asum.hpp

Go to the documentation of this file.

#pragma once
 
namespace monolish {
namespace {
template <typename F1> double Dasum_core(const F1 &x) {
  Logger &logger = Logger::get_instance();
  logger.func_in(monolish_func);
 
  double ans = 0;
  const double *xd = x.data();
  size_t size = x.size();
  const size_t xoffset = x.get_offset();
 
  if (x.get_device_mem_stat() == true) {
#if MONOLISH_USE_NVIDIA_GPU
    cublasHandle_t h;
    internal::check_CUDA(cublasCreate(&h));
#pragma omp target data use_device_ptr(xd)
    { internal::check_CUDA(cublasDasum(h, size, xd + xoffset, 1, &ans)); }
    cublasDestroy(h);
#else
    throw std::runtime_error(
        "error USE_GPU is false, but get_device_mem_stat() == true");
#endif
  } else {
    ans = cblas_dasum(size, xd + xoffset, 1);
  }
 
#if MONOLISH_USE_MPI
  mpi::comm &comm = mpi::comm::get_instance();
  ans = comm.Allreduce(ans);
#endif
 
  logger.func_out();
  return ans;
}
 
template <typename F1> float Sasum_core(const F1 &x) {
  Logger &logger = Logger::get_instance();
  logger.func_in(monolish_func);
 
  float ans = 0;
  const float *xd = x.data();
  size_t size = x.size();
  const size_t xoffset = x.get_offset();
 
  if (x.get_device_mem_stat() == true) {
#if MONOLISH_USE_NVIDIA_GPU
    cublasHandle_t h;
    internal::check_CUDA(cublasCreate(&h));
#pragma omp target data use_device_ptr(xd)
    { internal::check_CUDA(cublasSasum(h, size, xd + xoffset, 1, &ans)); }
    cublasDestroy(h);
#else
    throw std::runtime_error(
        "error USE_GPU is false, but get_device_mem_stat() == true");
#endif
  } else {
    ans = cblas_sasum(size, xd + xoffset, 1);
  }
 
#if MONOLISH_USE_MPI
  mpi::comm &comm = mpi::comm::get_instance();
  ans = comm.Allreduce(ans);
#endif
 
  logger.func_out();
  return ans;
}
 
} // namespace
} // namespace monolish