示例#1
0
文件: xasum.hpp 项目: gcp/CLBlast
 // Describes how to run the CLBlast routine
 static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
   auto queue_plain = queue();
   auto event = cl_event{};
   auto status = Asum<T>(args.n,
                         buffers.scalar(), args.asum_offset,
                         buffers.x_vec(), args.x_offset, args.x_inc,
                         &queue_plain, &event);
   clWaitForEvents(1, &event);
   return status;
 }
示例#2
0
文件: xdotu.hpp 项目: gpu/CLBlast
 // Describes how to run the CLBlast routine
 static StatusCode RunRoutine(const Arguments<T> &args, Buffers<T> &buffers, Queue &queue) {
   #ifdef OPENCL_API
     auto queue_plain = queue();
     auto event = cl_event{};
     auto status = Dotu<T>(args.n,
                           buffers.scalar(), args.dot_offset,
                           buffers.x_vec(), args.x_offset, args.x_inc,
                           buffers.y_vec(), args.y_offset, args.y_inc,
                           &queue_plain, &event);
     if (status == StatusCode::kSuccess) { clWaitForEvents(1, &event); clReleaseEvent(event); }
   #elif CUDA_API
     auto status = Dotu<T>(args.n,
                           buffers.scalar(), args.dot_offset,
                           buffers.x_vec(), args.x_offset, args.x_inc,
                           buffers.y_vec(), args.y_offset, args.y_inc,
                           queue.GetContext()(), queue.GetDevice()());
     cuStreamSynchronize(queue());
   #endif
   return status;
 }