb file and the CUDA Toolkit. eg:
 
On the device, install the .deb file and the CUDA Toolkit. eg:
 
  cd ~/Downloads
 
  cd ~/Downloads
Line 14: Line 14:
 
  # Download & install the actual CUDA Toolkit including the OpenGL toolkit from NVIDIA. (It only downloads around 15MB)
 
  # Download & install the actual CUDA Toolkit including the OpenGL toolkit from NVIDIA. (It only downloads around 15MB)
 
  sudo apt-get update
 
  sudo apt-get update
  sudo apt-get install cuda-toolkit-6-0
+
# Install "cuda-toolkit-6-0" if you downloaded CUDA 6.0, or "cuda-toolkit-6-5" if you downloaded CUDA 6.5, etc.
 +
  sudo apt-get install cuda-toolkit-6-5
 +
# Install the package full of CUDA samples (optional)
 +
sudo apt-get install cuda-samples-6-5
 
  # Add yourself to the "video" group to allow access to the GPU
 
  # Add yourself to the "video" group to allow access to the GPU
 
  sudo usermod -a -G video $USER
 
  sudo usermod -a -G video $USER
Line 24: Line 27:
 
Verify that the CUDA Toolkit is installed on your device:
 
Verify that the CUDA Toolkit is installed on your device:
 
  nvcc -V
 
  nvcc -V
 +
(note that the above flag is a capital "V" not lower-case "v").
  
 
== Installing & running the CUDA samples (optional) ==
 
== Installing & running the CUDA samples (optional) ==
 
If you think you will write your own CUDA code or you want to see what CUDA can do, then follow this section to build & run all of the CUDA samples.
 
If you think you will write your own CUDA code or you want to see what CUDA can do, then follow this section to build & run all of the CUDA samples.
  
Install writeable copies of the CUDA samples to your device's home directory (it will create a "NVIDIA_CUDA-6.0_Samples" folder):
+
Install writeable copies of the CUDA samples to your device's home directory (it will create a "NVIDIA_CUDA-6.5_Samples" folder):
  cuda-install-samples-6.0.sh /home/ubuntu
+
  cuda-install-samples-6.5.sh /home/ubuntu
 
Build the CUDA samples (takes around 15 minutes on [[Jetson TK1]]):
 
Build the CUDA samples (takes around 15 minutes on [[Jetson TK1]]):
  cd ~/NVIDIA_CUDA-6.0_Samples
+
  cd ~/NVIDIA_CUDA-6.5_Samples
 
  make
 
  make
 
Run some CUDA samples:
 
Run some CUDA samples:
Line 56: Line 60:
 
Note: Many of the CUDA samples use OpenGL GLX and open graphical windows. If you are running these programs through an SSH remote terminal, you can remotely display the windows on your desktop by typing "export DISPLAY=:0" and then executing the program. (This will only work if you are using a Linux/Unix machine or you run an X server such as the free "Xming" for Windows). eg:
 
Note: Many of the CUDA samples use OpenGL GLX and open graphical windows. If you are running these programs through an SSH remote terminal, you can remotely display the windows on your desktop by typing "export DISPLAY=:0" and then executing the program. (This will only work if you are using a Linux/Unix machine or you run an X server such as the free "Xming" for Windows). eg:
 
  export DISPLAY=:0
 
  export DISPLAY=:0
  cd ~/NVIDIA_CUDA-6.0_Samples/2_Graphics/simpleGL
+
  cd ~/NVIDIA_CUDA-6.5_Samples/2_Graphics/simpleGL
 
  ./simpleGL
 
  ./simpleGL
  cd ~/NVIDIA_CUDA-6.0_Samples/3_Imaging/bicubicTexture
+
  cd ~/NVIDIA_CUDA-6.5_Samples/3_Imaging/bicubicTexture
 
  ./bicubicTexture
 
  ./bicubicTexture
  cd ~/NVIDIA_CUDA-6.0_Samples/3_Imaging/bilateralFilter
+
  cd ~/NVIDIA_CUDA-6.5_Samples/3_Imaging/bilateralFilter
 
  ./bilateralFilter
 
  ./bilateralFilter
Note: the Optical Flow sample (HSOpticalFlow) and 3D stereo sample (stereoDisparity) take rougly 1 minute each to execute since they compare results with CPU code.
+
Note: the Optical Flow sample (HSOpticalFlow) and 3D stereo sample (stereoDisparity) take rouglhy 1 minute each to execute since they compare results with CPU code.
  
 
Some of the CUDA samples use other libraries such as OpenMP or MPI or OpenGL.  
 
