External IO C++ based generator

This example will demonstrate emulation of Xilinx Device’s IO with a C++ based external process .

KEY CONCEPTS: Traffic Generator, Traffic generation using C++

KEYWORDS: sim_ipc_axis_master, sim_ipc_axis_slave, gt_master, gt_slave

Some user applications make use of IO ports in Vitis flow to drive streaming traffic in and out of the platform . This example emulates such designs in HW-Emulation by mimicing the IO port behavior as in HW.

In this example, Master and Slave Stream IO models will be available as prepacked xo’s with different data width configuration. In order to instantiate the IP’s corresponding to the provided xo, the xos are specified in the v++ command line

XOS = $(XILINX_VITIS)/data/emulation/XO/sim_ipc_axis_master_32.xo  $(XILINX_VITIS)/data/emulation/XO/sim_ipc_axis_slave_32.xo
$(VPP) $(VPP_FLAGS) -l $(VPP_LDFLAGS) $(XOS) --temp_dir $(BUILD_DIR)  -o'$(BUILD_DIR)/increment.link.xclbin' $(+)

The C++ based Traffic Generator follows the following steps for blocking traffic generation, transportation and reception -

  1. Instantiate IPC socket with name matching in IPI diagram

xtlm_ipc::axis_initiator_socket_util<xtlm_ipc::BLOCKING> socket_util("gt_master");
xtlm_ipc::axis_target_socket_util<xtlm_ipc::BLOCKING> socket_util("gt_slave");

  1. Determine the number of transactions and create that many random packets

std::vector<char> data = generate_data();

//! Set packet attributes...
packet.set_data(data.data(), data.size());
packet.set_data_length(data.size());
packet.set_tlast(1);

  1. Send the data over ipc socket. Here the function call is blocked until response is available.

socket_util.transport(packet);

  1. Sample the data available on the socket.

socket_util.sample_transaction(packet)

  1. End the simulation using the following API.

socket_util.end_of_simulation();

The C++ based Traffic Generator follows the following steps for Non-blocking traffic generation, transportation and reception -

  1. Instantiate IPC socket with name matching in IPI diagram

xtlm_ipc::axis_initiator_socket_util<xtlm_ipc::NON_BLOCKING> socket_util("gt_master");
xtlm_ipc::axis_target_socket_util<xtlm_ipc::NON_BLOCKING> socket_util("gt_slave");

  1. Determine the number of transactions and create that many random data points

//! Initialize vector size...
std::vector<char> v_data(data_length);

//! Fill the data in vector...
std::generate(v_data.begin(), v_data.end(), []() { return std::rand() % 256; });

  1. Send the data over ipc socket. Here the function returns immediately without waiting for a response.

socket_util.transport(data.data(), data.size());

  1. Sample the data available on the socket.

socket_util.sample_transaction(data);

  1. End the simulation using the following API.

socket_util.end_of_simulation();

Connections to the Custom IP’s specified in xo’s are made using krnl_incr.cfg file as below:

[connectivity]
nk=sim_ipc_axis_master:1:gt_master
nk=sim_ipc_axis_slave:1:gt_slave
stream_connect=gt_master.M00_AXIS:increment_1.a
stream_connect=increment_1.output:gt_slave.S00_AXIS

Note : xo instance name specified on v++ command line becomes a key which will be used by the external process to refer to a particular stream port

EXCLUDED PLATFORMS:

  • All Xilinx DMA Platforms

  • Versal VCK5000

  • AWS VU9P F1

  • Samsung SmartSSD Computation Storage Drive

  • Samsung U.2 SmartSSD

DESIGN FILES

Application code is located in the src directory. Accelerator binary files will be compiled to the xclbin directory. The xclbin directory is required by the Makefile and its contents will be filled during compilation. A listing of all the files in this example is shown below

src/b_receive.cpp
src/b_send.cpp
src/host.cpp
src/increment.cpp
src/nb_receive.cpp
src/nb_send.cpp

Access these files in the github repo by clicking here.

COMMAND LINE ARGUMENTS

Once the environment has been configured, the application can be executed by

./external_io_cpp <increment XCLBIN>