cpp11thread.cu

#include <future>
#include <thread>
#include <map>
#include <vector>
#include <stdio.h>

const int N = 1 << 20;

__global__ void kernel(float *x, int n)
{
  int tid = threadIdx.x + blockIdx.x * blockDim.x;
  for (int i = tid; i < n; i += blockDim.x * gridDim.x) {
    x[i] = sqrt(pow(3.14159,i));
  }
}

void launch_kernel(std::promise<void*> p)
{
  try {
    float *data;
    cudaMalloc(&data, N * sizeof(float));

    kernel<<<1, 64>>>(data, N);

    cudaStreamSynchronize(0);
    p.set_value(NULL);
  } catch (...) {
    p.set_exception(std::current_exception());
  }
}

int main()
{
  const int num_threads = 8;
  std::vector<std::thread> threads;
  std::map<std::thread, std::future<void*>> futures;
    

  for (auto& t: threads) {
    try {
      std::primise<void*> p;
      futures.insert(std::make_pair(t, p.get_future()));
      t = std::thread(launch_kernel, std::move(p));
    } catch(...) {
      fprintf(stderr, "Error creating thread\n");
      return 1;
    }
  }

  for (auto& t: threads) {
    t.join();
    auto& f = futures[t];
    try {
      f.get();
    } catch(std::exception& e) {
      fprintf(stderr, "Error joining thread\n" + e.what());
    }

  }

  cudaDeviceReset();
  return 0;
}