cuda::experimental::launch

Defined in include/cuda/experimental/__launch/launch.cuh

template<typename ...Args, typename ...Config, typename Dimensions, typename Kernel>
void cuda::experimental::launch(::cuda::stream_ref stream, const kernel_config<Dimensions, Config...> &conf, const Kernel &kernel, Args&&... args)

Launch a kernel functor with specified configuration and arguments.

Launches a kernel functor object on the specified stream and with specified configuration. Kernel functor object is a type with device operator(). Functor might or might not accept the configuration as its first argument.

Snippet

#include <cstdio>
#include <cuda/experimental/launch.cuh>

struct kernel {
    template <typename Configuration>
    __device__ void operator()(Configuration conf, unsigned int thread_to_print) {
        if (conf.dims.rank(cudax::thread, cudax::grid) == thread_to_print) {
            printf("Hello from the GPU\n");
        }
    }
};

void launch_kernel(cuda::stream_ref stream) {
    auto dims    = cudax::make_hierarchy(cudax::block_dims<128>(), cudax::grid_dims(4));
    auto confing = cudax::make_config(dims, cudax::launch_cooperative());

    cudax::launch(stream, config, kernel(), 42);
}

Parameters

• stream – cuda::stream_ref to launch the kernel into

• conf – configuration for this launch

• kernel – kernel functor to be launched

• args – arguments to be passed into the kernel functor