#include <cstdio>
#include <iostream>
#define PRINTF_BINARY_PATTERN_INT8 "%c%c%c%c%c%c%c%c"
#define PRINTF_BYTE_TO_BINARY_INT8(i)                                          \
  (int)((((i) >> 7) & 1ull) + '0'), (int)((((i) >> 6) & 1ull) + '0'),          \
      (int)((((i) >> 5) & 1ull) + '0'), (int)((((i) >> 4) & 1ull) + '0'),      \
      (int)((((i) >> 3) & 1ull) + '0'), (int)((((i) >> 2) & 1ull) + '0'),      \
      (int)((((i) >> 1) & 1ull) + '0'), (int)((((i)) & 1ull) + '0')

#define PRINTF_BINARY_PATTERN_INT16                                            \
  PRINTF_BINARY_PATTERN_INT8 PRINTF_BINARY_PATTERN_INT8
#define PRINTF_BYTE_TO_BINARY_INT16(i)                                         \
  PRINTF_BYTE_TO_BINARY_INT8((i) >> 8), PRINTF_BYTE_TO_BINARY_INT8(i)
#define PRINTF_BINARY_PATTERN_INT32                                            \
  PRINTF_BINARY_PATTERN_INT16 PRINTF_BINARY_PATTERN_INT16
#define PRINTF_BYTE_TO_BINARY_INT32(i)                                         \
  PRINTF_BYTE_TO_BINARY_INT16((i) >> 16), PRINTF_BYTE_TO_BINARY_INT16(i)
#define PRINTF_BINARY_PATTERN_INT64                                            \
  PRINTF_BINARY_PATTERN_INT32 PRINTF_BINARY_PATTERN_INT32
#define PRINTF_BYTE_TO_BINARY_INT64(i)                                         \
  PRINTF_BYTE_TO_BINARY_INT32((i) >> 32), PRINTF_BYTE_TO_BINARY_INT32(i)
/* --- end macros --- */

__global__ void test() {

  auto last_known_point = 10981552822ULL, val = 10960094149ULL;
  auto x = last_known_point - val;
  printf("last_known_point =%llu, val = %llu " PRINTF_BINARY_PATTERN_INT64 "\n",
         last_known_point, val,
         PRINTF_BYTE_TO_BINARY_INT64(last_known_point - val));
}

int main() {
  std::cout << (PRINTF_BINARY_PATTERN_INT64) << std::endl;
  test<<<1, 1>>>();
  cudaDeviceSynchronize();
}