0% found this document useful (0 votes)
326 views82 pages

pg085 Axi4stream Infrastructure

The AXI4-Stream Infrastructure IP Suite v3.0 is a collection of modular IP cores designed for efficient connection of AXI4-Stream master/slave systems, offering features like buffering, transforms, and routing. It supports various device families and includes design files, example designs, and simulation models. The suite aims to facilitate the creation of complex AXI4-Stream systems while promoting inclusivity in language used across its products.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
326 views82 pages

pg085 Axi4stream Infrastructure

The AXI4-Stream Infrastructure IP Suite v3.0 is a collection of modular IP cores designed for efficient connection of AXI4-Stream master/slave systems, offering features like buffering, transforms, and routing. It supports various device families and includes design files, example designs, and simulation models. The suite aims to facilitate the creation of complex AXI4-Stream systems while promoting inclusivity in language used across its products.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 82

AXI4-Stream

Infrastructure IP Suite
v3.0
LogiCORE IP Product Guide

Vivado Design Suite


PG085 November 17, 2021

Xilinx is creating an environment where employees, customers,


and partners feel welcome and included. To that end, we’re
removing non-inclusive language from our products and related
collateral. We’ve launched an internal initiative to remove
language that could exclude people or reinforce historical biases,
including terms embedded in our software and IPs. You may still
find examples of non-inclusive language in our older products as
we work to make these changes and align with evolving industry
standards. Follow this link for more information.
Table of Contents
IP Facts

Chapter 1: Overview
Overview of Features. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
System Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Licensing and Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Chapter 2: Product Specification


AXI4-Stream Infrastructure IP Suite Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Resource Utilization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Port Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Register Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Chapter 3: Designing with the Core


General Design Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Clocking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Resets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Chapter 4: Design Flow Steps


Customizing and Generating the Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Constraining the Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Synthesis and Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Chapter 5: Example Design


Functionality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Chapter 6: Test Bench

Appendix A: Upgrading
Device Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


2
PG085 November 17, 2021 www.xilinx.com
Migrating to the Vivado Design Suite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Upgrading in the Vivado Design Suite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Appendix B: Debugging
Finding Help on Xilinx.com . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Debug Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Hardware Debug . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Interface Debug . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Appendix C: Additional Resources and Legal Notices


Xilinx Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Documentation Navigator and Design Hubs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Revision History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Please Read: Important Legal Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


3
PG085 November 17, 2021 www.xilinx.com
IP Facts

Introduction LogiCORE IP Facts Table

The AXI4-Stream Infrastructure IP Suite is a Core Specifics


collection of modular IP cores that can be used Supported UltraScale™ Architecture, 7 Series
Device Family(1)
to rapidly connect AXI4-Stream master/slave IP
systems in an efficient manner. All modules Supported User
AXI4-Stream, AXI4-Lite
Interfaces
have AXI4-Stream master and slave interfaces
Resources See Table 2-1.
that allow them to be daisy-chained
AXI4-Stream connections. The suite provides a Provided with Core
common set of functions including buffering, Design Files Verilog RTL
transforms, and routing. Together, these Example Design Verilog
modules provide the base-level functions to
Test Bench Verilog
create complex AXI4-Stream systems, allowing
Constraints File Xilinx Design Constraints (XDC)
system designers to create complex
AXI4-Stream systems in a timely manner. Simulation
Behavioral Verilog
Model
Supported
Features S/W Driver
N/A

Buffering Modules Tested Design Flows(2)


Design Entry Vivado® Design Suite
• AXI4-Stream Clock Converter
For supported simulators, see the
Provides clock crossing logic to bridge two Simulation
Xilinx Design Tools: Release Notes Guide.
clock domains. Synthesis Vivado Synthesis

• AXI4-Stream Data FIFO Support


Provides depth of 16 or deeper buffering www.xilinx.com/support
with support for multiple clocks, ECC, Notes:
different resource utilization types and 1. For a complete listing of supported devices, see the Vivado IP
optional FIFO Flags. Catalog.
2. For the supported versions of the tools, see the Xilinx Design
• AXI4-Stream Register Slice Tools: Release Notes Guide.

Creates timing isolation and pipelining


Features (continued)
master and slave using a two-deep register
buffer. • AXI4-Stream Data Width Converter

Transform Modules ° Increases the width of the TDATA signal


by combining a series of AXI4-Stream
• AXI4-Stream Combiner
transfers into one larger transfer.
° Aggregates multiple narrow
AXI4-Stream transfers in parallel into ° Decreases the width of a TDATA signal
by splitting an AXI4-Stream transfer into
one master by splicing the TDATA bits
a series of smaller transfers.
together in to create an AXI4-Stream
transfer with a wider output. • AXI4-Stream Subset Converter
Routing Modules
• AXI4-Stream Broadcaster
° Duplicates an AXI4-Stream transfer to
multiple slaves

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback 4


PG085 November 17, 2021 www.xilinx.com Product Specification
IP Facts

• AXI4-Stream Switch
° Allows multiple masters and slave to be connected by using the TDEST signal to route
transfers to different slaves.

° Optional control register routing mode that uses an AXI4-Lite interface to specify routing.
• AXI4-Stream Interconnect (Requires Vivado).

° Allows masters and slaves with differing AXI4-Stream characteristics to exchange


AXI4-Stream transfers.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


5
PG085 November 17, 2021 www.xilinx.com Product Specification
Chapter 1

Overview
The Arm® AMBA® 4 Specification builds on the AMBA 3 specifications by adding new
interface protocols to provide greater interface flexibility in designs with open standards.
Among the new interface protocols is the AXI4-Stream interface which is designed to
support low resource, high bandwidth unidirectional data transfers. It is well-suited for
FPGA implementation because the transfer protocol allows for high frequency versus clock
latency trade-offs to help meet design goals.

The AXI4-Stream protocol is derived from the single AXI3 write channel. It has no
corresponding address or response channel and is capable of non-deterministic burst
transactions (undefined length). The protocol interface signal set adds optional signals to
support data routing, end-of-transaction indicators, and null-beat modifiers to facilitate
management and movement of data across a system. These characteristics are suitable for
transferring large amounts of data efficiently while keeping gate count low.

The protocol interface consists of a master interface and a slave interface. The two
interfaces are symmetric and point-to-point, such that master interface output signals can
connect directly to the slave interface input signals. Utilizing this concept, it is possible to
design AXI4-Stream modules that have a slave interface input channel and a master
interface output channel. Because the master and slave interfaces are symmetric, any
number of these modules can be daisy-chained together by connecting the master
interface output channel of one module to the slave interface input channel of another
module and so on. The function of the modules can be a multitude of different options such
as buffering, data transforming or routing.

The AXI4-Stream Infrastructure IP Suite is a powerful collection of modules that provides a


rich set of functions for connecting together AXI4-Stream masters and slaves. The IP core is
capable of performing data switching/routing, data width conversion, pipelining, clock
conversion and data buffering. Parameters and IP configuration Graphical User Interfaces
(GUIs) are used to configure the core to suit each of the system designer's requirements.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


6
PG085 November 17, 2021 www.xilinx.com
Chapter 1: Overview

Overview of Features
The AXI4-Stream Infrastructure IP Suite consists of eight modular IP cores supporting the
full AXI4-Stream specification. Common features include:

AXI4-Stream compliant

• Supports all AXI4-Stream defined signals: TVALID, TREADY, TDATA, TSTRB, TKEEP,
TLAST, TID, TDEST, and TUSER.
• TDATA, TSTRB, TKEEP, TLAST, TID, TDEST, and TUSER are optional.
• Programmable TDATA, TID, TDEST, and TUSER widths (TSTRB and TKEEP width is
TDATA width/8).
• ACLK/ARESETn ports.
• Per port ACLKEN inputs (optional).

AXI4-Stream Broadcaster
• Replicates a master stream into multiple output slave streams.
• Provides TDATA/TUSER remap functionality.
• Supports 2-16 slaves.

AXI4-Stream Clock Converter


• Supports low latency and area synchronous 1:N and N:1 clock conversion.
• Supports asynchronous clock conversion.
• Supports configurable ACLKEN conversion.

AXI4-Stream Combiner
• Combines multiple "narrow" streams into one wide output stream.
• Supports 2-16 masters.
• Supports error detection for unmatched TLAST, TID, or TDEST signals slave interfaces.

AXI4-Stream Data FIFO


• Supports FIFO depths from 16-32,678 in powers of 2.
• Supports Distributed RAM, Block RAM, and UltraRAM (on select devices) memory
primitive types.
• Utilizes Xilinx™ Parameterized Macros for automatic constraint generation and FIFO
implementation.
• Supports independent read/write clocks and ACLKEN conversion.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


7
PG085 November 17, 2021 www.xilinx.com
Chapter 1: Overview

• Supports Packet Mode (Store and Forward based on TLAST).


• Supports error correction code (ECC) with optional ECC error injection inputs.
• Optional FIFO Flags: write data count, almost full, programmable full, read data count,
almost empty, and programmable empty.

AXI4-Stream Data Width Converter


• Supports 1:N TDATA width size increase in a single stage.
• Supports N:1 TDATA width size decrease in a single stage.
• Supports arbitrary M:N TDATA width conversion in multiple stages.

AXI4-Stream Register Slice


• Allows pipelining of AXI4-Streams.
• Provides timing isolation.
• Optional pipelining to cross super logic regions (SLRs) in stacked silicon interconnect
(SSI) devices.

AXI4-Stream Subset Converter


• Provides TDATA/TUSER remap functionality.
• Allows streams with different signal sets to be connected.
• Can generate a programmable TLAST.
• Can tie-off unused signals from masters.
• Can add signals based on default value rules.

AXI4-Stream Switch
• Supports 1-16 slaves.
• Supports 1-16 masters.
• Has slave side arbitrated crossbar switch.
• Supports multiple arbitration tuning points:

° Ability to arbitrate based on TLAST.

° Ability to arbitrate based on number of transfers.

° Ability to arbitrate based on a timeout (counts number of consecutive LOW TVALID


cycles).
• Supports Round-Robin, True Round-Robin, and Fixed Priority arbitration choices.
• Supports sparse connectivity.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


8
PG085 November 17, 2021 www.xilinx.com
Chapter 1: Overview

• Supports routing based on TDEST base/high pairs OR optional control register routing
with AXI4-Lite interface.

AXI4-Stream Interconnect
Note: AXI4-Stream Interconnect requires Vivado® IP integrator.
• Supports 1-16 slaves
• Combines AXI4-Stream Switch with buffering modules, AXI4-Stream Data Width
Converter and AXI4-Stream Subset Converter to allow masters and slaves with varying
AXI4-Stream characteristics to exchange AXI4-Stream transfers.

System Requirements
For a list of System Requirements, see the Xilinx Design Tools: Release Notes Guide.

Licensing and Ordering


This Xilinx ® LogiCORE™ IP module is provided at no additional cost with the Xilinx Vivado ®
Design Suite under the terms of the Xilinx End User License. Information about other Xilinx
LogiCORE IP modules is available at the Xilinx Intellectual Property page. For information
about pricing and availability of other Xilinx LogiCORE IP modules and tools, contact your
local Xilinx sales representative.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


9
PG085 November 17, 2021 www.xilinx.com
Chapter 2

Product Specification

AXI4-Stream Infrastructure IP Suite Modules


AXI4-Stream Broadcaster
The AXI4-Stream Broadcaster provides a solution for replicating a single inbound
AXI4-Stream interface into multiple outbound AXI4-Stream interfaces. Support for up to 16
outbound AXI4-Stream interfaces is provided. Each outbound interface also supports an
optional remapping feature that allows you to select which TDATA (or TUSER) bits from the
inbound interface are present on the TDATA (or TUSER) port of each outbound interface. A
block diagram of the broadcaster is shown in Figure 2-1.
X-Ref Target - Figure 2-1

TDATA/TUSER
Remapper

TDATA/TUSER
Remapper

TDATA/TUSER
Remapper

X13158
X-Ref Target - Figure 2-2

Figure 2-1: AXI4-Stream Broadcaster Block Diagram

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


10
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

AXI4-Stream Clock Converter


Clock converters are necessary in the AXI4-Stream protocol for converting masters
operating at different clock rates to slaves. Typically, the AXI4-Stream Infrastructure IP
should be clocked at the same rate as the fastest slave and devices not running at that same
rate need to be converted. Synchronous clock converters are ideal because they have the
lowest latency and smaller area. However, they are only viable if both clocks are
phase-aligned, integer clock ratios, and the F MAX requirements are able to be met.

Asynchronous clock converters are a generic solution able to handle both synchronous/
asynchronous clocks with arbitrary phase alignment. The trade-off is that there is a
significant increase in area and latency associated with asynchronous clock converters. If
global clock enables are configured, additional logic is generated to handle clock enables
independently for each clock domain. There is a Clock converter module available in every
datapath and it is instantiated if either the clocks are specified as asynchronous or have
different synchronous clock ratios. The Clock converter module performs the following
functions:

• A clock-rate reduction module performs integer (N:1) division of the clock rate from its
input (SI) side to its output (MI) side.
• A clock-rate acceleration module performs integer (1:N) multiplication of clock rate
from its input (SI) to output (MI) side.
• Asynchronous clock rate conversion between the input and output uses Xilinx
parameterized macros for automatic constraints generation.
• Clock enable crossing logic that handles different ACLKEN signals per clock domain.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


11
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Figure 2-2 shows the clock converter with support for independent ACLKEN signals on its
SI and MI.
X-Ref Target - Figure 2-2

Clock Converter Module

Endpoint Endpoint

Clock Crossing
Master Slave

Logic
MI SI MI SI

ACLKEN SI_ACLKEN MI_ACLKEN ACLKEN

Clk A Clk B

X12657

Figure 2-2: Clock Converter Module Block Diagram

AXI4-Stream Combiner
The AXI4-Stream Combiner provides a solution for aggregating multiple narrow inbound
AXI4-Stream interfaces into a single wide outbound AXI4-Stream interface. Support for up
to 16 inbound AXI4-Stream interfaces is provided. A block diagram of the AXI4-Stream
Combiner is shown in Figure 2-3. A common use case of this solution is to merge three
separate red, green, blue video streams into a single RGB stream.

