1. flash the TX2

download jetpack L4T 3.0

• do not run the script in sudo mode when installing host components
• do not install OpenCV4Tegra V2.4 for TX2, we will use OpenCV3
• for TX2 you may need to flash it tiwce, one just flash the OS, the other install cuda
• if the tx1/tx2 cannot boot the GUI, but shut down after finish loading BIOS, try to change a more powerful power source

basic tools

sudo apt-get install terminator -y

sudo apt-get install git

sudo apt-get install cmake -y

sudo apt-get install vim -y

sudo apt-get install htop

2. Eigen manually

install Eigen stable release V3.3.3

• http://eigen.tuxfamily.org/index.php?title=Main_Page

• it will fail 6 ctest cases out of 798

• or if you build BTL, fail 13 ctest cases out of 832

host:

scp ~/Downloads/eigen-eigen-67e894c6cd8f.tar.bz2 ubuntu@<YOUR_SSH PATH>:~/

TX1: cmake .. -DCMAKE_INSTALL_PREFIX=/usr/local -DCMAKE_INSTALL_PREFIX=/usr/local -DEIGEN_TEST_CXX11=ON -DEIGEN_CUDA_COMPUTE_ARCH=53 -DEIGEN_BUILD_BTL=ON

TX2: cmake .. -DCMAKE_INSTALL_PREFIX=/usr/local -DCMAKE_INSTALL_PREFIX=/usr/local -DEIGEN_TEST_CXX11=ON -DEIGEN_CUDA_COMPUTE_ARCH=62 -DEIGEN_BUILD_BTL=ON

make check

TX1 result:

Label Time Summary:
Official       = 618.29 sec (660 tests)
Unsupported    = 294.54 sec (164 tests)

Total Test time (real) = 963.72 sec

The following tests FAILED:
	384 - qr_colpivoting_3 (OTHER_FAULT)
	479 - bdcsvd_9 (OTHER_FAULT)
	661 - failtests (Failed)
	662 - NonLinearOptimization (OTHER_FAULT)
	684 - matrix_function_1 (OTHER_FAULT)
	714 - sparse_extra_3 (OTHER_FAULT)
	826 - btl_eigen3_linear (Not Run)
	827 - btl_eigen3_vecmat (Not Run)
	828 - btl_eigen3_matmat (Not Run)
	829 - btl_eigen3_adv (Not Run)
	830 - btl_tensor_linear (Not Run)
	831 - btl_tensor_vecmat (Not Run)
	832 - btl_tensor_matmat (Not Run)

this step will take really a long time

3. ceres manly

normal install, without eigen, need pass all ctest cases

typical Ceres test bench mark is 131s for TX2

ceres-solver.org/ceres-solver-1.12.0.tar.gz

sudo apt-get install libgoogle-glog-dev

sudo apt-get install libatlas-base-dev -y

sudo apt-get install libsuitesparse-dev -y

tar zxf ceres-solver-1.12.0.tar.gz

mkdir ceres-bin

cd ceres-bin

cmake ../ceres-solver-1.12.0 -DCMAKE_INSTALL_PREFIX=/usr/local

make -j4

make test

TX1 test result:L 100% tests passed, 0 tests failed out of 68

Total Test time (real) = 392.52 sec

make install

4. Download buildOpencvTX2, remove libeigen3-dev, install Opencv manuly

https://github.com/jetsonhacks/buildOpenCVTX2/blob/master/buildOpenCV.sh

Enable opengl, add -WITH_OPENGL=ON sudo apt-get install qtbase5-dev https://devtalk.nvidia.com/default/topic/965134/jetson-tx1/opencv-3-1-compilation-on-tx1-lets-collect-the-/post/4983805/#4983805

patch opengl headers: https://devtalk.nvidia.com/default/topic/1007290/jetson-tx2/building-opencv-with-opengl-support-/post/5141945/#5141945

in opencv V3.2.0 opencv_test_cudev will fail, but acturally, it success, ctest made a wrong test compare between 2.0 and 2.

TX1 result:

