How to write configuration file to reduce CPU usage while DiscoveryServerExample‘s publisher are used？

[greenhandZ]

background

device A: a embeded divice
device B: ubuntu 22 AMD cpu

	device type	hardware	software	program
device A	A embeded device	rv1106(ARMv7 Processor rev 5 (v7l)), 182MB memory	ubuntu22	fast-discovery-server -i 0 -t 0.0.0.0 -q 42100 DiscoveryServerExample publisher -c 127.0.0.1 -p 42100 --transport=tcpv4 --interval=100 --data-size=61440
device B	A normal PC	x86 AMD 5600 CPU, 32GB memory	ubuntu22	DiscoveryServerExample subscriber -c 192.168.31.78 -p 42100 --transport=tcpv4

modify publisher QOS

according to https://fast-dds.docs.eprosima.com/en/stable/fastdds/dynamic_types/dynamic_types.html
15.1
15.4
15.6

// participant
    DomainParticipantQos pqos;
  pqos.allocation().participants = eprosima::fastrtps::
        ResourceLimitedContainerConfig::fixed_size_configuration(3u);
    pqos.allocation().writers = eprosima::fastrtps::
        ResourceLimitedContainerConfig::fixed_size_configuration(3U);
    pqos.allocation().readers = eprosima::fastrtps::
        ResourceLimitedContainerConfig::fixed_size_configuration(3u);
    pqos.allocation().data_limits.max_user_data = 256u;
    pqos.allocation().data_limits.max_partitions = 256u;
    pqos.allocation().data_limits.max_properties = 512u;
    pqos.allocation().content_filter.expression_initial_size = 512u;
    pqos.allocation().content_filter.expression_parameters = eprosima::fastrtps
        ::ResourceLimitedContainerConfig::fixed_size_configuration(4u);
    std::cout << "send_socket_buffer_size:"<< pqos.transport().send_socket_buffer_size
        << std::endl ;
    std::cout << "listen_socket_buffer_size:" << pqos.transport().listen_socket_buffer_size
        << std::endl;
    BuiltinTransportsOptions large_data_options;
    large_data_options.maxMessageSize = 200000;
    large_data_options.sockets_buffer_size = 200000;
    large_data_options.non_blocking_send = true;
    large_data_options.tcp_negotiation_timeout = 0;//ms
    pqos.setup_transports(BuiltinTransports::LARGE_DATA, large_data_options);

// datawriter
    DataWriterQos wqos;
    wqos.writer_resource_limits().matched_subscriber_allocation = eprosima::
        fastrtps::ResourceLimitedContainerConfig::fixed_size_configuration(3u);
    wqos.writer_resource_limits().reader_filters_allocation = eprosima::fastrtps
        ::ResourceLimitedContainerConfig::fixed_size_configuration(1u);

CPU usage test on device A

one package size	period	Ethernet	cpu usage with native tcp	device A cpu usage,before modify QOS	device A cpu usage, after modify QOS
61440B	100ms	600KiB/s	0.7%	3~5%	1~2%
61440B	50ms	1.2MiB/s	<1%	5~7.8%	2~3%
61440B	10ms	5.9MiB/s	about 3%	17~20%	about 10%

question

How to write configuration file to reduce CPU usage on device A?
How to modify QoS with CPU usage approaching TCP load?