The combiner concatenates the incoming streams signals TDATA/TSTRB/TKEEP/TUSER to


create a single output stream that is the combination of the input streams. The TLAST/TID/
TDEST signals are taken from a single slave interface and the primary slave interface is
configurable.

IMPORTANT: All slave interfaces must assert the TVALID signal before the TVALID signal on the output
is asserted.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


12
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

X-Ref Target - Figure 2-3

axis_combiner

AXI4-Stream[0] Payload[0]

Payload[1]
AXI4-Stream[1] AXI4-
Stream

AXI4-Stream[15]

Payload[15]

X13159

Figure 2-3: AXI4-Stream Combiner Block Diagram

AXI4-Stream Data FIFO


The FIFO module is capable of providing temporary storage (a buffer) of the AXI4-Stream
data. The FIFO Buffer module should be used in between two endpoints when:

• More buffering than a register slice is desired.


• Store and forward: to accumulate a certain number of bytes from the master before
forwarding them to the slave (packet mode).

Based on the Xilinx Parameterized Module xpm_fifo_axis, this IP allows you to easily use
the macro within Vivado® IP integrator with an easy to configure GUI. See the UltraScale
Architecture Libraries Guide (UG974) [Ref 11] for more details on the xpm_fifo_axis
macro. This supports native AXI4-Stream with the following features:

• Configurable FIFO depths.


• Total FIFO data width of up to 4096 bits.
• Independent or common clock domains.
• Symmetric aspect ratios.
• Asynchronous active-Low reset.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


13
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

• Selectable memory type. Distributed RAMs are suitable for shallow depths and only
consume LUTs/Registers. Block RAMs are suitable for medium depth FIFOs and
consume only block RAM resources. UltraRAMs are best for very deep FIFOs, but are
limited to certain architectures and cannot use independent master/slave clocks.
• ECC support for single-bit error correction, and double-bit error detection. Optional
error injection for debug.
• FIFO Flags including almost full/empty, and programmable full/empty.
• Data counts for both writes (synchronous to s_axis clock) and reads (synchronous to
m_axis clock) interfaces.

AXI4-Stream Data Width Converter


Data width converters (upsizer/downsizer) are required when interfacing different data
width cores with each other. One data width conversion module is available to handle all
supported combinations of data widths.

The conversion follows the AMBA® AXI4-Stream Protocol Specification with regards to
ordering and expansion of TUSER bits. The width converter does not process any special
TUSER encoding formats; it only maps TUSER bits across the width conversion function
using the algorithm specified in the AXI4-Stream protocol specification. Depending on the
usage/meaning of TUSER to the endpoint IP, additional external logic might be required to
manipulate TUSER bits that have been transformed by the width converter.

IMPORTANT: The number of TUSER bits per TDATA bytes must remain constant between input and
output.

Up-conversion requires that each incoming beat that is composed of the new larger beat
consists of identical TID and TDEST bits and no intermediate TLAST assertions. Partial data
may be flushed when either the TLAST bit is received or TID/TDEST changes before
enough data is accumulated to send out a complete beat. Unassigned bytes are flushed out
as null bytes.

RECOMMENDED: Monitor the TKEEP signal output if TID/TDEST/TLAST signal is present.

Any non-integer multiple byte ratio conversion (N:M) is accomplished by calculating the
lowest common multiple (LCM) of N and M and then up-converting from N:LCM then
down-converting from LCM:M.

Up-conversion features:

• Range: Input 1-256 Bytes, Output 2-512 Bytes


• Supports full range of 1:N byte ratio conversions
• Minimum latency of 2 + N clock cycles in 1:N byte ratio up-conversion.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


14
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Down-conversion features:

• Range: Input 2-512 bytes, output 256-1 bytes


• Supports full range of N:1 byte ratio conversions
• Minimum Latency: 2 clock cycles
X-Ref Target - Figure 2-4

Data Converter Module

Endpoint 64 bits 32 bits Endpoint


Master MI SI MI SI Slave

X12656

Figure 2-4: Data Width Converter (Down Conversion) Module Block Diagram

AXI4-Stream Register Slice


The register slice is a multipurpose pipeline register that is able to isolate timing paths
between master and slave. The register slice is designed to trade-off timing improvement
with area and latency necessary to form a protocol compliant pipeline stage. Implemented
as a two-deep FIFO buffer by default, the register slice supports throttling by the master
(channel source) and/or slave (channel destination) as well as back-to-back transfers
without incurring unnecessary idle cycles. The module can be independently instantiated at
all port boundaries. A configuration parameter allows for the trade off of performance vs.
area efficiency, including a mode that adds extra pipeline stages to optimally cross super
logic regions (SLR) boundaries in stacked silicon interconnect (SSI) devices.

AXI4-Stream Subset Converter


The AXI4-Stream Subset Converter provides a solution for connecting slightly incompatible
AXI4-Stream signal sets together. The IP has configurable AXI4-Stream signals for each
interface that allows one to convert one signal set to another in consistent manner.

All signals can be configured to be removed or added and additionally the TDATA, TUSER,
TSTRB, TKEEP, TID, TDEST, and TLAST signals can be remapped.

CAUTION! Due to the inherent data loss, care should be taken to fully understand the payload content
when converting to a signal set with fewer signals.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


15
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

When signals are added, they are assigned default values as specified by the AMBA
specification. The TLAST signal can be added with a configurable assertion counter that
allows one to packetize their data. The REMAP functionality can be used to re-order TDATA
bytes when working with core that have slightly different notations for data storage and
propagation.

AXI4-Stream Switch
The AXI4-Stream Switch provides configurable routing between masters and slaves. It
supports up to 16 masters and 16 slaves, two routing options, and multiple arbitration
options.

The two routing options available are TDEST routing and control register routing. The
TDEST based routing uses RTL parameters configured before synthesis to control the
routing. Each master interface is assigned a base/high TDEST pair that is used to generate
a decode table. Each slave interface decodes the incoming transfer based on the valid
TDEST value and routes a request to an arbiter of one of the master interfaces. When the
arbiter responds with a grant, then the slave interface proceeds with the transfer.
Arbitration can be performed at the transfer level or at the transaction level. (A transaction
is a series of two or more transfers.) Transaction level arbitration can be set to either
arbitrate at either fixed lengths or at TLAST boundaries. An optional timeout option is
available that terminates the transaction before the fixed length or TLAST is received if the
connection is idle for too long. This can help avoid deadlock in certain system topologies.
TDEST based routing requires that the signal has at least log2(Number of Slave Interfaces)
number of bits.

Control register routing introduces an AXI4-Lite interface to configure the routing table.
There is one register for each of the master interfaces to control each of the selectors. Once
the registers have been programmed, a commit register transfers the programmed values
from the register interface into the switch. During this period, the AXI4-Stream interfaces
are held in reset. This routing mode requires that there is precisely only one path between
master and slave. When attempting to map the same slave interface to multiple master
interfaces, only the lowest master interface is able to access the slave interface. Unused
master interfaces may be disabled and any unmapped slave interfaces are disabled.

Sparse connectivity between slave and master interfaces can be configured. This allows
resources to be conserved when they are not needed or to prevent invalid routes. For each
slave interface and master interface, a grid is created to allow you to deselect connectivity
during IP configuration. Invalid routes stall if using control register based routing and drop
transfers in the TDEST based routing. When transfers are dropped, the appropriate
decode_err signal is asserted.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


16
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Standards
The IP cores have bus interfaces that comply with the Arm® AMBA AXI4-Stream Protocol
Specification Version 1.0.

Performance
The performance of an AXI4-Stream Infrastructure IP core is limited only by the FPGA logic
speed. Each core utilizes only block RAMs, LUTs, and registers and contains no I/O elements.
The values presented in this section should be used as an estimation guideline, actual
performance can vary.

Maximum Frequencies
Each core is designed to meet the maximum target frequency of 250 MHz on a Kintex ®-7
FPGA (xc7k325tffg900-1.) It can be expected that an -2 speed grade part can achieve 5%
higher maximum target frequency and that a -3 speed grade part can achieve 10% higher
maximum target frequency. For AXI4-Stream Switch configurations with more than
approximately four masters or slaves, the target maximum frequency can be reduced by
20-25%.

Latency
The latency in the IP cores can vary on an interface-to-interface basis, depending on how
the IP cores are configured. The latency is calculated in clocked cycles and is measured as
the time that it takes from the assertion of the slave interface TVALID signal to the first
assertion of the master interface TVALID signal. The latency for each of the individual
modules is listed in Table 2-1. To obtain the minimum latency for the system, add up the
values shown in the following tables for the modules in your system. The latency
specifications assume that the master interface TREADY signal input is always asserted. The
back-to-back delay is the number of clock cycles that back-to-back transfers can be
accepted by the module. This can be observed by counting how many cycles slave interface
TREADY is Low after a transfer is accepted on the interface.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


17
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-1: Latency by Module Type


Back-to-
Latency Back
Module Type Description
(Clocks) Delay
(Clocks)
AXI4-Stream 0 0 The datapath of the broadcaster is combinatorial. It exhibits no
Broadcaster latency if all M_AXIS interfaces have TREADY asserted.
AXI4-Stream Clock 1 0 The synchronous clock converter latency is reported as units of the
Converter slave interface clock.
(synchronous,
speed-up)
AXI4-Stream Clock 1 [clock The synchronous clock converter latency is reported as units of the
Converter ratio]-1 slave interface clock. The back-to-back delay varies based on the
(synchronous, clock ratio. Example: If using a synchronous 150 MHz-to-50 MHz
speed-down) 3:1 clock converter (clock ratio of 3), the back-to-back delay is 2
clock cycles.
AXI4-Stream Clock Not 0 The latency associated with an asynchronous clock converter can
Converter Defined vary greatly depending on the clocks. It can be expected to see
(asynchronous) latencies of 5 clock cycles or more.
AXI4-Stream 0 0 The datapath of the Combiner module is combinatorial and thus
Combiner has no latency if all ready/valid inputs are asserted.
AXI4-Stream Data TBD 0 The FIFO when configured in normal mode outputs data as soon
FIFO as it is possible.
AXI4-Stream Data Until 0 When configured in packet mode, the FIFO outputs data only when
FIFO (packet mode) TLAST is a TLAST is received or the FIFO has filled.
receive
d or
FIFO is
full.
AXI4-Stream Data [data 0 The latency varies based on the data width ratio. Example: If a 32
Width Converter width to 128-bit data converter is used (1:4 ratio), the latency of the
(upsizer) ratio] module is 4 clock cycles.
AXI4-Stream Data 1 [data The back-to-back delay varies based on the data width ratio.
Width Converter width Example: If a 32 to 16-bit data converter is used (2:1 ratio), then
(downsizer) ratio]-1 the module can only accept transfers every other cycle.
AXI4-Stream Register 1 0 Adding a register slice adds one cycle of latency. There is no
Slice (default or back-to-back delay.
Fully-registered
mode)
AXI4-Stream Register 1 1 Adding a register slice adds one cycle of latency. Light-weight
Slice Lightweight mode inserts one bubble cycle after each transfer.
mode
AXI4-Stream Register 3 0 SLR Crossing mode incurs 3 latency cycles and adds no
Slice SLR Crossing back-to-back delay.
mode
AXI4-Stream Register 3 0 SLR TDM Crossing mode incurs 3 aclk latency cycles and adds no
Slice SLR TDM back-to-back delay.
Crossing mode

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


18
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-1: Latency by Module Type


Back-to-
Latency Back
Module Type (Clocks) Delay Description
(Clocks)
AXI4-Stream Register 0 0 Bypass mode directly connects the SI to the MI.
Slice Bypass mode
AXI4-Stream Subset 0-1 0 A register slice is inserted when there is a m_axis_tready signal, but
Converter not a s_axis_tready signal to avoid violation of the AXI4-Stream
protocol. In this configuration, the latency is 1 cycle, otherwise it is
0.
AXI4-Stream Switch 2 0-1 The output latency of the switch is 2 clock cycles. There is 1 cycle
of latency for the TDEST decode and 1 cycle of latency for the
arbiter grant (if idle.) The back-to-back delay for an already
granted arbitration is 0. Back-to-back arbitration results in 1 cycle
delays between transactions.

Throughput
The throughput of a datapath through each AXI4-Stream Infrastructure IP is calculated as
TDATA width x clock frequency of each of the paths determined by the SI interface, and MI
interface. The minimum throughput of an individual path in a system for which the transfer
will traverse determines the overall throughput of the datapath.

Resource Utilization
The resource utilization of each AXI4-Stream Infrastructure IP is primarily a function of the
payload width of the stream. The payload width of the stream is calculated as the width of
the TDATA, TSTRB, TKEEP, TLAST, TID, TDEST, and TUSER signals. For example, consider
the design that has the following signal widths listed in Table 2-2.

