Software to AXI4-Stream

Stream AXI4 data from software to FPGA

Since R2022b

• 

Libraries:
SoC Blockset / Memory

Description

The Software to AXI4-Stream block models a connection between hardware logic and a software task through external memory. The writer (processor) streams data into the channel through a DMA driver using a MathWorks^® simplified AXI stream protocol. The block models the datapath and software stack of that connection, including a FIFO, DMA engine, interconnect and external memory, interrupts, kernel buffer management of the DMA driver, and data transfers from the software task.

This block is equivalent to a Memory Channel block with the Channel type parameter set to Software to AXI4-Stream via DMA connected to a Memory Controller block.

Since the memory controller is implicit to the design - you can instantiate several memory blocks that connect to the same memory unit via a memory controller. Valid blocks are:

• AXI4 Random Access Memory

• AXI4-Stream to Software

• Software to AXI4-Stream

• AXI4 Video Frame Buffer

• Memory Traffic Generator

The maximum number of manager interfaces in a model is 16.

For more information about the MathWorks simplified AXI stream protocol, see AXI4-Stream Interface.

Examples

Streaming Data from Software to Hardware

Design the data-path from an embedded processor to hardware logic (FPGA) using SoC Blockset™. Design and simulate the entire application comprising of FPGA and processor algorithms, memory interface, and task scheduling to meet the system requirements. Then, validate the design on hardware by generating code from the model and implementing it on a System-on-Chip (SoC) device.

• Open Script

Ports

Input

msg — Input data signal from software

scalar | vector | matrix

This signal contains the data sent from the processor to the algorithm.

Data Types: SoCData

rdCtrlIn — Reader input control signal

bus

This port accepts a control bus from a data consumer block, signaling that the consumer block is ready to accept read data. The rdCtrlIn port is a backpressure signal from a data consumer to the Software to AXI4-Stream block. To create this control bus, use the SoC Bus Creator block.

Data Types: StreamS2MBusObj

Output

rdData — Output data signal to data consumer

scalar | vector | matrix

This signal contains the data read from the processor through the memory.

Data Types: single | double | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64 | Boolean | fixed point

rdCtrlOut — Reader control signal to data consumer

bus

This signal represents the protocol bus from the memory channel to the data consumer. Connect this signal to the data consumer. To separate the signal from the control bus, use the SoC Bus Selector block.

Data Types: StreamM2SBusObj

event — Task write event signal

scalar

This port sends a task event signal that triggers the Task Manager block to execute the associated event-driven write.

Data Types: rteEvent

done — Notification of freed buffer in memory

scalar

This message port sends a notification to the connected Stream Write block. This notification indicates that a read transaction completed, and that a buffer in memory is available for writing.

Data Types: Boolean

Parameters

Memory simulation — Set simulation timing accuracy

Burst accurate (default) | Protocol accurate | Behavioral

Memory simulation provides three levels of timing resolution. Select one of these options:

• Burst accurate — Simulates memory contention and high-resolution timing.

• Protocol accurate — Simulates AXI4 protocol hand-shaking sequencing and low-resolution timing.

• Behavioral — Simulates data transactions only and no timing.

  When this parameter is set to Behavioral mode, the value of the TLAST signal is ignored.

Main

Memory selection — Choose between memory regions

PS memory (default) | PL memory

Choose between processing subsystem (PS) or programming logic (PL) memory.

• If the selected board supports only PL memory, then the default value is PL memory.

• If the selected board supports only PS memory or only PL memory, then this parameter is read-only.

• If the selected board is not a supported SoC board, then this parameter is not visible.

Region size (bytes) — Size of the allocated memory region, in bytes

calculated

This parameter is read-only.

The size of the region in bytes. This value is calculated as the number of buffers multiplied by the size of the buffer. The size of the buffer is defined as the size of the data multiplied by the software frame size. Define this value by setting the Data type and Dimensions parameters on the Signal Attributes tab.

Example: A software frame size of 1024 uint32 defines a buffer of 4096 Bytes. If the number of buffers is set to 2, the region size is 8192 bytes.

Number of buffers — Number of buffers in region

8 (default) | integer

Define the number of buffers in the memory.

The memory access has a ring-buffer pattern. The writer can continually write as long as buffers are available. When a buffer is completed, it becomes available for the reader. The writer and reader traverse the buffers in a circular pattern. As long as the writer and reader maintain similar rates, the buffering prevents blockage.

A disparate rate between a reader and a writer slows down the faster device. For example, a slow reader causes the writer to run out of buffers and blocks the writer, effectively slowing down the writer to the reader rate. Likewise, a slow writer causes the reader to run out of buffers and blocks the reader, effectively slowing down the reader to the writer rate.

The Number of buffers parameter must be an integer from 3 to 64.

