OpenCL basic examples

here is my feeble attempt at learning OpenCL, please don't make fun of me too much 🍔

Configuration

This code uses OpenCL 1.1 on a NVIDIA GPU.

Linux

(Only tested on Ubuntu). For NVIDIA GPUs, I've installed the following packages: nvidia-346 nvidia-346-dev nvidia-346-uvm nvidia-libopencl1-346 nvidia-modprobe nvidia-opencl-icd-346 nvidia-settings. Since the opencl-headers package in the main repository is for OpenCL 1.2, you can get the OpenCL 1.1 header files from here.

Then to compile the C++ code:

g++ -std=c++0x main.cpp -o main.out -lOpenCL

To compile the C code:

gcc main.c -o main.out -lOpenCL

For examples 04 and 05, you can run

make ex04
make ex05
make       # makes both!

OS X

OpenCL is installed on OS X by default, but since this code uses the C++ bindings, you'll need to get that too. Get the official C++ bindings from the OpenCL registr and copy it to the OpenCL framework directory, or do the following:

wget https://www.khronos.org/registry/cl/api/1.1/cl.hpp
sudo cp cl.hpp /System/Library/Frameworks/OpenCL.framework/Headers/

To compile:

clang++ -std=c++0x -framework OpenCL main.cpp -o main.out

example 00

this example is based off of this example (example-ception), but it goes a bit further. In the blogspot example, two 10-element vectors are created and a thread is used for each pair of elements. In this example, 10 threads are spawned but two 100-element vectors are used, and it is shown how to split up a specific number of elements per thread.

example 01

Measures the duration of adding two vectors. See the README in the folder for more details.

example 02

Demonstrates that one array can be modified several times without having to re-read and re-write data to and from the GPU.

example 03

A simple example using the cl_khr_fp64 extension which allows for usage of doubles instead of floats.

example 04

An example of the CLFFT library for an in-place complex-planar transform. There is also Python code to check the answer, which requires numpy / scipy. The C code requires the CLFFT library to be installed in the root of the repository. See more details in the folder's readme.

example 05

Another CLFFT example where an in-place real transform is performed. There's also Python code for checking the answer, which requires numpy / scipy. The C code requires the CLFFT library to be installed in the root of the repository. For instructions on doing this, check out the readme of example04.

Some Notes

From the guide on programming OpenCL for NVIDIA:

• CUDA streaming multiprocessor corresponds to an OpenCL compute unit
• CUDA thread corresponds to an OpenCL work-item
• CUDA thread block corresponds to an OpenCL work-group