#pragma once
#include <torch/extension.h>
#include <c10/cuda/CUDAGuard.h>
#include <ATen/cuda/CUDAContext.h>
#include <cuda_runtime.h>
#include <cuda_fp16.h>
#include <cstdint>
#include <cstdio>

#include "config.h"
#include "cpp/util.h"

#define TORCH_CHECK_DTYPE(__x, __dtype) TORCH_CHECK((__x).dtype() == torch::__dtype, #__x " is incorrect datatype, must be " #__dtype)
#define TORCH_CHECK_DTYPE_OPT(__x, __dtype) TORCH_CHECK((__x).device().is_meta() || (__x).dtype() == torch::__dtype, #__x " is incorrect datatype, must be " #__dtype)
#define TORCH_CHECK_SHAPES(__x, __dim_x, __y, __dim_y, __scale_y) TORCH_CHECK((__x).size(__dim_x) == (__y).size(__dim_y) * __scale_y, #__x " and " #__y " have incompatible shapes")
#define TORCH_CHECK_SHAPES_OPT(__x, __dim_x, __y, __dim_y, __scale_y) TORCH_CHECK((__x).device().is_meta() || (__x).size(__dim_x) == (__y).size(__dim_y) * __scale_y, #__x " and " #__y " have incompatible shapes")