Label Time Summary:
Accuracy                 = 2187.12 sec (27 tests)
Main                     = 6063.93 sec (71 tests)
Performance              = 3202.87 sec (22 tests)
Sanity                   = 673.94 sec (22 tests)
opencv_calib3d           = 303.71 sec (3 tests)
opencv_core              = 1201.93 sec (3 tests)
opencv_cudaarithm        = 203.99 sec (3 tests)
opencv_cudabgsegm        =  10.81 sec (3 tests)
opencv_cudacodec         =   0.45 sec (3 tests)
opencv_cudafeatures2d    =  24.15 sec (3 tests)
opencv_cudafilters       = 137.54 sec (3 tests)
opencv_cudaimgproc       = 371.67 sec (3 tests)
opencv_cudalegacy        = 137.41 sec (3 tests)
opencv_cudaobjdetect     =  11.42 sec (3 tests)
opencv_cudaoptflow       =  28.72 sec (3 tests)
opencv_cudastereo        =  25.07 sec (3 tests)
opencv_cudawarping       = 605.62 sec (3 tests)
opencv_cudev             =   1.99 sec (1 test)
opencv_features2d        = 125.05 sec (3 tests)
opencv_flann             =   0.02 sec (1 test)
opencv_highgui           =   0.05 sec (1 test)
opencv_imgcodecs         =  75.52 sec (3 tests)
opencv_imgproc           = 761.49 sec (3 tests)
opencv_ml                =  84.47 sec (1 test)
opencv_objdetect         =  22.82 sec (3 tests)
opencv_photo             = 236.25 sec (3 tests)
opencv_shape             = 1089.02 sec (1 test)
opencv_stitching         = 328.15 sec (3 tests)
opencv_superres          =  30.62 sec (3 tests)
opencv_video             = 124.65 sec (3 tests)
opencv_videoio           = 121.32 sec (3 tests)

Total Test time (real) = 6064.09 sec

The following tests FAILED:
	  1 - opencv_test_cudev (Failed)
	 17 - opencv_perf_cudabgsegm (Failed)
	 23 - opencv_perf_cudaimgproc (Failed)
	 26 - opencv_perf_cudawarping (Failed)
	 27 - opencv_sanity_cudawarping (Failed)
	 35 - opencv_test_videoio (Failed)
	 41 - opencv_test_highgui (Failed)
	 54 - opencv_test_cudalegacy (Failed)
	 55 - opencv_perf_cudalegacy (Failed)
	 67 - opencv_perf_stitching (Failed)
	 68 - opencv_sanity_stitching (Failed)

5. ros-desktop source install, remove eigen & opencv3

ROSTX2

you could install ROS through apt, but it might conflict with OPENCV3-CUDA

1 if ROS deb init fall try

sudo c_rehash /etc/ssl/certs

2 to enable best performence mode use:

~$ sudo nvpmodel -m 0

~$ sudo ./jecson_clock.sh

SDL

`sudo apt-get install libsdl-image1.2-dev

and

sudo apt-get install libsdl-dev

some useful scripts

monitor the CPU temperature

watch -n 1 cat /sys/devices/virtual/thermal/thermal_zone1/temp

check CUDA version

nvcc -V

Enable /dev/ttyTHS2 (UART1) for TX2

install dtc

sudo apt-get install device-tree-compiler

rewrite kernel

enable tty THS2

sudo -s
cd /tmp
dtc -I dtb -O dts -o extracted.dts /boot/tegra186-quill-p3310-1000-c03-00-base.dtb
# Search for "serial@c280000" where it is a block of code and not just a single line...
# Change status = "disabled" to status = "okay";
# Build a modified version:
dtc -I dts -O dtb -o /boot/modified_tegra186-quill-p3310-1000-c03-00-base.dtb extracted.dts
cd /boot/extlinux
# edit extlinux.conf...add this line between MENU LABEL line and LINUX line:
FDT /boot/modified_tegra186-quill-p3310-1000-c03-00-base.dtb

enable CAN buss

ref: http://www.jetsonhacks.com/2017/03/25/build-kernel-and-modules-nvidia-jetson-tx2/

$ git clone https://github.com/jetsonhacks/buildJetsonTX2Kernel.git

use Manifold2 board USBs

todo write notes

test Ethernet speeed

$ sudo apt-get install iperf
#start server
$ sudo iperf -s
#start client
$ sudo iperf -c 192.168.1.xxx -i 5

install SSH keys

ssh-copy-id user@123.45.56.78

bash tools

devlist

devlist is

aliased to `sudo arp-scan --interface=eth0 192.168.1.0/24 | grep NVIDIA'

dev

devlist is aliased to `sudo arp-scan --interface=eth0 192.168.1.0/24 | grep NVIDIA' dev is a function

dev () 
{ 
    local ip="$(sudo arp-scan --interface=eth0 192.168.1.0/24 | grep NVIDIA | sed -n $1p | cut -d'	' -f 1)";
    echo "ssh into $ip";
    ssh nvidia@$ip
}

max_performance.sh

nvidia@tegra-ubuntu:~$ cat max_performance.sh 
echo "Enabling fan for safety..."
if [ ! -w /sys/kernel/debug/tegra_fan/target_pwm ] ; then
echo "Cannot set fan -- exiting..."
fi
echo 255 > /sys/kernel/debug/tegra_fan/target_pwm

nvpmodel -m 0
./jetson_clocks.sh