Author Topic: Boost.Compute: GPU computing library for C++ based on OpenCL  (Read 1166 times)



0 Members and 1 Guest are viewing this topic.

JeGX

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1266
    • View Profile
    • Geeks3D.com
Boost.Compute is a GPU/parallel-computing library for C++ based on OpenCL.

The core library is a thin C++ wrapper over the OpenCL API and provides access to compute devices, contexts, command queues and memory buffers.

On top of the core library is a generic, STL-like interface providing common algorithms (e.g. transform(), accumulate(), sort()) along with common containers (e.g. vector<T>, flat_set<T>). It also features a number of extensions including parallel-computing algorithms (e.g. exclusive_scan(), scatter(), reduce()) and a number of fancy iterators (e.g. transform_iterator<>, permutation_iterator<>, zip_iterator<>).

Links:
- https://github.com/boostorg/compute/
- http://boostorg.github.io/compute/


The development branch brings the full support of OpenCL 2.1:
https://github.com/boostorg/compute/tree/develop



Code sample:
Code: [Select]
#include <vector>
#include <algorithm>
#include <boost/compute.hpp>

namespace compute = boost::compute;

int main()
{
    // get the default compute device
    compute::device gpu = compute::system::default_device();

    // create a compute context and command queue
    compute::context ctx(gpu);
    compute::command_queue queue(ctx, gpu);

    // generate random numbers on the host
    std::vector<float> host_vector(1000000);
    std::generate(host_vector.begin(), host_vector.end(), rand);

    // create vector on the device
    compute::vector<float> device_vector(1000000, ctx);

    // copy data to the device
    compute::copy(
        host_vector.begin(), host_vector.end(), device_vector.begin(), queue
    );

    // sort data on the device
    compute::sort(
        device_vector.begin(), device_vector.end(), queue
    );

    // copy data back to the host
    compute::copy(
        device_vector.begin(), device_vector.end(), host_vector.begin(), queue
    );

    return 0;
}