Table 2-2: Signal Widths Used for Resource Utilization Estimation


AXI4-Stream Signal Width
TDATA 64
TSTRB 8
TKEEP 8
TLAST 1
TID 5
TDEST 6
TUSER 8
Total (W p) 100

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


19
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

The payload width W P is calculated as 64 + 8 + 8 + 1 + 5 + 6 + 8 = 100. The register slice


works as a double buffer and is able to hold two AXI4-Stream transfers at one time.
Therefore, a rough estimate of utilization can be achieved by multiplying the payload width
by two. This signal configuration from Table 2-2 is used in Table 2-3 as the basis for the
resource utilizations of the individual modules on a Kintex-7 FPGA (xc7k325tffg900-1) using
the Vivado synthesis tool. UltraScale™ results are expected to be similar to 7 series results.

Table 2-3: Resource Utilization by Module Type


Block
Module Feature LUTs FFs
RAMs
2 Master Interfaces 5 2 0
AXI4-Stream Broadcaster 4 Master Interfaces 9 4 0
8 Master Interfaces 21 8 0
Asynchronous 104 287 0
AXI4-Stream Clock
Synchronous 2:1 109 211 0
Converter
Synchronous 1:2 107 209 0
2 Slave Interfaces 1 1 0
AXI4-Stream Combiner 4 Slave Interfaces 2 1 0
8 Slave Interfaces 4 1 0
TDATA: 32 to 64 bits (W P= 56 to W P =100) 35 164 0
TDATA: 32 to 128 bits ( W P =56 to W P =188) 44 254 0
TDATA: 64 to 32 bits (W P =100 to W P =56) 51 164 0
AXI4-Stream Data Width TDATA: 64 to 128 bits ( W P =100 to W P 35 296 0
Converter =188)
TDATA: 128 to 32 bits ( W P =188 to W P =56) 129 257 0
TDATA: 128 to 64 bits ( W P =188 to W P 75 296 0
=100)
AXI4-Stream Register Slice Default 110 206 0
AXI4-Stream Subset No SI TREADY -> MI TREADY 105 198 0
Converter
1 slave interface x 2 master interfaces 123 216 0
1 slave interface x 4 master interfaces 130 220 0
2 slave interface x 1 master interfaces 68 13 0
AXI4-Stream Switch
2 slave interface x 2 master interfaces 376 454 0
4 slave interface x 1 master interfaces 133 27 0
4 slave interface x 4 master interfaces 1028 978 0

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


20
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Port Descriptions
Global Signals
These signals are always present when there is a common clock between all interfaces of the
IP core.

Table 2-4: Global Signals


Signal Direction Description
aclk Input Global Clock Signal. Drives the clocks on the AXI4-Stream Switch and
is the primary clock to the system.
aresetn Input Global Reset Signal. This active-Low signal drives the reset pins on
the AXI4-Stream Switch and is the primary reset of the system.
aclken Input Global ACLK Enable signals. Drives the ACLKEN pins on the
AXI4-Stream Switch and is the primary ACLKEN of the system.
aclk2x Input This auxiliary clock input is only enabled on the AXI4-Stream Register
Slice when configured in SLR TDM Crossing mode, and must be
exactly twice the frequency of aclk and generated from the same
clock source with zero phase shift.

Slave Interface Signals


The following table lists the signals associated with each slave interface. If the number of
interfaces is configurable, then the signals in Table 2-5 are replicated for each port. The nn
denoted for the signals starts at 00 and increments by one up to 15 for each slave interface
instantiated. For IPs that contain only one slave interface the nn value is dropped. For
example, the Snn_AXIS_TVALID would be S_AXIS_TVALID. IP cores that do not support
multiple clocks do not have the Snn_AXIS_ACLK, Snn_AXIS_ARESETN, or the
Snn_AXIS_ACLKEN signals.

Table 2-5: Signals Associated with the Slave Interface


Signal Direction Description
Clock signal. All inputs/outputs of this bus
snn_axis_aclk Input interface are rising edge aligned with this
clock.
Active-Low synchronous reset signal (all
cores, except AXI4-Stream Data FIFO.)
snn_axis_aresetn Input
Active-Low asynchronous reset signal
(AXI4-Stream Data FIFO only.)
snn_axis_aclken Input Clock enable signal

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


21
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-5: Signals Associated with the Slave Interface (Cont’d)


Signal Direction Description
TVALID indicates that the master is driving
(1) a valid transfer.
snn_axis_tvalid Input
A transfer takes place when both TVALID
and TREADY are asserted.
TREADY indicates that the slave can accept
snn_axis_tready(1) Output
a transfer in the current cycle.
TDATA is the primary payload that is used
snn_axis_tdata to provide the data that is passing across
Input
[(C_MNN_AXIS_TDATA_WIDTH-1):0](1) the interface. The width of the data payload
is an integer number of bytes.
TSTRB is the byte qualifier that indicates
snn_axis_tstrb whether the content of the associated byte
Input
[((C_MNN_AXIS_TDATA_WIDTH/8)-1):0] (1) of TDATA is processed as a data byte or a
position byte.
TKEEP is the byte qualifier that indicates
whether the content of the associated byte
of TDATA is processed as part of the data
snn_axis_tkeep
Input stream.
[((C_MNN_AXIS_TDATA_WIDTH/8)-1):0] (1)
Associated bytes that have the TKEEP byte
qualifier deasserted are null bytes and can
be removed from the data stream.
snn_axis_tlast (1) Input TLAST indicates the boundary of a packet.
snn_axis_tid TID is the data stream identifier that
Input
[C_NATIVE_TID_WIDTH-1:0] (1) indicates different streams of data.
snn_axis_tdest TDEST provides routing information for the
Input
[(C_NATIVE_TDATA_WIDTH-1):0] (1) data stream.
TUSER is user-defined sideband
snn_axis_tuser
Input information that can be transmitted
[(C_SNN_AXIS_TUSER_WIDTH-1):0] (1) alongside the data stream.
AXI4-Stream Switch only signal.
Active-High signal to skip this bus on the
next arbitration cycle. While the signal is
s_req_suppress[C_NUM_SI_SLOTS-1:0] Input asserted, this bus does not receive the next
arbitration. If this bus already has
arbitration granted, it remains granted until
the arbitration cycle is completely normally.
AXI4-Stream Switch only signal. One-hot
output indicates that a incoming transfer
has a TDEST value that not map to a valid
s_decode_err[C_NUM_SI_SLOTS-1:0] Output
Master Interface. Invalid TDEST transfers
are dropped. Only valid if the TDEST signal
is present and used for routing.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


22
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-5: Signals Associated with the Slave Interface (Cont’d)


Signal Direction Description
AXI4-Stream Subset Converter only signal.
This signal is only present if the Slave
Interface TREADY signal is not enabled and
transfer_dropped Output the Master interface TREADY signal is
enabled. This signal indicates if there is an
AXI-S transfer that has been dropped due
to a de-asserted MI TREADY.
AXI4-Stream Subset Converter only signal.
This signal is only present if the Slave
Interface TKEEP is enabled and the Master
sparse_tkeep_removed Output Interface TKEEP is not enabled. This signal
signals if there is a Slave Interface TKEEP
that has been removed and null data bytes
were present.
AXI4-Stream Combiner only. This output is
s_cmd_err[(C_NUM_SI_SLOTS*3)-1:0] Output
not defined and may change in the future.
AXI4-Stream Data FIFO Only. Indicates the
write count inside the DATA
axis_wr_data_count[31:0] Output FIFO. This signal can be used when using
asynchronous clocking and can be sampled
on the posedge of the s_axis_aclk.
Almost Full: When asserted, this signal
indicates that only one more write can be
almost_full Output
performed before the FIFO is full. Not
available when Packet Mode is used.
Programmable Full: This signal is asserted
when the number of words in the FIFO is
greater than or equal to the programmable
prog_full Output
full threshold value. It is de-asserted when
the number of words in the FIFO is less than
the programmable full threshold value.
Single Bit Error Injection- Injects a single bit
error if the ECC feature is used. The signal
injectsbiterr Input should be asserted with TVALID and can
only transition with a TVALID/TREADY
handshake.
Double Bit Error Injection- Injects a double
bit error if the ECC feature is used. Signal
injectdbiterr Input should be asserted with TVALID and can
only transition with a TVALID/TREADY
handshake.

Notes:
1. This signal description is taken from the Arm AMBA Protocol Specification.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


23
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Master Interface Signals


Table 2-6 lists the signals associated with each master interface. If the number of interfaces
is configurable, the signals are then replicated for each port. The nn denoted for the signals
starts at 00 and increments by one up to 15 for each master interface instantiated. For IPs
that contain only one master interface the nn value is dropped. For example, the
Mnn_AXIS_TVALID would be M_AXIS_TVALID. IPs that do not support multiple clocks do
not have the Mnn_AXIS_ACLK, Mnn_AXIS_ARESETN, or the Mnn_AXIS_ACLKEN signals.

Table 2-6: Signals Associated with the Master Interface


Signal Direction Description
Clock signal. All inputs/outputs of this
mnn_axis_aclk Input bus interface are rising edge aligned
with this clock.
mnn_axis_aresetn Input Active-Low synchronous reset signal
mnn_axis_aclken Input Clock enable signal
TVALID indicates that the master is
(1) driving a valid transfer.
mnn_axis_tvalid Output
A transfer takes place when both TVALID
and TREADY are asserted.
TREADY indicates that the slave can
mnn_axis_tready(1) Input
accept a transfer in the current cycle.
TDATA is the primary payload that is
used to provide the data that is passing
mnn_axis_tdata
Output across the interface. The width of the
[(c_mnn_axis_tdata_width-1):0] (1)
data payload is an integer number of
bytes.
TSTRB is the byte qualifier that indicates
mnn_axis_tstrb whether the content of the associated
Output
[((c_mnn_axis_tdata_width/8)-1):0](1) byte of TDATA is processed as a data byte
or a position byte.
TKEEP is the byte qualifier that indicates
whether the content of the associated
byte of TDATA is processed as part of the
mnn_axis_tkeep data stream.
Output
[((c_mnn_axis_tdata_width/8)-1):0](1) Associated bytes that have the TKEEP
byte qualifier deasserted are null bytes
and can be removed from the data
stream.
TLAST indicates the boundary of a
mnn_axis_tlast (1) Output
packet.
mnn_axis_tid TID is the data stream identifier that
Output
[c_native_tid_width-1:0](1) indicates different streams of data.
mnn_axis_tdest TDEST provides routing information for
Output
[(c_native_tdata_width-1):0](1) the data stream.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


24
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-6: Signals Associated with the Master Interface (Cont’d)


Signal Direction Description
TUSER is user-defined sideband
mnn_axis_tuser
Output information that can be transmitted
[(c_snn_axis_tuser_width-1):0](1)
alongside the data stream.
AXI4-Stream Data FIFO Only. Indicates
the read count inside the DATA
FIFO. This signal can be used when using
axis_rd_data_count[31:0] Output
asynchronous clocking and can be
sampled on the posedge of the
m_axis_aclk.
Almost Empty : When asserted, this
signal indicates that only one more read
almost_empty Output can be performed before the FIFO goes
to empty. Not available when Packet
Mode is used.
Programmable Empty- This signal is
asserted when the number of words in
the FIFO is less than or equal to the
prog_empty Output programmable empty threshold value. It
is de-asserted when the number of
words in the FIFO exceeds the
programmable empty threshold value.
Single Bit Error- Indicates that the ECC
decoder detected and fixed a single-bit
sbiterrOutput Output
error. This signal is asserted for the
transfer in which the error was detected.
Double Bit Error- Indicates that the ECC
decoder detected a double-bit error and
dbiterrOutput Output data in the FIFO core is corrupted. This
signal is asserted for the transfer in
which the error was detected.

Notes:
1. This signal description is taken from the Arm AMBA Protocol Specification.

AXI4-Lite Interface Signals


Table 2-7 lists the signals associated with the optional AXI4-Lite control register interface.
This interface is optional and only present on a subset of IPs.

Table 2-7: Optional AXI4-Lite Interface Signals

Signal(1) Direction Description

s_axi_ctrl_aclk Input Clock signal. All inputs/outputs of this bus interface are rising edge
aligned with this clock.
s_axi_ctrl_aresetn Input Active-Low synchronous reset signal

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


25
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-7: Optional AXI4-Lite Interface Signals (Cont’d)

Signal(1) Direction Description

s_axi_ctrl_awalid Input Write address valid. This signal indicates that the channel is signaling
valid write address.
s_axi_ctrl_awready Output Write address ready. This signal indicates that the slave is ready to
accept an address.
s_axi_ctrl_awaddr Input Write address. The write address gives the address of the transaction.
s_axi_ctrl_wvalid Input Write valid. This signal indicates that valid write data are available.
s_axi_ctrl_wready Output Write ready. This signal indicates that the slave can accept the write
data.
s_axi_ctrl_wdata Input Write data.
s_axi_ctrl_bvalid Output Write response valid. This signal indicates that the channel is signaling
a valid write response.
s_axi_ctrl_bready Input Write response ready. This signal indicates that the master can accept
a write response.
s_axi_ctrl_bresp Output Write response. This signal indicate the status of the write transaction.
s_axi_ctrl_arvalid Input Read address valid. This signal indicates that the channel is signaling
valid read address.
s_axi_ctrl_arready Output Read address ready. This signal indicates that the slave is ready to
accept an address.
s_axi_ctrl_araddr Input Read address. The read address gives the address of the transaction.
s_axi_ctrl_rvalid Output Read valid. This signal indicates that the channel is signaling the
required read data.
s_axi_ctrl_rready Input Read ready. This signal indicates that the master can accept the read
data and response information.
s_axi_ctrl_rdata Output Read data.
s_axi_ctrl_rresp Output Read response. This signal indicate the status of the read transfer.