Some of the CUDA samples use other libraries such as OpenMP or MPI or OpenGL.  

Latest revision as of 05:00, 14 November 2018

You have two options for developing CUDA applications for Jetson TK1:

  • native compilation (compiling code onboard the Jetson TK1)
  • cross-compilation (compiling code on an x86 desktop in a special way so it can execute on the Jetson TK1 target device).

Native compilation is generally the easiest option, but takes longer to compile, whereas cross-compilation is typically more complex to configure and debug, but for large projects it will be noticeably faster at compiling. The CUDA Toolkit currently only supports cross-compilation from an Ubuntu 12.04 or 14.04 Linux desktop. In comparison, native compilation happens onboard the Jetson device and thus is the same no matter which OS or desktop you have.

Installing the CUDA Toolkit onto your device for native CUDA development

Download the .deb file for the CUDA Toolkit for L4T either using a web browser on the device, or download on your PC then copy the file to your device using a USB flash stick or across the network. (Make sure you download the Toolkit for L4T and not the Toolkit for Ubuntu since that is for cross-compilation instead of native compilation). Some more direct links for Jetson TK1: CUDA 6.5 Toolkit for L4T Rel 21.2, and CUDA 6.5 Toolkit including the CUDA 6.5 Getting Started Linux Guide.

On the device, install the .deb file and the CUDA Toolkit. eg:

cd ~/Downloads
# Install the CUDA repo metadata that you downloaded manually for L4T
sudo dpkg -i cuda-repo-l4t-r19.2_6.0-42_armhf.deb
# Download & install the actual CUDA Toolkit including the OpenGL toolkit from NVIDIA. (It only downloads around 15MB)
sudo apt-get update
# Install "cuda-toolkit-6-0" if you downloaded CUDA 6.0, or "cuda-toolkit-6-5" if you downloaded CUDA 6.5, etc.
sudo apt-get install cuda-toolkit-6-5
# Install the package full of CUDA samples (optional)
sudo apt-get install cuda-samples-6-5
# Add yourself to the "video" group to allow access to the GPU
sudo usermod -a -G video $USER

Add the 32-bit CUDA paths to your .bashrc login script, and start using it in your current console:

echo "# Add CUDA bin & library paths:" >> ~/.bashrc
echo "export PATH=/usr/local/cuda/bin:$PATH" >> ~/.bashrc
echo "export LD_LIBRARY_PATH=/usr/local/cuda/lib:$LD_LIBRARY_PATH" >> ~/.bashrc
source ~/.bashrc

Verify that the CUDA Toolkit is installed on your device:

nvcc -V

(note that the above flag is a capital "V" not lower-case "v").

Installing & running the CUDA samples (optional)

If you think you will write your own CUDA code or you want to see what CUDA can do, then follow this section to build & run all of the CUDA samples.

Install writeable copies of the CUDA samples to your device's home directory (it will create a "NVIDIA_CUDA-6.5_Samples" folder):

cuda-install-samples-6.5.sh /home/ubuntu

Build the CUDA samples (takes around 15 minutes on Jetson TK1):

cd ~/NVIDIA_CUDA-6.5_Samples
make

Run some CUDA samples:

1_Utilities/deviceQuery/deviceQuery
1_Utilities/bandwidthTest/bandwidthTest
cd 0_Simple/matrixMul
./matrixMulCUBLAS
cd ../..
cd 0_Simple/simpleTexture
./simpleTexture
cd ../..
cd 3_Imaging/convolutionSeparable
./convolutionSeparable
cd ../..
cd 3_Imaging/convolutionTexture
./convolutionTexture
cd ../..

Note: Many of the CUDA samples use OpenGL GLX and open graphical windows. If you are running these programs through an SSH remote terminal, you can remotely display the windows on your desktop by typing "export DISPLAY=:0" and then executing the program. (This will only work if you are using a Linux/Unix machine or you run an X server such as the free "Xming" for Windows). eg:

export DISPLAY=:0
cd ~/NVIDIA_CUDA-6.5_Samples/2_Graphics/simpleGL
./simpleGL
cd ~/NVIDIA_CUDA-6.5_Samples/3_Imaging/bicubicTexture
./bicubicTexture
cd ~/NVIDIA_CUDA-6.5_Samples/3_Imaging/bilateralFilter
./bilateralFilter

Note: the Optical Flow sample (HSOpticalFlow) and 3D stereo sample (stereoDisparity) take rouglhy 1 minute each to execute since they compare results with CPU code.

Some of the CUDA samples use other libraries such as OpenMP or MPI or OpenGL.

If you want to compile those samples then you'll need to install these toolkits like this:

(to be added)