/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

#pragma once
#include <cuda.h>

namespace paddle {
namespace platform {

#define CUDA_1D_KERNEL_LOOP(i, n)                            \
  for (int i = blockIdx.x * blockDim.x + threadIdx.x; i < n; \
       i += blockDim.x * gridDim.x)

#define CUDA_ATOMIC_WRAPPER(op, T) \
  __device__ __forceinline__ T CudaAtomic##op(T* address, const T val)

#define USE_CUDA_ATOMIC(op, T) \
  CUDA_ATOMIC_WRAPPER(op, T) { return atomic##op(address, val); }

// For atomicAdd.
USE_CUDA_ATOMIC(Add, float);

#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 600
USE_CUDA_ATOMIC(Add, double);
#else
// Custom implementation of atomicAdd for double.
// This implementation is copied from CUDA manual.
CUDA_ATOMIC_WRAPPER(Add, double) {
  unsigned long long int* address_as_ull =
      reinterpret_cast<unsigned long long int*>(address);
  unsigned long long int old = *address_as_ull, assumed;

  do {
    assumed = old;
    old = atomicCAS(address_as_ull, assumed,
                    __double_as_longlong(val + __longlong_as_double(assumed)));

    // Note: uses integer comparison to avoid hang in case of NaN
  } while (assumed != old);

  return __longlong_as_double(old);
#endif
}

}  // namespace platform
}  // namespace paddle