OpenCL Practice Repository

• Using OpenCL CPP, the following examples are implemented

-[x] Vector addition
-[ ] Regression
-[ ] Convolution

Installing OpenCL on Ubuntu 22.04

• Verify that you have a CUDA-capable GPU

lspci | grep -i nvidia

• Remove existing nvidia driver

dpkg -l | grep nvidia-driver
sudo apt purge nvidia-driver-xxx
sudo apt autoremove
sudo apt autoclean

• Install the latest nvidia drivers for your system

sudo apt update
sudo apt install nvidia-driver-xxx
sudo reboot

• Check OpenCL version

/usr/bin/clinfo -l

• Install the cuda toolkit

sudo apt update && sudo apt upgrade
sudo apt install nvidia-cuda-toolkit

• Check cuda install

nvcc --version

Running OpenCL CPP code

• On Ubuntu 22.04

• On terminal 1

lttng-sessiond daemonize
lttng create vadd_session
lttng enable-event --userspace vadd:*
lttng start

• On terminal 2

cd ~
git clone https://github.com/acceleration-robotics/opencl-practice.git
cd opencl-practice/vector_add/cpu_vadd_tracing
make all
./cpu_vadd
cd ../gpu_vadd_tracing
make all
./gpu_vadd

• On terminal 1 again

lttng stop
babeltrace ~/lttng-traces/vadd_session*

Running Image Processing examples

1. Copy Image

• On Ubuntu 22.04

• On terminal

cd ~/opencl-practice/image_processing
make all
./read_image <image_name> 

• <image_name> can be :

1. image1.png
2. image2.png
3. None : leave blank for default image ("lenna.png")

• Compare outputs in input_images and output_images of same name

• To cleanup :

make clean