Notes:
1. This signal description is taken from the Arm AMBA Protocol Specification.

Register Space
AXI4-Stream Switch
The AXI4-Stream switch has a Control Register interface option that can be enabled when
the Use control register routing option is set to Yes. Table 2-8 describes the register map.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


26
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Table 2-8: Register Map


Access Double
Address Offset Name Type Buffered Default Value Description

0x0000 Control R/W Yes 0x0 General control


0x0004-0x003C Reserved 0x0
0x0040-0x007F MI_MUX[0-15] R/W Yes 0x80000000 MI selector value

Control Register
This register is responsible for committing the MI selector values from the control register
block to the AXI4-Stream Switch block. This register has a single bit that commits the
change. The commit causes the AXI4-Stream switch to go into a soft reset for approximately
16 cycles. This register is self-clearing when the commit/reset is complete.

Table 2-9: Control Register


Name Bits Description
Reserved 0 Reserved

REG_UPDATE 1 Register Update. MUX registers are double buffered. Writing '1'
updates the registers and issues a soft reset to the core (for
approximately 16 cycles.)
Reserved 31:2 Reserved

MI_MUX[0-15] Register
There is one MI_MUX register for each number of master interfaces ports in the design. Each
MIx_MUX value controls slave interface selection. For example, MI4_MUX value of 0x1
would route slave interface 1 to master interface 4. The MIx_DISABLE value can be set to
disable the master interface. Each slave interface can only be selected once. If more than 1
MIx_MUX value is set to the same slave interface, then the lower master interface wins
control and the higher master interface(s) is disabled. MI0_MUX register is at address offset
0x40, MI1_MUX_register is at address offset 0x44, …, and MI15_MUX register is at address
offset 0x7C.

Table 2-10: MI Mux Registers


Name Bits Description
MIx_MUX 3:0 MIx Mux Value
MIx_DISABLE 31 Set to 1 to explicitly disable

Notes:
1. x is 0-15

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


27
PG085 November 17, 2021 www.xilinx.com
Chapter 2: Product Specification

Usage
To configure a 4x4 switch where MI0 is sourced from SI1, MI1 is unused, MI2 is sourced from
SI3 and MI3 is sourced from SI0, use the following code example:

# Setup registers
Write address offset 0x40, Data 0x1
Write address offset 0x44, Data 0x8000_0000
Write address offset 0x48, Data 0x3
Write address offset 0x4C, Data 0x0
# Commit registers
Write address offset 0x0, data 0x2

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


28
PG085 November 17, 2021 www.xilinx.com
Chapter 3

Designing with the Core


This chapter includes guidelines and additional information to make designing with the
core easier.

General Design Guidelines


When designing systems using AXI4-Stream Infrastructure IP Suite, the first step is to
establish the topology of the system. This requires an understanding of the main interface
characteristics of each AXI4-Stream master and slave that needs to be able to communicate
together. AXI4-Stream masters and slaves then need to be grouped by desired connectivity
into a system with one or more AXI4-Stream Infrastructure IP blocks tying them together so
they can exchange data.

In general, try to establish the system partitioning/topology to use multiple smaller/simpler


interconnects than a single large interconnect, especially in systems with a large number of
devices. For example, combinations of N master x 1 slaves interconnects and 1 master x N
slave interconnects are generally preferable to MxN interconnects in terms of area, latency,
and throughput. MxN interconnect when required for performance or connectivity
requirements should try to limit the number of endpoints or specify sparse connectivity to
reduce resource utilization.

The Xilinx Vivado AXI Reference Guide (UG1037) [Ref 3] provides information about
AXI4-Stream protocol usage guidelines and conventions; much of the AXI system
optimizations information described for AXI Interconnect is applicable to AXI4-Stream
Infrastructure IP Suite. The Xilinx AXI Reference Guide should be reviewed and consulted
before designing or structuring systems around the AXI4-Stream Infrastructure IP.

After the number and topology of AXI4-Stream Infrastructure IP systems have been
determined, the next step is to tailor each AXI4-Stream Infrastructure IP system to have the
correct set of optional interface signals and set signals’ widths as needed. This sets up the
interface signal set for the AXI4-Stream Infrastructure IP to ensure that data can be
exchanged and routed as needed across the system.

Finally, the AXI4-Stream Infrastructure IP system should be optimized and fine-tuned to fit
its application. This includes tuning FIFOs, width converters, clock converters, arbiters, and
register slices (pipeline stages) as needed to balance area, timing, performance, and
ease-of-use.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


29
PG085 November 17, 2021 www.xilinx.com
Chapter 3: Designing with the Core

Clocking
Each AXI4-Stream Infrastructure IP module has a single clock input that must be driven with
the exception of AXI4-Stream Clock Converter and AXI4-Stream Data FIFO. These modules
allow for different clocks and must have both S_AXIS_ACLK and M_AXIS_ACLK
connected.

The AXI4-Stream Clock Converter and Data FIFO allows systems with different clock
domains to be designed. AXI4-Stream Clock Converter synchronous mode can be used
when the endpoint IP has a phase-aligned, integer multiple clock ratio to the core switch's
clock. This is often the case when the same MMCM or PLL is driving synchronous integer
ratio clocks because the MMCM/PLL can also ensure phase alignment across their clock
outputs. The AXI4-Stream Clock Converter and AXI4-Stream Data FIFO asynchronous
clocking mode allows the attached endpoint IP to run at a completely unrelated clock
frequency or phase to the core switch clock.

An optional feature for clock enables (ACLKEN ports) allows an extra level of control for
essentially gating clocks. The clock enable signals can be used to control which clock edges
are seen as real transfer cycles. Clock enables can be used for purposes like preserving
global clock buffers, debug by stepping the clock enable to step through operational states
in the system, and dynamic power savings. Clock enables can be controlled independently
on each interface port and for the core switch.

The optional AXI4-Lite control register interface operates asynchronously from the
AXI4-Stream clocks.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


30
PG085 November 17, 2021 www.xilinx.com
Chapter 3: Designing with the Core

Resets
The AXI4-Stream Infrastructure IP Suite provides active-Low reset inputs for every clock
input on the IP. Each reset input must be synchronized to the associated ACLK input of the
interface. To ensure data is not lost during reset deassertion across multiple interfaces of
the AXI4-Stream Infrastructure IP systems (operating in potentially different clock domains),
the AXI4-Stream Infrastructure IP deasserts all TREADY and TVALID outputs until the clock
cycle after their source logic has internally exited reset. Any endpoint IP driving TREADY or
TVALID inputs to the AXI4-Stream Infrastructure IP should also deassert these signals until
the clock cycle after they have exited reset internally.

These guidelines ensure that endpoint IPs can internally come out of reset at different times
(due to internal reset pipelining) and no data will be exchanged until both are internally out
of reset.

RECOMMENDED: Any endpoint should deassert TREADY and TVALID within 8 clock cycles of reset
assertion. ARESETn should also be asserted for at least 16 cycles of the slowest system clock to ensure
that all AXI4-Stream interfaces in the system enter reset and have time to deassert their TREADY/
TVALID outputs before coming back out of reset.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


31
PG085 November 17, 2021 www.xilinx.com
Chapter 4

Design Flow Steps


This chapter describes customizing and generating the core, constraining the core, and the
simulation, synthesis and implementation steps that are specific to this IP core. More
detailed information about the standard Vivado® design flows in the Vivado IP integrator
can be found in the following Vivado Design Suite user guides:

• Vivado Design Suite User Guide: Designing IP Subsystems using IP Integrator (UG994)
[Ref 9]
• Vivado Design Suite User Guide: Designing with IP (UG896) [Ref 5]
• Vivado Design Suite User Guide: Getting Started (UG910) [Ref 7]
• Vivado Design Suite User Guide: Logic Simulation (UG900) [Ref 8]

Customizing and Generating the Core


This section includes information about using Xilinx® tools to customize and generate the
core in the Vivado Design Suite.

If you are customizing and generating the core in the IP integrator, see the Vivado Design
Suite User Guide: Designing IP Subsystems using IP Integrator (UG994) [Ref 9] for detailed
information. IP integrator might auto-compute certain configuration values when
validating or generating the design. To check whether the values do change, see the
description of the parameter in this chapter. To view the parameter value, run the
validate_bd_design command in the Tcl console.

You can customize the IP for use in your design by specifying values for the various
parameters associated with the IP core using the following steps:

1. Select the IP from the IP catalog.


2. Double-click on the selected IP or select the Customize IP command from the toolbar or
popup menu.

For details, see the sections, “Working with IP” and “Customizing IP for the Design” in the
Vivado Design Suite User Guide: Designing with IP (UG896) [Ref 5] and the “Working with the
Vivado IDE” section in the Vivado Design Suite User Guide: Getting Started (UG810) [Ref 7].

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


32
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

If you are customizing and generating the core in the IP integrator, see the Vivado Design
Suite User Guide: Designing IP Subsystems Using IP Integrator (UG994) [Ref 9] for detailed
information. IP integrator might auto-compute certain configuration values when
validating or generating the design. To check whether the values do change, see the
description of the parameter in this chapter. To view the parameter value you can run the
validate_bd_design command in the Tcl console.

Note: Figures in this chapter are illustrations of the Vivado IDE. This layout might vary from the
current version.

Creating a Project
First, create a new project using the Vivado Design Suite. For detailed information on
starting and using the Vivado Design Suite, see the Vivado documentation.

Perform the following steps:

1. Start the Vivado Design Suite.


2. Choose File > New Project from the menu.
3. Using the Create New Vivado Project wizard. Click Next.
4. Select a Project Name:
a. Modify the project name, if desired.
b. Specify the Project location using the text box or the directory navigator.
c. Click Next.
5. Select a Project Type:
a. Choose RTL Project.
b. Ensure the "Do not specify sources at this time" checkbox is checked.
c. Click Next.
6. Select the Default Part.
a. Select the target family, device, package, and speed grade.
Example: Kintex-7, xc7k325t-ffg900-1
Note: If an unsupported family is selected, the IP core does not appear in the IP catalog.
b. Click Next.
7. Review Project Summary.
a. If the summary does not look correct, click the Back button and resolve the issue.
b. Click Finish to create the project.

After creating the project, the IP cores are available for selection in the IP catalog, located
at AXI Infrastructure Taxonomy.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


33
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Customizing and Generating the Core


Locate the IP core in the Vivado Design Suite and click it once to select it. Details regarding
the solution are displayed in the Details window. To configure the IP, double-click the IP
name in the IP catalog.

This launches the customization dialog box. Each dialog box is described in detail in their
respective sections.

AXI4-Stream Broadcaster
The AXI4-Stream Broadcaster GUI is shown in Figure 4-1.
X-Ref Target - Figure 4-1

Figure 4-1: AXI4-Stream Broadcaster Customization Dialog Box

Review each of the available options in Figure 4-1 and modify them as desired so that the
AXI4-Stream Broadcaster solution meets the requirements of the larger project into which it
is integrated. The following subsections discuss the options in detail to serve as a guide.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


34
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Global Parameters

Component Name: The base name of the output files generated for the core. Names must
begin with a letter and can be composed of any of the following characters: a to z, 0 to 9,
and "_".

Number of Master Interfaces: This parameter specifies the number of AXI4-Stream master
interfaces present on the IP. The value can be 2 and 16. AXI4-Stream slave interface
transfers are replicated to each of the AXI4-Stream master interfaces specified by this
parameter.

Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

Enable TREADY: If set to Yes, this parameters specifies if the optional TREADY signal is
present on all the AXI4-Stream interfaces. Set to No to omit the signal.

TIP: The AXI4-Stream specification recommends that TREADY is always included.

SI TDATA Width (bytes): This parameter specifies the width in bytes of the TDATA signal
on the inbound AXI4-Stream Interface (the Slave Interface). This parameter is an integer and
can vary from 0 to 512. Set to 0 to omit the TDATA signal. If the TDATA signal is omitted,
then the TKEEP and TSTRB signals are also omitted. The width of the port is multiplied by
8 to get the width in bits.

MI TDATA Width (bytes): This parameter specifies the width in bytes of the TDATA signal
on each of the outbound AXI4-Stream Interfaces (the Master Interfaces). This parameter is
an integer and can vary from 0 to 512, but it may only be set to 0 if the SI TDATA Width is
also 0. When both parameters are set to 0, the TDATA signal is omitted from the interfaces
of the IP. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the TDATA signal on each port is multiplied by eight to get the width
in bits. If the MI TDATA width is not equal to the SI TDATA Width then TKEEP and TSTRB
signals should not be enabled.

Enable TSTRB: If set to Yes, this parameters specifies if the optional TSTRB signal is present
on all the AXI4-Stream interfaces. This option can only be enabled if the SI and MI TDATA
width (bytes) parameter are both greater than 0 and each have the same width value.

Enable TKEEP: If set to Yes, this parameters specifies if the optional TKEEP signal is present
on all the AXI4-Stream interfaces. This option can only be enabled if the SI and MI TDATA
width (bytes) parameter are both greater than 0 and each have the same width value.

