testC++

Building C++ Unit Tests for PYNQ: CMake based sds++ cross-compilation

List of Supported Components

Image Arithmetic	Filters	Geometric Transform	Features	Input Processing	Analysis	Flow and Depts
absdiff	filter2D	remap	canny	* split	calcHist	stereoBM
bitwise_and	dilate	resize	cornerHarris	* merge	equalizeHist	calcOpticalFlowDenseNonPyrLK
bitwise_or	erode	warpAffine	fast		integral
bitwise_xor	medianBlur	warpPerspective			meanStdDev
bitwise_not	boxFilter				minMaxLoc
threshold	pyrUp				LUT
add	pyrDown
subtract
multiply
accumulate
accumulateWeighted
accumulateSquare
magnitude
phase

* These designs require a patch to xfopencv (tag: 2018.2_release) in order to build properly. Please clone https://github.com/denolf/xfopencv.git and checkout the v2018.2Fixed branch to pick up a temporary fix.

Setup Environment on Host

• clone Pynq-ComputerVision repository:

  $ git clone https://github.com/Xilinx/PYNQ-ComputerVision.git <your_pynqcv_folder>
  

• clone xfOpenCV repository and checkout the correct version (e.g. 2018.2_release):

  $ git clone https://github.com/Xilinx/xfopencv.git <your_xfopencv_folder>
  $ cd <your_xfopencv_folder>
  $ git checkout <release_number>
  

• Prepare the Ultra96 bare platform package in /your_PynqPlatform_folder.
  • download ultra96_platform_sysroot_2018.2.tar.gz in /your_PynqPlatform_folder
  • untar the package. This will create an ultra subfolder in /your_PynqPlatform_folder

    $ tar -zxvf ultra96_platform_sysroot_2018.2.tar.gz
    

  • Note that you can adapt /your_PynqPlatform_folder to your preference, but the deepest subfolder should match the platform name, in this case 'ultra'.
• set an environmental variable to xfOpenCV

  $ setenv XFOPENCV_PATH <your_xfopencv_folder>
  

• set up Xilinx SDx tools, version 2018.2 by running its setup script

Building unit tests

Individual unit tests

• For an indidivual unit test, navigate to the the test you wish to build (e.g. testXfFilter2D), create a folder there and run cmake.

  $ cd /<your_pynqcv_folder>/applicationCode/unitTests/testC++/testXfFilter2D
  $ mkdir build; cd build
  $ cmake .. -DCMAKE_TOOLCHAIN_FILE=../../../frameworks/cmakeModules/toolchain_sdx2018.2.cmake -DSDxPlatform=/your_PynqPlatform_folder/ultra -DSDxClockID=1 -DusePL=ON -DnoBitstream=OFF -DnoSDCardImage=ON -DSDxArch=arm64
  

• Note that the clock ID for the platform selects the desired clock frequency for the design. In the case of the Ultra96 platform, the following IDs are available: 0=100MHz, 1=150MHz, 2=250MHz, 3=300MHz.
• run make for the unit test

  $ make testSDxXfFiter2D
  

• The bitstream needs to be reformatted to the format used to program the FPGA on the board. This is done by first creating a simple .bif file (conv.bif) as shown below:

  all:
  {
       [destination_device = pl] testSDxXfFilter2D.bit
  }
  

  Then you run bootgen as follows:

  $ bootgen -image conv.bif -arch zynqmp -o ./testSDxXfFilter2D.bit.bin -w
  

• Copy the reformatted bitstream and executable to a test folder (for instance ~/proj/test) on your pynq board:

  $  scp testSDxXfFilter2D.bit.bin xilinx@<pynq-board-ip>:/home/xilinx/proj/test
  $  scp testSDxXfFilter2D xilinx@<pynq-board-ip>:/home/xilinx/proj/test
  

Entire unit test suite

• To build all the available tests, you can run the build script under applicationCode/unitTests/testC++. Be sure to edit the buildUnitTest.py so the platform points to /ultra.

  $  cd /<your_pynqcv_folder>/applicationCode/unitTests/testC++/
  $  ./buildUnitTest.py
  

• This creates individual reformatted bitstream and executables for each test and copies them to applicationCode/unitTests/testC++/unitTestFiles. You can then the files onto the board using the same method as in the individual test case
• You can also pass in the name of text file that newline separate list the units you wish to build as follows:

  $  ./buildUnitTest.py -l myTestList
  

Running unit test on Board

• To run the unit tests, program the FPGA with the target bitstream (.bin) with the sudo command on the board

  $ cd ~/proj/test
  $ su
  $ echo 0 > /sys/class/fpga_manager/fpga0/flags
  $ cp testSDxXfFilter2D.bit.bin /lib/firmware
  $ echo testSDxXfFilter2D.bit.bin > /sys/class/fpga_manager/fpga0/firmware
  

• Prior to running the tests, the required image files must first be present (e.g. bigBunny_1080.png). These files can be copied to the local directory on the board such as ~/proj/images. Then, the individual tests can be executed in sudo mode or while still under su mode from before.

  $ sudo ./testSDxXfFilter2D ~/proj/images/bigBunny_1080.png
  

  you should see ~140fps measured result.