How to run program with use NCCL backend (torch.distributed)

I am trying to run a project that use nccl architecture from pytorch(multiprocessing architecture) could I run this project with only one GPU? If the answer is No could I run with one system with 2 graphic card on it with SLI or I should run this project in some different PC? some of script python in this project is as follows :

import torch
from torch import distributed as dist


def synchronize():
    if not dist.is_nccl_available():
        return
    if not dist.is_initialized():
        return

    world_size = dist.get_world_size()

    if world_size == 1:
        return

    dist.barrier()


def get_rank():
    if not dist.is_nccl_available():
        return 0
    if not dist.is_initialized():
        return 0
    return dist.get_rank()


def is_main_process():
    return get_rank() == 0


def get_world_size():
    if not dist.is_nccl_available():
        return 1
    if not dist.is_initialized():
        return 1
    return dist.get_world_size()


def broadcast_tensor(tensor, src=0):
    world_size = get_world_size()
    if world_size < 2:
        return tensor

    with torch.no_grad():
        dist.broadcast(tensor, src=0)

    return tensor


def broadcast_scalar(scalar, src=0, device="cpu"):
    if get_world_size() < 2:
        return scalar
    scalar_tensor = torch.tensor(scalar).long().to(device)
    scalar_tensor = broadcast_tensor(scalar_tensor, src)
    return scalar_tensor.item()


def reduce_tensor(tensor):
    world_size = get_world_size()

    if world_size < 2:
        return tensor

    with torch.no_grad():
        dist.reduce(tensor, dst=0)
        if dist.get_rank() == 0:
            tensor = tensor.div(world_size)

    return tensor


def gather_tensor(tensor):
    world_size = get_world_size()

    if world_size < 2:
        return tensor

    with torch.no_grad():
        tensor_list = []

        for _ in range(world_size):
            tensor_list.append(torch.zeros_like(tensor))

        dist.all_gather(tensor_list, tensor)
        tensor_list = torch.stack(tensor_list, dim=0)
    return tensor_list


def reduce_dict(dictionary):
    world_size = get_world_size()
    if world_size < 2:
        return dictionary

    with torch.no_grad():
        if len(dictionary) == 0:
            return dictionary

        keys, values = zip(*sorted(dictionary.items()))
        values = torch.stack(values, dim=0)

        dist.reduce(values, dst=0)

        if dist.get_rank() == 0:
            # only main process gets accumulated, so only divide by
            # world_size in this case
            values /= world_size
        reduced_dict = {k: v for k, v in zip(keys, values)}
    return reduced_dict


def print_only_main(string):
    if is_main_process():
        print(string)