Enable TLAST: If set to Yes, this parameters specifies if the optional TLAST signal is present
on all the AXI4-Stream interfaces.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


35
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TID Width (bits): If greater than 0, this parameter specifies if the optional TID signal is
present on all the AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets
the width of this signal accordingly.

TDEST Width (bits): If greater than 0, this parameter specifies if the optional TDEST signal
is present on all the AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32
sets the width of this signal accordingly.

TUSER Width (bits): If greater than 0, this parameter specifies if the optional TUSER signal
is present on all the AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32
sets the width of this signal accordingly.

Enable ACLKEN: If set to Yes, this parameter specifies if the optional ACLKEN signal is
present with all the AXI4-Stream interfaces clocks.

Stream Splitting Options

The second page of the AXI4-Stream Broadcaster GUI is shown in Figure 4-2.

X-Ref Target - Figure 4-2

Figure 4-2: AXI4-Stream Broadcaster Customization Dialog Box, Page 2

The second page of configuration parameters control the remapping feature. Two remap
strings are shown for each enabled interface. The TDATA Remap String for each port
controls which bits from the S_AXIS interface's TDATA signal are present on the
corresponding master interface's TDATA signal. Similarly the TUSER Remap String for each
port controls which bits from the S_AXIS interface's TUSER signal are present on the
corresponding master interface's TUSER signal.

The remap string parameters are edited as regular text and are encoded in a syntax very
similar Verilog vector concatenation. Each remap string is a comma-separated list of
elements and each element is either a bit-constant (eg, 1'b0) or a bit slice of the source
signal (for example, tdata[7:0] for TDATA remap strings or tuser[1] for TUSER remap
strings). The same bit from the source signal may be mapped multiple times. For example,
the remap string "tdata[7:0],tdata[7:0],tdata[7:0]" replicates the first byte of the S_AXIS
tdata signal three times in the tdata signal of the corresponding master interface. It is also
acceptable to mix bit-constant and bit-slice elements in the remap string. For example, the
remap string "tdata[7:0],8'b0,tdata[7:0]" replicates the first byte from the S_AXIS interface
with a 1 byte constant in the second byte position of the corresponding master interface.
The total width of the expression specified must match the width of the master interface
signal specified in the M_TDATA_NUM_BYTES or M_TUSER_WIDTH parameter. If the

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


36
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

S_AXIS signal width is 0 and the width of the corresponding master interface signal is
greater than 0, the remap expression must be a bit-constant covering the full width of the
master interface signal.

The IP configuration GUI automatically updates the remap strings to safe defaults whenever
the Number of Master Interfaces, SI TDATA Width, MI TDATA Width, SI TUSER Width
or MI TUSER Width parameters are changed. If any custom remap strings are to be used in
the IP configuration, it is recommended the custom remap strings are entered specified
after the above parameters have been set to their desired value.

AXI4-Stream Clock Converter


Review each of the available options in Figure 4-3 and modify them as desired so that the
AXI4-Stream Clock Converter solution meets the requirements of the larger project into
which it is integrated.
X-Ref Target - Figure 4-3

Figure 4-3: AXI4-Stream Clock Converter Customization


The following subsections discuss the options in detail to serve as a guide.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


37
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Global Parameters

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

Asynchronous Clocks

If set to Yes, then the S_AXIS_ACLK and M_AXIS_ACLK clock signals are assumed to be
asynchronous to each other and the IP operates in asynchronous mode. The Slave
Interface Clock Frequency Ratio and Master Interface Clock Frequency Ratio options
are disabled when the clocks are specified as asynchronous. Asynchronous clock mode
should be used unless S_AXIS_ACLK and M_AXIS_ACLK are phase aligned and have
frequencies that have either 1:N or N:1 clock frequency ratio.

SI to MI Clock Ratio

This parameter set the ratio in frequency between the S_AXIS_ACLK and the
M_AXIS_ACLK inputs if they are not asynchronous.

Synchronization Stages across Cross Clock Domain Logic

When S_AXIS_ACLK and M_AXIS_ACLK are asynchronous to each other, then this
parameter specifies the number of synchronization stages to use for cross-clock domain
logic. Increasing this value increases the MTBF of design, but incurs increased latency and
logic utilization.

ACLKEN Conversion Mode

This pull-down option selects the conversion mode for the ACLKEN signal. Extra latency and
logic is incurred when ACLKEN conversion is performed. The options are:

• None - There are no ACLKEN signals associated with the IP.


• S AXIS Only - There is an S_AXIS_ACLKEN signal associated with the S_AXIS_ACLK
clock signal and no M_AXIS_ACLKEN signal.
• M AXIS Only - There is an M_AXIS_ACLKEN signal associated with the M_AXIS_ACLK
clock signal and no S_AXIS_ACLKEN signal.
• S AXIS and M AXIS - Both clocks have ACLKEN signals associated with them.

Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


38
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on all the AXI4-Stream
interfaces. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


39
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

AXI4-Stream Combiner
Review each of the available options in Figure 4-4 and modify them as desired so that the
AXI4-Stream Combiner solution meets the requirements of the larger project into which it is
integrated.
X-Ref Target - Figure 4-4

Figure 4-4: AXI4-Stream Combiner Customization


The following subsections discuss the options in detail to serve as a guide.

Global Parameters

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

Number of Slave Interfaces

This parameter specifies the number of AXI4-Stream slave interfaces present on the IP. The
value can be 2 and 16. The AXI4-Stream master interface signals TDATA, TSTRB, TKEEP, and
TUSER width are multiplied by this value.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


40
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Signal Properties
When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on each of the AXI4-Stream
slave interfaces. The AXI4-Stream master interface TDATA width is this value multiplied by
the Number of Slave Interfaces parameter. This parameter is an integer and can vary from 0
to (512/Number of Slave Interfaces). Set to 0 to omit the TDATA signal. If the TDATA signal
is omitted, then the TKEEP and TSTRB signals are also omitted. The width of the port is
multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly on the AXI4-Stream slave interfaces. The AXI4-Stream master interface
TUSER width is to set to this value multiplied by Number of Slave Interfaces.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


41
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

Primary Slave Interface

This option specifies the Slave Interface that is used to pass the TLAST, TID, and TDEST
signals to the master interface.

Enable Command Port Error

If set to Yes, this option enables the s_cmd_err port if TID, TDEST, TLAST signals do not
match the slave interface specified by the Primary Slave Interface option.

AXI4-Stream Data FIFO


Review each of the available options in Figure 4-5 and modify them as desired so that the
AXI4-Stream Data FIFO solution meets the requirements of the larger project into which it is
integrated.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


42
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

X-Ref Target - Figure 4-5

Figure 4-5: AXI4-Stream Data FIFO Customization


The following subsections discuss the options in detail to serve as a guide.

General Options

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

FIFO depth

This option specifies the depth of the FIFO to be instantiated. FIFO depth can vary between
16 and 32768 (powers of 2 up to 2 n for whichever value of n is the current maximum.) Large
FIFO depths may not be realizable on certain devices.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


43
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Memory type

Designate the FIFO memory primitive (resource type) to use. Allowable options: Auto -
Allow Vivado Synthesis to choose. Distributed RAM - Distributed RAM FIFO (does not
support ECC.) Block RAM - Block RAM FIFO. UltraRAM - UltraRAM FIFO (does not support
independent clocks.)

Independent clocks

If set to Yes, then the S_AXIS_ACLK and M_AXIS_ACLK clock signals are assumed to be
asynchronous to each other and the IP operates in asynchronous mode.

CDC (Clock Domain Crossing) sync stages

When S_AXIS_ACLK and M_AXIS_ACLK are asynchronous to each other, then this
parameter specifies the number of synchronization stages to use for cross clock domain
logic. Increasing this value increases the MTBF of design, but incurs increased latency and
logic utilization.

Enable packet mode

This option enables the Packet Mode option of the FIFO when set to Yes. This option
requires the TLAST signal to be enabled. The FIFO operation in Packet Mode is modified to
store transfers until the TLAST signal is asserted. When the TLAST signal is asserted or the
FIFO is full, the store transfers are presented on the AXI4-Stream master interface.

ACLKEN conversion mode

This pull-down option selects the conversion mode for the ACLKEN signal. Extra latency and
logic is incurred when ACLKEN conversion is performed. The options are:

• None - There are no ACLKEN signals associated with the IP.


• S AXIS Only - There is an S_AXIS_ACLKEN signal associated with the S_AXIS_ACLK
clock signal and no M_AXIS_ACLKEN signal.
• M AXIS Only - There is an M_AXIS_ACLKEN signal associated with the M_AXIS_ACLK
clock signal and no S_AXIS_ACLKEN signal.
• S AXIS & M AXIS - Both clocks have ACLKEN signals associated with them.

Enable ECC

Enables both ECC Encoder and Decoder. ECC enablement only supported on Block RAM and
UltraRAM primitive types.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


44
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Include ECC error injection

Signal properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

TDATA width (bytes)

This parameter specifies the width in bytes of the TDATA signal on all the AXI4-Stream
interfaces. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

TID width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


45
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Flags
X-Ref Target - Figure 4-6

Figure 4-6: Flags


Enable write data count

Include write data (s_axis) data count output port.

Enable almost full

Include almost full output port. Can only be enabled if Packet Mode is not used.

Enable programmable full

Include programmable full output port.

Programmable full threshold

Configures the programmable full threshold value. Minimum value: 5+(Number of CDC
sync stages if using independent clocks), maximum value: (FIFO depth)-5.

Enable read data count

Include read data (m_axis) data count output port.

Enable almost empty

Include almost empty output port. Can only be enabled if Packet Mode is not used.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


46
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Enable programmable empty

Include programmable empty output port.

Programmable empty threshold

Configures the programmable empty threshold value. Minimum value: 5, maximum value:
(FIFO depth)-5.

AXI4-Stream Data Width Converter


Review each of the available options in Figure 4-7 and modify them as desired so that the
AXI4-Stream Data Width Converter solution meets the requirements of the larger project
into which it is integrated. The following subsections discuss the options in detail to serve
as a guide.
X-Ref Target - Figure 4-7

Figure 4-7: AXI4-Stream Data Width Converter Customization


Global Parameters

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


47
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Slave Interface TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on the AXI4-Stream slave
interface. This parameter is an integer and can vary from 1 to 512. The width of the port is
multiplied by 8 to get the width in bits. This value cannot be the same as the value of Master
Interface TDATA Width (bytes). When using Vivado IP integrator, the Vivado IDE
automatically computes the value of this parameter.

Master Interface TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on the AXI4-Stream master
interface. This parameter is an integer and can vary from 1 to 512. The width of the port is
multiplied by 8 to get the width in bits. This value cannot be the same as the value of Slave
Interface TDATA Width (bytes).

Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters except for Enable Aclken.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. Certain configurations may introduce null-bytes into the system. In
these situations, the m_axis_tkeep signal is always present regardless of the value of this
parameter.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


48
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

AXI4-Stream Register Slice


Review each of the available options in Figure 4-8 and modify them as desired so that the
AXI4-Stream Register Slice solution meets the requirements of the larger project into which
it is integrated.
X-Ref Target - Figure 4-8

Figure 4-8: AXI4-Stream Register Slice Customization


The following subsections discuss the options in detail to serve as a guide.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


49
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

Enable TREADY

If set to Yes, this parameters specifies if the optional TREADY signal is present on all the
AXI4-Stream interfaces. Set to No to omit the signal.

TIP: The AXI4-Stream specification recommends that TREADY is always included.

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on all the AXI4-Stream
interfaces. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


50
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

Advanced Properties

Register Pipeline Type

This property allows for the trade-off between performance and area efficiency.

Default: A two-deep registered mode (supports back-to-back transfers) that balances


performance with low input fanout, as used in the AXI4-Stream Switch.

Fully-registered: Similar to Default (supports back-to-back transfers), except all payload


and handshake outputs are driven directly from registers.

Light-weight: Inserts one bubble cycle after each transfer.

SLR Crossing: Adds extra pipeline stages to optimally cross one super logic region (SLR)
boundary in stacked silicon interconnect (SSI) devices. All SLR crossings are flop-to-flop
with fanout=1. See Floor Planning Constraints for AXI4-Stream Register Slice SLR Crossing
Modes for floorplanning guidance.

SLR TDM Crossing: Similar to SLR Crossing, except it consumes half number of payload
wires across the SLR boundary and propagates the cross-SLR signals at twice the frequency
of the AXI interfaces.

Bypass: Directly connects the slave interface to the master interface.

Bypass-Endpoint: Similar to Bypass, except for allowing interface properties to be treated


as fixed values during IP integrator design validation.

Multi SLR Crossing: Supports spanning zero or more SLR boundaries using a single
Register Slice instance. Also inserts additional pipeline stages within each SLR to help meet
timing goals.

Number of SLR Crossings: When using Multi SLR Crossing mode, select the number of SLR
boundaries to be crossed within the core.

Pipeline Stages within Master-side SLR: When using Multi SLR Crossing mode, select the
number of additional pipeline stages to insert within the master-side SLR (between the SLR
boundary and the SI interface).

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


51
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Pipeline Stages within Slave-side SLR: Select the number of additional pipeline stages to
insert within the slave-side SLR (between the SLR boundary and the MI interface).

Pipeline Stages within Middle SLR: Select the number of additional pipeline stages to
insert within each middle SLR (between the two SLR boundaries). Enabled only when
Number of SLR Crossings is >1.

