Write matrix size estimate when running model_maker

            write_obj(sconf, gconf, max_rad)

        try:

            max_gpu_ram = int(model['system_settings']['max_gpu_ram'])

            if (max_gpu_ram > 0):

                max_gpu_ram_bytes = max_gpu_ram * 1024 * 1024 * 1024

            matrix_dim = 2 * gconf['nsph'] * gconf['li'] * (gconf['li'] + 2)

            matrix_size_bytes = 16 * matrix_dim * matrix_dim

            matrix_size_Gb = float(matrix_size_bytes) / 1024.0 / 1024.0 / 1024.0

            print("INFO: estimated matrix size is {0:.3g} Gb.".format(matrix_size_Gb))

            if (max_gpu_ram > 0):

                max_gpu_ram_bytes = max_gpu_ram * 1024 * 1024 * 1024

                if (matrix_size_bytes < max_gpu_ram_bytes):

                    max_gpu_processes = int(max_gpu_ram_bytes / matrix_size_bytes)

                    print("INFO: system supports up to %d simultaneous processes on GPU."%max_gpu_processes)

                    print("INFO: only %d GPU processes allowed, if using refinement."%(max_gpu_processes / 3))

                else:

                    print("WARNING: estimated matrix size is larger than available GPU memory!")

            else: