TK1配置cuda6.5

一、下載

http://developer.download.nvidia.com/embedded/L4T/r21_Release_v3.0/cuda-repo-l4t-r21.3-6-5-prod_6.5-42_armhf.deb

注意：ubuntu14.04 32bit

二、安裝

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

具體請參考以下鏈接:https://developer.nvidia.com/embedded/linux-tegra

三、配置

（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

(補充: 運行yolo)

修改Makefile:
源代碼：
ARCH= –gpu-architecture=compute_52 –gpu-code=compute_52

改爲：
ARCH= –gpu-architecture=compute_20 –gpu-code=compute_20

---

sudo cp /usr/local/cuda-6.5/lib/libcudart.so /usr/local/lib/libcudart.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libcublas.so /usr/local/lib/libcublas.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libcurand.so /usr/local/lib/libcurand.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libcudart.so.6.5 /usr/local/lib/libcudart.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libcublas.so.6.5 /usr/local/lib/libcublas.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libcurand.so.6.5 /usr/local/lib/libcurand.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libnppi.so /usr/local/lib/libcudart.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libnpps.so /usr/local/lib/libcublas.so && sudo ldconfig
sudo cp /usr/local/cuda-6.5/lib/libnppc.so /usr/local/lib/libcurand.so && sudo ldconfig

四、示例

cuda-install-samples-6.5.sh /home/ubuntu
cd ~/NVIDIA_CUDA-6.5_Samples
make