AXI4-Stream Subset Converter


Review each of the available options in Figure 4-9 and modify them as desired so that the
AXI4-Stream Subset Converter solution meets the requirements of the larger project into
which it is integrated.
X-Ref Target - Figure 4-9

Figure 4-9: AXI4-Stream Subset Converter Customization


The following subsections discuss the options in detail to serve as a guide.

Global Parameters

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

Slave Interface Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


52
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Enable TREADY

If set to Yes, this parameters specifies if the optional TREADY signal is present on the
AXI4-Stream slave interface. Set to No to omit the signal.

TIP: The AXI4-Stream specification recommends that TREADY is always included.

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on the AXI4-Stream slave
interface. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on the
AXI4-Stream slave interface. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on the
AXI4-Stream slave interface. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on the
AXI4-Stream slave interface.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


53
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Master Interface Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

Enable TREADY

If set to Yes, this parameters specifies if the optional TREADY signal is present on the
AXI4-Stream slave interface. Set to No to omit the signal.

TIP: The AXI4-Stream specification recommends that TREADY is always included.

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on the AXI4-Stream slave
interface. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on the
AXI4-Stream slave interface. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on the
AXI4-Stream slave interface. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on the
AXI4-Stream slave interface.

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


54
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on the
AXI4-Stream slave interface. A value of 0 omits this signal. Values 1 and 32 sets the width of
this signal accordingly.

Extra Settings

TDATA/TUSER/TID/TDEST/TSTRB/TKEEP/TLAST Remap String

The remap string parameters are used to remap input to output bytes/bits of the
TDATA/TUSER/TID/TDES/TSTRB/TKEEP/TLAST signals, respectively.

The format of the remap user parameter follows syntax similar to Verilog vector
concatenation.

• The remap parameter is a comma separated list of elements.


• Each element is either a constant or a bit slice of one or more SI signals.

° For example, if SI and MI TDATA width have the same value 32, the remap string
tdata[31:0] passes the entire TDATA signal from SI to MI unmodified.

° If the MI TDATA signal were 24 bits and the SI TDATA signal were 16 bits, a remap
string of 8b00000000,tdata[15:0] assigns a constant 0 to the upper 8 bits of
the MI signal and pass the SI TDATA through on the lower 16 bits.
• A bit-slice element can reference a single bit of an SI signal (e.g., TDATA[0]) or a vector
slice of an SI signal (e.g., tdata[11:8]).

° The index values must not exceed the indices of the SI signal.
• The same SI bit can be mapped multiple times into the MI output.

° For example, tdata[7:0], tdata[7:0] repeats the SI TDATA least significant


byte twice on MI signal.
• The combined width of the constants and the SI bit slices in the remap parameter must
match the MI signal width.
• The right-most element defines the least significant bits of the MI signal and the
left-most element defines the most significant bits of the MI signal.
• The remap string can reference bits of the corresponding SI signal or bits from another
remappable SI signal.

° For example, the remap string for TDATA may reference bits from the TUSER SI
signal.
• Constant elements and bit-slice elements can be freely mixed in the comma separated
list of elements.

° For example, tdata[7:0],8b00000000,tdata[15:8] is a valid expression.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


55
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

• Only binary format constant elements are supported.

° h format is not supported.


• The constant element must specify the number of bits that follow b and the number of
bits that follow b must match the number of bits specified. If not, a validation error is
given.
• If the SI signal width is 0, the remap string must be a constant element.
• If the MI signal width is 0, the remap string is 1b0.
• Bit slices are only valid when the SI signal width is greater than 0.
• When using IP inside of Vivado IP integrator, both m_axis and s_axis port widths of
the remapped value should be set to manual override if write_bd_tcl is used.

Generate TLAST

This parameter can be set if the TLAST signal is enabled on the master interface, but not on
the slave interface. The number specifies how many transfers to count before asserting the
TLAST signal. A value of 0 indicates that the TLAST signal should always be de-asserted.
Conversely, a value of 1 indicates that the TLAST signal should be asserted on every
transfer. A value of 2 indicates that the TLAST signal should be asserted on every other
transfer and so on. Values 0 and 256 are accepted.

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

AXI4-Stream Switch
Review each of the available options in Figure 4-10 and modify them as desired so that the
AXI4-Stream Switch solution meets the requirements of the larger project into which it is
integrated.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


56
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

X-Ref Target - Figure 4-10

Figure 4-10: AXI4-Stream Switch Customization


The following subsections discuss the options in detail to serve as a guide.

Global Parameters

Component Name

The base name of the output files generated for the core. Names must begin with a letter
and can be composed of any of the following characters: a to z, 0 to 9, and "_".

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


57
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Switch Properties

Number of Slave Interfaces

This parameter specifies the number of AXI4-Stream slave interfaces present on the IP. The
value can be 1 and 16. This value cannot be set to 1 when Number of Master Interfaces is 1.

Number of Master Interfaces

This parameter specifies the number of AXI4-Stream master interfaces present on the IP.
The value can be 1 and 16. This value cannot be set to 1 when Number of Slave Interfaces
is 1.

Use Control Register Routing

This parameter specifies the routing mode. Control register routing enables the AXI4-Lite
control register interface to handle transfer routing. If this option is set to No, then the
inline TDEST value is used for routing. Enabling this option disables the Routing tab.

Signal Properties

When using Vivado IP integrator, the Vivado IDE automatically computes the values of these
parameters.

TDATA Width (bytes)

This parameter specifies the width in bytes of the TDATA signal on all the AXI4-Stream
interfaces. This parameter is an integer and can vary from 0 to 512. Set to 0 to omit the
TDATA signal. If the TDATA signal is omitted, then the TKEEP and TSTRB signals are also
omitted. The width of the port is multiplied by 8 to get the width in bits.

Enable TSTRB

If set to Yes, this parameters specifies if the optional TSTRB signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TKEEP

If set to Yes, this parameters specifies if the optional TKEEP signal is present on all the
AXI4-Stream interfaces. This option can only be enabled if the TDATA Width (bytes)
parameter is greater than 0.

Enable TLAST

If set to Yes, this parameters specifies if the optional TLAST signal is present on all the
AXI4-Stream interfaces.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


58
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

TID Width (bits)

If greater than 0, this parameter specifies if the optional TID signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TDEST Width (bits)

If greater than 0, this parameter specifies if the optional TDEST signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

TUSER Width (bits)

If greater than 0, this parameter specifies if the optional TUSER signal is present on all the
AXI4-Stream interfaces. A value of 0 omits this signal. Values 1 and 32 sets the width of this
signal accordingly.

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

Data Flow Properties

Data Flow Properties options are available to modify only if the Number of Slave Interfaces
is greater than 1 and Use control register routing is set to No.

Arbitrate on Maximum Number of Transfers

This setting specifies how many transfers to count before relinquishing the granted
arbitration. If set to zero, then the number is infinite and Arbitrate on TLAST transfer must
be set. If set to one, then after each transfer, the interconnect switch relinquishes control
and requests another arbitration. If set to a value greater than one, then after the specified
number of transfers, the arbitration is relinquished. For example, setting this value to 16
allows 16 transfers to pass from slave interface to master interface per each arbitration
grant.

Arbitrate on Number of LOW TVALID Cycles

This setting allows relinquishing of a granted arbitration without a transfer. A watchdog


timer is instantiated and counts the number of consecutive LOW TVALID signals the
granted SI and MI interfaces. When the requisite number of LOW TVALID cycles is counted
as specified here, the switch relinquishes the granted arbitration and signals that the
transaction is complete to the arbiter. If this setting is set to 0, no watchdog timer is
instantiated. If there is more than one slave interface, more than one master interface, and
the Arbitrate on maximum number of transfers setting is greater than 1, this setting cannot
be set to zero. This ensures that deadlock cannot occur.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


59
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Arbitrate on TLAST Transfer

If checked, this setting indicates that a transaction is complete when a transfer with TLAST
asserted passes from the slave interface to the master interface of the interconnect switch.
The arbiter is then able to arbitrate to the next arbitration winner.

Arbiter Algorithm

There are three arbitration algorithms available to choose from. The True Round-Robin
algorithm arbitrates between all the slave interfaces in a round robin fashion. If all slave
interfaces are not active, then it adapts to provide an equal weighting for each active slave
interface. The ordering starts with S00, then S01, down to S15. The slave interface following
the one that was last granted has the highest priority next. The Round-Robin algorithm
operates similarly to the True Round-Robin option except that after a successful grant, it
proceeds to the next arbitration order regardless of which slave interface was granted.
Fixed-Priority algorithm issues the highest priority to S00, and the lowest priority to S15 for
every arbitration cycle. System-level traffic profiles for Fixed-Priority algorithm must be
carefully understood to avoid starvation of lower-priority ports.

Consider the scenario where there is a 4 SI to 1 MI switch with ports S00, S02, and S03
continuously requesting transfers with large transactions. After a large number of
arbitrations, the algorithms produces different utilizations. The True Round-Robin provides
33% of the available bandwidth to each of the ports S00, S02, and S03. The Round-Robin
algorithm provides 25% of the available bandwidth to ports S00 and S03, and 50% to port
S02. This is because port S02 gains ports S01 bandwidth. The Fixed-Priority algorithm
provides 100% bandwidth to port S00, and starve the other 2 requesting ports.

Regardless of arbiter algorithm, the arbiter does not support back-to-back transfers from
the same slave interface. This configuration has two implications. If only one interface is
requesting to a particular master, then only 50% of the bandwidth will be realized. If using
the fixed-priority arbitration in the scenario described previously, then the ports S00 and
S02 would get 50% bandwidth allocation.

Pipeline Registers

Pipeline register options are available to set when using the Use control register routing
is set to Yes.

Enable Input Pipeline Register

When enabled, the slave interface side of the switch has register slices for each port at the
IP boundary.

Enable Output Pipeline Register

When enabled, the master interface side of the switch has register slices for each port at the
IP boundary.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


60
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Connectivity
X-Ref Target - Figure 4-11

Figure 4-11: AXI4-Stream Switch Customization Dialog Box - Connectivity Tab


Slave to Master Interface Switch Connectivity Map

The table shown Figure 4-11 allows the configuration of the switch to optimize
connectivity. By default full connectivity is enabled meaning that every slave interface is
connected to every master interface. Deselecting a checkbox removes the connection from
the slave interface listed in the column to the master interface in the row. Removing
connections is desirable when it is known that a slave interface is never going to send
transfers to a particular master interface. This removes unnecessary logic resulting in
smaller area utilization, routing and logic complexity.

IMPORTANT: No interface should be left without connectivity. Every slave interface must have
connectivity to at least one master interface. Conversely, every master interface must have connectivity
to at least one slave interface.

Routing
X-Ref Target - Figure 4-12

Figure 4-12: AXI4-Stream Switch Customization - Routing Tab

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


61
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Routing Parameters

The routing parameters shown in Figure 4-12 set up the decoding used by the switch to
route slave interface transfers to master interfaces based on the TDEST signal. Each master
interface must have a BASE/HIGH range that is within the valid width of the TDEST width
specified previously. The BASE/HIGH pair must not overlap master interfaces. When a
transfer is received at the slave interface, the TDEST is decoded based on this table. A
request is then sent to arbiter for the master interface corresponding to the TDEST. If the
slave interface is chosen by the arbiter, the transfer proceeds to the master interface. This
tab is only available when Use control register routing is set to No.

AXI4-Stream Interconnect
Note: AXI4-Stream Interconnect requires IP integrator.

Figure 4-13 shows the top-most AXI4-Stream Interconnect core block diagram. Inside the
AXI4-Stream Interconnect, an AXI4-Stream Switch core routes traffic between the Slave
Interfaces (SI) and Master Interfaces (MI). Along each pathway connecting a SI or MI to the
Switch, an optional series of AXI4-Stream Infrastructure cores (couplers) can perform
various conversion and buffering functions. The couplers include: AXI4-Stream Register
Slice, AXI4-Stream Data FIFO, AXI4-Stream Clock Converter, AXI4-Stream Data Width
Converter and AXI4-Stream Protocol Converter.
X-Ref Target - Figure 4-13

Figure 4-13: AXI4-Stream Interconnect


The AXI4-Stream Interconnect core can be configured to have up to 16 Slave Interfaces (SI)
and up to 16 Master Interfaces (MI). Each SI connects to one AXI4-Stream master device and
accepts transfers from the connected master device. Each MI connects to one AXI4-Stream
slave device and issues transfers to slave devices. At the center is the Switch core that routes
transfers between the SI and MI. Along each of the pathways between an SI and the Switch,

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


62
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

or between the Switch and an MI, there can be one or more AXI4-Stream infrastructure
cores to perform various conversion and storage functions.

The Switch effectively splits the AXI4-Stream Interconnect core down the middle between
the SI-related functional units (SI hemisphere) and the MI-related units (MI hemisphere).
Where possible, Vivado system design tools automatically insert couplers into the SI or MI
hemisphere to resolve differences in the configuration of the connected master and slave
devices.

