Batch Load Speed From Disk Benchmarking

benchmark_iteration_speed.py

def run_one_epoch(data, conn):
    epoch_start = time.time()
    for batch in data:
        pass
    epoch_end = time.time()

    conn.send(round(epoch_end - epoch_start, 6))


def measure_data_load_speed(data, num_epochs):
    """
    :param data: can be a DataSet or a DataLoader
    """ 
    epoch_times = []
    for i in range(num_epochs):
        parent_conn, child_conn = Pipe()
        p = Process(target=run_one_epoch, args=(data, child_conn))
        p.start()
        epoch_time = parent_conn.recv()
        p.join()
        epoch_times.append(epoch_time)

    results = {'data': epoch_times,
               'avg': round(stats.mean(epoch_times), 4),
               'std': round(stats.stdev(epoch_times), 4)}

    return results