Burst length — Burst length for memory transactions

256 (default) | scalar

The length of bursts for this connection on the memory bus in units of scalar data. The scalar unit is the packed data type. Specify the burst size for both writer and reader access to the channel.

The channel data is always transferred to the memory model using burst transactions. For the AXI4 configuration, the algorithm logic is responsible for defining the burst through the protocol signals.

The Burst length parameter determines the burst size to the memory, and the rdData signal defines the size of each transfer on the interface.

FIFO depth (number of bursts) — Depth of FIFO for data

8 (default) | scalar

Specify the depth of the data FIFO, in units of bursts. When the writer has no buffers to write to, the FIFO can absorb data until a buffer becomes available. This value is the maximum number of bursts that the FIFO can buffer before it drops data.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

FIFO almost full depth (number of bursts) — Depth of FIFO when backpressure is asserted

6 (default) | scalar

Specify a number that asserts a backpressure signal from the channel to the data source. To avoid dropping data, set a high watermark, allowing the data producer enough time to react to backpressure. This number must be smaller than the FIFO depth.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

Clock frequency (MHz) — Interconnect frequency of manager datapath

100 (default)

Specify the frequency of the manager datapath in MHz.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

Data width (bits) — Data width of manager datapath

64 (default) | scalar

Specify the data width of manager datapath to the interconnect controller in bits.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

Signal Attributes

Software data signal

Dimensions — Dimensions of software data

1024 (default) | scalar

Specify the dimension for the software data (writer) as a whole number.

The default value is 1024.

Data type — Data type of software data

uint32 (default) | uint16 | uint64 | fixdt(1,16,0)

Specify the data type of the software data (writer). For help, click the ... button and select Data Type Assistant. By default, this value is set to inherit the data type from the source signal.

Stream data signal

Dimensions — Dimensions of output data signal

1 (default) | scalar | array

rdData can be a multidimensional array. Specify the dimension for the array as a whole number.

Example: 1 — A scalar sample.

Example: [10 1] — A vector of ten scalars.

Example: [1080 1920 3] — A 1080p frame. The frame includes 1080 lines of 1920 pixels per line, and each pixel is represented by three values (for red, green, and blue).

Data type — Data type of reader data

Inherit: Same as input (default) | double | single | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64 | Boolean | fixed point

Specify the data type of the rdData port. For help, click the ... button and select Data Type Assistant. By default, this value is set to inherit the data type from the source signal.

Sample time — Time interval of sampling

1 (default) | positive scalar | vector

Specify a time interval in seconds to define how often the block updates.

When you do not want the output to have a time offset, specify the Sample time parameter as a scalar. To add a time offset to the output, specify the Sample time parameter as a 1-by-2 vector, where the first element is the sampling period and the second element is the offset. For more information about sample times in Simulink^®, see Specify Sample Time.

Enable sample packing (last signal dimension as channel) — Pack data on last dimension of signal

off (default) | on

Select this parameter to enable data packing across the last dimension of the signal. The AXI4-Stream to Software block packs the data along the last dimension of the signal. For example, if the channel data type is uint32, the dimensions are [1024 4]. If you select this sample packing parameter, then the memory channel generates 1024 read or write transactions of 128 bits. If you clear this sample packing parameter, the memory channel generates 4096 transactions of 32 bits each.

This figure shows how data is aligned for a signal with data type fixdt10[4x3]. When the data is packed, three 10-bit words are concatenated and extended by 2 bits to a 32-bit word. When the data is not packed, each 10-bit word is extended to a 16-bit word.

This figure shows how data is aligned for a signal with data type uint8[8x3]. When the data is packed, three 8-bit words are concatenated and extended by 8 bits to a 32-bit word. When the data is not packed, each 8-bit word is represented as an 8-bit sample.

The combined width of the flattened signal must not exceed 512 bits.

Performance

Local Master

View performance plots — Display performance metrics

button

Clicking the button opens performance plots for the local manager in a new window. For more information about performance graphs, see Memory Channel Latency Plots.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

Memory Controller

View performance plots — Display performance metrics

button

Clicking the button opens the Performance Plots for Memory Controller window. You can then select to plot bandwidth, bursts, or latencies. For more information about performance graphs, see Memory Controller Latency Plots.

Dependencies

To enable this parameter, select Burst accurate under Memory simulation.

Extended Capabilities

HDL Code Generation
Generate VHDL, Verilog and SystemVerilog code for FPGA and ASIC designs using HDL Coder™.

Fixed-Point Conversion
Design and simulate fixed-point systems using Fixed-Point Designer™.

Version History

Introduced in R2022b

R2023b: Support for multiple AXI streams

Use multiple AXI4-Stream to Software, Software to AXI4-Stream, or Memory Channel blocks to model, simulate, and deploy designs with multiple streams to and from software. In previous releases, SoC Blockset™ supported multiple streams for modeling only. This release adds the ability to generate binaries and software and deploy it on SoC devices.