Generating a Project for AXI4-Stream Interconnect

AXI4-Stream Interconnect v2.1 requires IP integrator. To access the IP, first create a Vivado
project, then select Create Block Design from the Vivado Flow Navigator. In the block
design canvas, select the Add IP option from the toolbar and choose AXI4-Stream
Interconnect from the IP integrator IP catalog window. An AXI4-Stream Interconnect IP
instance is added to the block design canvas. Double-clicking on the IP instance in the
diagram opens its customization window. The user-configurable options of the
AXI4-Stream Interconnect are described in the following sections.
X-Ref Target - Figure 4-14

Figure 4-14: Top-level Settings


Top-level Settings

Number of Slave Interfaces

This parameter specifies the number of AXI4-Stream slave interfaces present on the IP. The
value can be between 1 and 16.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


63
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Number of Master Interfaces

This parameter specifies the number of AXI4-Stream master interfaces present on the IP.
The value can be between 1 and 16.

Enable ACLKEN

If set to Yes, this parameter specifies if the optional ACLKEN signal is present with all the
AXI4-Stream interfaces clocks.

Use Control Register Routing

This parameter specifies the routing mode. Control register routing enables the AXI4-Lite
control register interface to handle transfer routing. If this option is set to No, then the
inline TDEST value is used for routing.

Data Flow Properties

Data Flow Properties options are available to modify only if the Number of Slave Interfaces
is greater than 1.

Arbitrate on Maximum Number of Transfers

This setting specifies how many transfers to count before relinquishing the granted
arbitration and is passed directly to the AXI4-Stream Switch within the AXI4-Stream
Interconnect instance. Consult the AXI4-Stream Switch parameter descriptions for more
details on the valid value range and use of this parameter.

Arbitrate on Number of LOW TVALID Cycles

This setting allows relinquishing of a granted arbitration without a transfer and is passed
directly to the AXI4-Stream Switch within the AXI4-Stream Interconnect instance. Consult
the AXI4-Stream Switch parameter descriptions for more details on the valid value range
and use of this parameter.

Arbitrate on TLAST Transfer

This setting allows relinquishing of a granted arbitration when a transfer with TLAST
asserted is received and is passed directly to the AXI4-Stream Switch within the
AXI4-Stream Interconnect instance. Consult the AXI4-Stream Switch parameter descriptions
for more details on the valid value range and use of this parameter.

Arbiter Algorithm

This setting allows the selection of an arbitration algorithm and is passed directly to the
AXI4_Stream Switch within the AXI4-Stream Interconnect instance. Consult the AXI4-Stream
Switch parameter descriptions for more details on the valid value range and use of this
parameter.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


64
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

X-Ref Target - Figure 4-15

Figure 4-15: Slave Interfaces


Enable Register Slice

If checked, an AXI4-Stream Register slice is inserted in the SI hemisphere couplers between


the Snn_AXIS interface and the AXI4-Stream Switch.

FIFO Depth

If a value greater than 0 is selected, an AXI4-Stream Data FIFO is inserted in the SI


hemisphere couplers between the Snn_AXIS interface and the AXI4-Stream Switch. The
selected depth of the FIFO is configured directly into the AXI4-Stream Data FIFO instance.

FIFO Packet Mode

If checked and if a FIFO depth greater than 0 is selected for Snn_AXIS, the AXI4-Stream Data
FIFO's packet mode is enabled. Consult the AXI4-Data FIFO parameter descriptions for more
details on packet mode operation.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


65
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

X-Ref Target - Figure 4-16

Figure 4-16: Master Interfaces


Enable Register Slice

If checked, an AXI4-Stream Register slice is inserted in the MI hemisphere couplers between


the AXI4-Stream Switch and Mnn_AXIS interface.

FIFO Depth

If a value greater than 0 is selected, an AXI4-Stream Data FIFO is inserted in the MI


hemisphere couplers between the AXI4-Stream Switch and the Mnn_AXIS interface. The
selected depth of the FIFO is configured directly into the AXI4-Stream Data FIFO instance.
Consult the AXI4-Stream Data FIFO parameter descriptions for more details on the valid
range of FIFO depth.

FIFO Packet Mode

If checked and if a FIFO depth greater than 0 is selected for Snn_AXIS, the AXI4-Stream
Data FIFO's packet mode is enabled. Consult the AXI4-Stream Data FIFO parameter
descriptions for more details on packet mode operation.

Base TDEST and High TDEST

The routing parameters shown in Figure 4-16 set up the decoding to route slave interface
transfers to master interfaces based on the TDEST signal and are passed directly to the
AXI4-Stream Switch instance within the AXI4-Stream Interconnect. Consult the AXI4-Stream
Switch parameter descriptions for more details on the valid range and use of routing

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


66
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

parameters. These parameters are not configurable when Use control register routing is
set to Yes.
X-Ref Target - Figure 4-17

Figure 4-17: Advanced Options Tab


Enable FIFO Count Ports

When 'Yes' is selected, the data count pins of any SI and MI AXI4-Stream Data FIFO
instances are made available as pins on the boundary of AXI4-Stream Interconnect, as
shown in Figure 4-17. Two sets of data counts per FIFO are available: the read count and the
write count. If the interface has enabled clock conversion, then these data counts are
flopped on their respective clocks. For slave interfaces, the write data count is synchronous
to the clock input for that interface and the read data count is synchronous to the ACLK
global clock. For master interfaces, the read data count is synchronous to the clock input for
that interface and the write data count is synchronous to the ACLK global clock.

Enable URAM Memory Type

When 'Yes' is selected, the AXI4-Stream Data FIFO inserted inside the AXI4-Stream
Interconnect gets configured with UltraRAM memory. This option is only visible for devices
supporting URAM

Synchronization Stages

Specifies the number of synchronization stages used in any asynchronous clock domain
conversion couplers instantiated within the AXI4-Stream Interconnect.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


67
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Output Generation
For details, see the Vivado Design Suite User Guide: Designing with IP (UG896) [Ref 5].

Constraining the Core


This section contains information about constraining the core in the Vivado Design Suite.

Required Constraints
The AXI4-Stream Infrastructure IP does not generally require any additional timing
constraints other than clock period constraints on its clocks inputs. When an asynchronous
clock converter is utilized, the underlying Xilinx Parameterized Macro is instantiated and
generates an internal constraint file to prevent timing paths crossing clock domains from
causing false timing errors. Those constraints apply only to internal logic inside the
AXI4-Stream Infrastructure IP and are automatically generated with the IP.

Floor Planning Constraints for AXI4-Stream Register Slice SLR


Crossing Modes
When using the AXI4-Stream Register Slice core in either the SLR Crossing, SLR TDM
Crossing, or Multi SLR Crossing modes, the constraints can be applied to explicitly floor
plan the submodules of the core into adjacent SLRs. This ensures that the SLR crossing
occurs between the intended flop-to-flop, unit-fanout, internal wires across all payload and
handshake pathways within the core.

SLR Crossing and SLR TDM Crossing Modes


After synthesis, all logic and registers placed into the "source" SLR (where the AXI4-Stream
master connected to the SI interface is located) contain the cell name pattern "*slr_source*".
All logic and registers placed into the "destination" SLR (where the AXI4-Stream slave
connected to the MI interface is located) contain the cell name pattern "*slr_dest*".
Constraints that combine the instance name of the AXI4-Stream Register Slice and either of
these sub-module name patterns can then be used to group all cells in the core into their
respective PBLOCKs for floorplanning.

In the following example, an AXI4-Stream Register Slice instance named "my_reg" is


constrained to cross a SLR boundary that exists in the target device between row Y4 (the
top of the lower SLR) and row Y5 (the bottom of the upper SLR):

create_pblock master_slr
add_cells_to_pblock [get_pblocks master_slr] [get_cells -hierarchical -filter
"NAME=~*my_reg*slr_source*"]
resize_pblock [get_pblocks master_slr] -add {CLOCKREGION_X0Y0:CLOCKREGION_X5Y4}
create_pblock slave_slr

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


68
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

add_cells_to_pblock [get_pblocks slave_slr] [get_cells -hierarchical -filter


"NAME=~*my_reg*slr_dest*"]
resize_pblock [get_pblocks slave_slr] -add {CLOCKREGION_X0Y5:CLOCKREGION_X5Y9}

Multi SLR Crossing Mode


After synthesis, all logic and registers that should be placed into the master-side SLR (where
the AXI4-Stream master connected to the SI interface is located) contain the cell name
pattern *slr_master*. All logic and registers that should be placed into the slave-side SLR
(where the AXI4-Stream slave connected to the MI interface is located) contain the cell
name pattern *slr_slave*. When spanning three SLRs, all logic and registers that should be
placed into the middle SLR contain the cell name pattern *slr_middle*. When spanning four
SLRs, all logic and registers that should be placed into the middle SLR adjacent to the
master contain the cell name pattern *slr_middle_master*, and all logic and registers that
should be placed into the middle SLR adjacent to the slave contain the cell name pattern
*slr_middle_slave*.

Constraints that combine the instance name of the Register Slice and any of these
submodule name patterns can then be used to group all cells in the core into their
respective PBLOCKs for floorplanning.

In the following example, an AXI4-Stream Register Slice instance named my_reg is


configured in MultiSLR Crossing mode to cross two SLR boundaries that exist in the target
device.

One of the boundaries exists between row Y4 (the top of the lower SLR) and row Y5 (the
bottom of the middle SLR). The other boundary exists between row Y9 (the top of the
middle SLR) and row Y10 (the bottom of the upper SLR).

create_pblock lower_slr
add_cells_to_pblock [get_pblocks lower_slr] [get_cells -hierarchical -filter
"NAME=~*my_reg*slr_master*"]
resize_pblock [get_pblocks lower_slr] -add {CLOCKREGION_X0Y0:CLOCKREGION_X5Y4}
create_pblock center_slr
add_cells_to_pblock [get_pblocks center_slr] [get_cells -hierarchical -filter
"NAME=~*my_reg*slr_middle*"]
resize_pblock [get_pblocks center_slr] -add {CLOCKREGION_X0Y5:CLOCKREGION_X5Y9}
create_pblock upper_slr
add_cells_to_pblock [get_pblocks upper_slr] [get_cells -hierarchical -filter
"NAME=~*my_reg*slr_slave*"]
resize_pblock [get_pblocks upper_slr] -add {CLOCKREGION_X0Y10:CLOCKREGION_X5Y14}

Device, Package, and Speed Grade Selections


This section is not applicable for this IP core.

Clock Frequencies
This section is not applicable for this IP core.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


69
PG085 November 17, 2021 www.xilinx.com
Chapter 4: Design Flow Steps

Clock Management
This section is not applicable for this IP core.

Clock Placement
This section is not applicable for this IP core.

Banking
This section is not applicable for this IP core.

Transceiver Placement
This section is not applicable for this IP core.

I/O Standard and Placement


This section is not applicable for this IP core.

Simulation
For comprehensive information about Vivado simulation components, as well as
information about using supported third party tools, see the Vivado Design Suite User
Guide: Logic Simulation (UG900) [Ref 8].

Synthesis and Implementation


For details about synthesis and implementation, see the Vivado Design Suite User Guide:
Designing with IP (UG896) [Ref 5].

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


70
PG085 November 17, 2021 www.xilinx.com
Chapter 5

Example Design
An example design that demonstrates basic core functionality for the customized IP is
available for AXI4-Stream Infrastructure IP cores. The example design is an independent
Vivado® project populated with the customized IP along with additional IPs including
example master(s), example slave(s), clocking and reset blocks. A synthesizable top-level
HDL file is provided that instantiates and wires together the IPs shown in Figure 5-1. If the
parent Vivado project is configured for a Xilinx supported board, then the physical board
constraints are also provided. A simulation-only demo test bench for the example design is
also provided and discussed in further in the Chapter 6, Test Bench.

IMPORTANT: The example design does not exhaustively demonstrate all the features of the IP. It is not
a verification test bench.

X-Ref Target - Figure 5-1

Figure 5-1: Schematic of the Example Design Top Level

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


71
PG085 November 17, 2021 www.xilinx.com
Chapter 5: Example Design

Functionality
The example design shows basic functionality by instantiating a synthesizable AXI4-Stream
example master which sends transfers to an AXI4-Stream example slave. After a fixed
number of transfers, the master completes and asserts the done output signal. The example
slave receives the transfers and performs a null operation on the payload to emulate data
processing. Once all transfers are received by the example slave the idle output signal is
asserted. The example master and example slave generates a subset of AXI4-Stream
protocol compliant transfers only. When the example master done output signal and the
example slave idle output signal are both asserted then the done output pin is asserted. If
this project is configured for a Xilinx® reference board then the done signal is tied to an LED
output.

A Clocking Wizard core is present to interface with an external differential clock input and
to provide a clock output suitable for the design to easily meet timing closure. If the project
is configured for a Xilinx reference board, then the Clocking Wizard is configured to specify
the differential clock input constraints for the board.

A LogiCore Processor System Reset block is present to interface with an external reset input
and to provide a de-bounced active Low system reset. If the project is configured for a
Xilinx reference board, then the Processor System Reset (Proc Sys Reset) core is configured
to specify the reset input constraint for the board.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


72
PG085 November 17, 2021 www.xilinx.com
Chapter 6

Test Bench
A behavioral Verilog test bench that wraps around the example design top level is provided
when the example design output product is generated. The test bench provides clocking
and reset stimulus to the example design top level to run simulations on the example
design. It monitors the done output to signal simulation completion. The test bench is
useful for getting familiar with the signaling on the core by observing the simulation
waveforms. The test bench can be used with all simulation outputs from behavioral RTL
through post-implementation timing.

In the example design, the simulation sources file set includes the test bench. To run the
test bench, select the Run Simulation option in the Vivado® Flow Navigator. When the
simulation is open, enter the run all command to run the simulation to completion.
Output similar to code shown below should be generated if the simulation completes
successfully.

1937.60ns: exdes_tb: Starting testbench


2017.60ns: exdes_tb: Asserting reset for 16 cycles
2097.60ns: exdes_tb: Reset complete
68480.00ns: exdes_tb: SIMULATION PASSED
68480.00ns: exdes_tb: Test Completed Successfully

For more information with running the simulation test bench please see Vivado Design Suite
User Guide: Logic Simulation (UG900) [Ref 8].

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


73
PG085 November 17, 2021 www.xilinx.com
Appendix A

Upgrading
This appendix contains information about migrating a design from ISE® to the Vivado®
Design Suite, and for upgrading to a more recent version of the IP core. For customers
upgrading in the Vivado Design Suite, important details (where applicable) about any port
changes and other impact to user logic are included.

Device Migration
If you are migrating from a 7 series GTX or GTH device to an UltraScale GTH device, the
prefixes of the optional transceiver debug ports for single-lane cores are changed from
“gt0”, “gt1” to “gt”, and the suffix “_in” and “_out” are dropped. For multi-lane cores,
the prefixes of the optional transceiver debug ports gt(n) are aggregated into a single port.
For example: gt0_gtrxreset and gt1_gtrxreset now become gt_gtrxreset [1:0].
This is true for all ports, with the exception of the DRP buses which follow the convention
of gt(n)_drpxyz.

It is important to update your design to use the new transceiver debug port names. For
more information about migration to UltraScale devices, see the UltraScale Architecture
Migration Methodology Guide (UG1026) [Ref 10].

Migrating to the Vivado Design Suite


For information about migrating to the Vivado Design Suite, see the ISE to Vivado Design
Suite Migration Guide (UG911) [Ref 4].

Upgrading in the Vivado Design Suite


This section provides information about any changes to the user logic or port designations
that take place when you upgrade to a more current version of this IP core in the Vivado
Design Suite.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


74
PG085 November 17, 2021 www.xilinx.com
Appendix A: Upgrading

Parameter Changes
There are no parameter changes.

Port Changes
There are no port changes.

Other Changes
There are no other changes.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


75
PG085 November 17, 2021 www.xilinx.com
Appendix B

Debugging
This appendix includes details about resources available on the Xilinx ® Support website and
debugging tools.

TIP: If the IP generation halts with an error, there may be a license issue. See Licensing and Ordering
in Chapter 1 for more details.

Finding Help on Xilinx.com


To help in the design and debug process when using the AXI4-Stream Infrastructure IP
Suite, the Xilinx Support web page contains key resources such as product documentation,
release notes, answer records, information about known issues, and links for obtaining
further product support.

Documentation
This product guide is the main document associated with the AXI4-Stream Infrastructure IP
Suite. This guide, along with documentation related to all products that aid in the design
process, can be found on the Xilinx Support web page or by using the Xilinx Documentation
Navigator.

Download the Xilinx Documentation Navigator from the Design Tools tab on the Downloads
page. For more information about this tool and the features available, open the online help
after installation.

Solution Centers
See the Xilinx Solution Centers for support on devices, software tools, and intellectual
property at all stages of the design cycle. Topics include design assistance, advisories, and
troubleshooting tips.

Answer Records
Answer Records include information about commonly encountered problems, helpful
information on how to resolve these problems, and any known issues with a Xilinx product.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


76
PG085 November 17, 2021 www.xilinx.com
Appendix B: Debugging

Answer Records are created and maintained daily ensuring that users have access to the
most accurate information available.

Answer Records for this core are listed below, and can also be located by using the Search
Support box on the main Xilinx support web page. To maximize your search results, use
proper keywords such as

• Product name
• Tool message(s)
• Summary of the issue encountered

A filter search is available after results are returned to further target the results.

Technical Support
Xilinx provides technical support at the Xilinx Support web page for this LogiCORE™ IP
product when used as described in the product documentation. Xilinx cannot guarantee
timing, functionality, or support if you do any of the following:

• Implement the solution in devices that are not defined in the documentation.
• Customize the solution beyond that allowed in the product documentation.
• Change any section of the design labeled DO NOT MODIFY.

To contact Xilinx Technical Support, navigate to the Xilinx Support web page.

Debug Tools
There are many tools available to address AXI4-Stream Infrastructure IP Suite design issues.
It is important to know which tools are useful for debugging various situations.

Vivado Design Suite Debug Feature


The Vivado® Design Suite debug feature inserts logic analyzer and virtual I/O cores directly
into your design. The debug feature also allows you to set trigger conditions to capture
application and integrated block port signals in hardware. Captured signals can then be
analyzed. This feature in the Vivado IDE is used for logic debugging and validation of a
design running in Xilinx devices.

The Vivado logic analyzer is used with the logic debug IP cores, including:

• ILA 2.0 (and later versions)


• VIO 2.0 (and later versions)

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


77
PG085 November 17, 2021 www.xilinx.com
Appendix B: Debugging

See the Vivado Design Suite User Guide: Programming and Debugging (UG908) [Ref 6].

Reference Boards
Various Xilinx development boards support AXI4-Stream Infrastructure IP Suite. These
boards can be used to prototype designs and establish that the core can communicate with
the system.

• 7 series evaluation boards

° KC705

° KC724

Hardware Debug
Hardware issues can range from link bring-up to problems seen after hours of testing. This
section provides debug steps for common issues. The Vivado debug feature is a valuable
resource to use in hardware debug. The signal names mentioned in the following individual
sections can be probed using the ChipScope tool for debugging the specific problems.

General Checks
Ensure that all the timing constraints for the core were properly incorporated from the
example design and that all constraints were met during implementation.

• Does it work in post-place and route timing simulation? If problems are seen in
hardware but not in timing simulation, this could indicate a PCB issue. Ensure that all
clock sources are active and clean.
• If using MMCMs in the design, ensure that all MMCMs have obtained lock by
monitoring the LOCKED port.
• If your outputs go to 0, check your licensing.

Interface Debug
AXI4-Stream Interfaces
If data is not being transmitted or received, check the following conditions:

• If transmit <interface_name>_tready is stuck Low following the


<interface_name>_tvalid input being asserted, the core cannot send data.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


78
PG085 November 17, 2021 www.xilinx.com
Appendix B: Debugging

• If the receive <interface_name>_tvalid is stuck Low, the core is not receiving


data.
• Check that the ACLK inputs are connected and toggling.
• Check core configuration.
• Add appropriate core specific checks.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


79
PG085 November 17, 2021 www.xilinx.com
Appendix C

Additional Resources and Legal Notices

Xilinx Resources
For support resources such as Answers, Documentation, Downloads, and Forums, see Xilinx
Support.

Documentation Navigator and Design Hubs


Xilinx ® Documentation Navigator provides access to Xilinx documents, videos, and support
resources, which you can filter and search to find information. To open the Xilinx
Documentation Navigator (DocNav):

• From the Vivado ® IDE, select Help > Documentation and Tutorials.
• On Windows, select Start > All Programs > Xilinx Design Tools > DocNav.
• At the Linux command prompt, enter docnav.

Xilinx Design Hubs provide links to documentation organized by design tasks and other
topics, which you can use to learn key concepts and address frequently asked questions. To
access the Design Hubs:

• In the Xilinx Documentation Navigator, click the Design Hubs View tab.
• On the Xilinx website, see the Design Hubs page.
Note: For more information on Documentation Navigator, see the Documentation Navigator page
on the Xilinx website.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


80
PG085 November 17, 2021 www.xilinx.com
Appendix C: Additional Resources and Legal Notices

References
These documents provide supplemental material useful with this product guide:

1. Arm AMBA AXI4-Stream Protocol Specification


2. PCI-SIG Documentation (www.pcisig.com/specifications)

° PCI Express Base Specification 1.1

° PCI Express Card Electromechanical (CEM) Specification 1.1


3. Vivado AXI Reference Guide (UG1037)
4. ISE to Vivado Design Suite Migration Guide (UG911)
5. Vivado Design Suite User Guide: Designing with IP (UG896)
6. Vivado Design Suite User Guide: Programming and Debugging (UG908)
7. Vivado Design Suite User Guide: Getting Started (UG910)
8. Vivado Design Suite User Guide: Logic Simulation (UG900)
9. Vivado Design Suite User Guide: Designing IP Subsystems Using IP Integrator (UG994)
10. UltraScale Architecture Migration Methodology Guide (UG1026)
11. UltraScale Architecture Libraries Guide (UG974)

Revision History
The following table shows the revision history for this document.

Date Version Revision


11/17/2021 3.0 • Updated Enable URAM Memory Type.
12/05/2018 3.0 • Updated for AXI4-Stream Data FIFO core changes:
• Added Xilinx Parameterized Macro support.
• Updated FIFO depth support, and support for multiple clocks, ECC, various
resource utilization types, and optional FIFO flags.
• Removed FIFO Generator support.
04/04/2018 2.2 Updated AXI4-Stream Switch Arbiter Algorithm, AXI4-Steam Slice parameters,
and core constraints.
11/29/2017 2.2 Updated AXI4-Stream Register Slice to add user-selectable performance vs.
area tradeoff and optional pipelining to cross SLRs in SSI devices.
04/05/2017 2.2 Updated for Enable FIFO Count Ports section.
04/06/2016 2.2 Updated AXI4-Stream Subset Converter section.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


81
PG085 November 17, 2021 www.xilinx.com
Appendix C: Additional Resources and Legal Notices

Date Version Revision


04/01/2015 2.2 Updated AXI4-Stream Switch section.
10/01/2014 2.2 Updated Table 2-6. Added Device Migration section.
12/18/2013 2.2 Added UltraScale Architecture support.
10/02/2013 2.2 Added Example Design and Test Bench chapters.
03/20/2013 2.1 Updated Tables 2-3 and 2-5 and revised Debugging appendix.
12/18/2012 2.0 Updated for characterization numbers and added Debugging appendix.
10/16/2012 1.0 Initial Xilinx release.

Please Read: Important Legal Notices


The information disclosed to you hereunder (the "Materials") is provided solely for the selection and use of Xilinx products. To the
maximum extent permitted by applicable law: (1) Materials are made available "AS IS" and with all faults, Xilinx hereby DISCLAIMS
ALL WARRANTIES AND CONDITIONS, EXPRESS, IMPLIED, OR STATUTORY, INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY, NON-INFRINGEMENT, OR FITNESS FOR ANY PARTICULAR PURPOSE; and (2) Xilinx shall not be liable (whether
in contract or tort, including negligence, or under any other theory of liability) for any loss or damage of any kind or nature related
to, arising under, or in connection with, the Materials (including your use of the Materials), including for any direct, indirect,
special, incidental, or consequential loss or damage (including loss of data, profits, goodwill, or any type of loss or damage
suffered as a result of any action brought by a third party) even if such damage or loss was reasonably foreseeable or Xilinx had
been advised of the possibility of the same. Xilinx assumes no obligation to correct any errors contained in the Materials or to
notify you of updates to the Materials or to product specifications. You may not reproduce, modify, distribute, or publicly display
the Materials without prior written consent. Certain products are subject to the terms and conditions of Xilinx's limited warranty,
please refer to Xilinx's Terms of Sale which can be viewed at https://www.xilinx.com/legal.htm#tos; IP cores may be subject to
warranty and support terms contained in a license issued to you by Xilinx. Xilinx products are not designed or intended to be
fail-safe or for use in any application requiring fail-safe performance; you assume sole risk and liability for use of Xilinx products
in such critical applications, please refer to Xilinx's Terms of Sale which can be viewed at https://www.xilinx.com/legal.htm#tos.
AUTOMOTIVE APPLICATIONS DISCLAIMER
AUTOMOTIVE PRODUCTS (IDENTIFIED AS “XA” IN THE PART NUMBER) ARE NOT WARRANTED FOR USE IN THE DEPLOYMENT OF
AIRBAGS OR FOR USE IN APPLICATIONS THAT AFFECT CONTROL OF A VEHICLE (“SAFETY APPLICATION”) UNLESS THERE IS A
SAFETY CONCEPT OR REDUNDANCY FEATURE CONSISTENT WITH THE ISO 26262 AUTOMOTIVE SAFETY STANDARD (“SAFETY
DESIGN”). CUSTOMER SHALL, PRIOR TO USING OR DISTRIBUTING ANY SYSTEMS THAT INCORPORATE PRODUCTS, THOROUGHLY
TEST SUCH SYSTEMS FOR SAFETY PURPOSES. USE OF PRODUCTS IN A SAFETY APPLICATION WITHOUT A SAFETY DESIGN IS FULLY
AT THE RISK OF CUSTOMER, SUBJECT ONLY TO APPLICABLE LAWS AND REGULATIONS GOVERNING LIMITATIONS ON PRODUCT
LIABILITY.
© Copyright 2012-2021 Xilinx, Inc. Xilinx, the Xilinx logo, Alveo, Artix, Kintex, Kria, Spartan, Versal, Vitis, Virtex, Vivado, Zynq, and
other designated brands included herein are trademarks of Xilinx in the United States and other countries. AMBA, AMBA Designer,
ARM, ARM1176JZ-S, CoreSight, Cortex, and PrimeCell are trademarks of ARM in the EU and other countries. All other trademarks
are the property of their respective owners.

AXI4-Stream Infrastructure IP Suite v3.0 Send Feedback


82
PG085 November 17, 2021 www.xilinx.com

You might also like