CUdeviceptr dev_ptr;
    bool valid;
} DevicePtrInfo;

static inline DevicePtrInfo getPointer(PyObject *obj, int idx) {
  DevicePtrInfo ptr_info;
  ptr_info.dev_ptr = 0;
  ptr_info.valid = true;
  if (PyLong_Check(obj)) {
    ptr_info.dev_ptr = PyLong_AsUnsignedLongLong(obj);
    return ptr_info;
  }
  if (obj == Py_None) {
    // valid nullptr
    return ptr_info;
  }
  PyObject *ptr = PyObject_GetAttrString(obj, "data_ptr");
  if(ptr){
    PyObject *empty_tuple = PyTuple_New(0);
    PyObject *ret = PyObject_Call(ptr, empty_tuple, NULL);
    Py_DECREF(empty_tuple);
    Py_DECREF(ptr);
    if (!PyLong_Check(ret)) {
      PyErr_SetString(PyExc_TypeError, "data_ptr method of Pointer object must return 64-bit int");
      ptr_info.valid = false;
      return ptr_info;
    }
    ptr_info.dev_ptr = PyLong_AsUnsignedLongLong(ret);
    unsigned attr;
    CUresult status =
        cuPointerGetAttribute(&attr, CU_POINTER_ATTRIBUTE_MEMORY_TYPE, ptr_info.dev_ptr);
    if (ptr_info.dev_ptr &&
        (!(attr == CU_MEMORYTYPE_DEVICE || attr == CU_MEMORYTYPE_UNIFIED) ||
         !(status == CUDA_SUCCESS))) {
        PyErr_Format(PyExc_ValueError,
                     "Pointer argument (at %d) cannot be accessed from Triton (cpu tensor?)", idx);
        ptr_info.valid = false;
    }
    Py_DECREF(ret);  // Thanks ChatGPT!
    return ptr_info;
  }
  PyErr_SetString(PyExc_TypeError, "Pointer argument must be either uint64 or have data_ptr method");
  return ptr_info;
}

static PyObject* launch(PyObject* self, PyObject* args) {
  int gridX, gridY, gridZ;
  uint64_t _stream;
  uint64_t _function;
  int num_warps;
  int shared_memory;
  PyObject *launch_enter_hook = NULL;
  PyObject *launch_exit_hook = NULL;
  PyObject *compiled_kernel = NULL;
  PyObject *hook_ret = NULL;
  r›