Overview
Using this Guide
Chapter 1: Introduction to VFS
What Is VFS?
Why Use VFS?
Architecture Overview
Next Steps
Chapter 2: Installation and Environment Setup
Enabling Editor Autocomplete and Docstrings for VFS
Problem
Prerequisites
Solution
Step 1: Source the Vitis environment
Step 2: Create the vfs.pth file
Step 3: Restart your editor
Removing
Troubleshooting
VFS Standalone Installation
Overview
Benefits
Prerequisites
Creating a Standalone Installation
Step 1: Set Up the Environment
Step 2: Run the Script
What the Script Does
Using the Standalone Installation
Output Contents
Portability
Troubleshooting
XILINX_VITIS Not Set
Missing Component Warnings
Supported Python and MATLAB Versions
Related Information
Chapter 3: Quick Start / Hello VFS
Prerequisites
Hello HLS Kernel
Python
MATLAB
3.2 Hello AIE Graph
Python
MATLAB
Next Steps
Chapter 4: The varray Data Model
Supported Data Types at a Glance
varray AXI4-Stream Axis User Guide
Introduction
Key Features
The varray.axis Configuration Class
Constructor Parameters
Conditionally-Present Fields
Optional Fields
Creating Arrays with Axis
Basic Axis Configuration
Axis with Optional Fields
Sideband Field Reference
tlast
tkeep
tstrb
tuser
tid
tdest
Reading and Writing Sideband Fields
Accessing Fields (Python)
Accessing Fields (MATLAB)
Indexing and Slicing
Modifying Axis Configuration with set_axis()
Adding Axis to a Plain Array
Removing Axis from an Array
Replacing Axis with a Different Configuration
Display and Representation
Default Axis (All Sidebands Enabled)
TLAST-only Axis (tkeep and tstrb Disabled)
Axis with Optional Fields
Byte-Level Packing
Relationship to HLS Interfaces
ap_axis
hls::axis
Supported Configurations
Input Axis Validation
What the Validation Checks
Error: port expects axis but varray has none
Error: port has no axis but varray has one
Error: axis field mismatch
AIE Graphs
MATLAB Error IDs
Python Error Types
Examples
Marking Packet Boundaries with tlast
Fixed-Point and Complex Types with Axis
End-to-End HLS Kernel Example
TUSER-Only Streams (No TDATA)
Best Practices
8. Match enable_tkeep, enable_tstrb, and strict to the kernel
Troubleshooting
Python: "The 'axis' parameter must be an instance of varray.axis"
Python: AttributeError for tuser, tid, or tdest
MATLAB: Error Accessing tuser, tid, or tdest
Unexpected tkeep/tstrb default Values
MATLAB: Wide Sideband Values Silently Truncate
TypeError: "Input port N expects AXI-Stream axis sidebands"
TypeError: "Input port N does not expect axis sidebands"
TypeError: "Input port N axis config mismatch"
TypeError: "Input port N axis config mismatch" for enable_tdata
Error When Passing Axis varray to AIE Graph
API Reference
varray.axis Class
Constructor
Methods and Properties
varray.array Methods
Array Construction with Axis
Sideband Attributes
Related Information
MATLAB varray
Setup and Version Check
Supported Data Types
Integer
Floating Point
Complex
Fixed-point (fi, cfi)
Computing Byte Width
Microscaling (mx4, mx6, mx9)
Creating Arrays
Array Properties
Type-specific Properties
Array Operations
Indexing, Assignment, and Slicing
Reshape, Concatenate, Transpose
Arithmetic and Math Operations (Not Supported)
Comparisons
MATLAB Interoperability
Casting to Native MATLAB Types
Casting between varray Types
MATLAB Functions that Work on varray Directly
Fixed-Point, Microscaling, and Half Types
Fixed-point Quantization
Microscaling Blocks
HLS half vs. float16
Troubleshooting
API Reference
Constructors
Properties Common to All Types
Methods common to all types
AXI4-Stream Sideband Helpers
Type-Specific Properties and Methods
Helpers
Version
Related Information
Python varray
Setup and Import
Supported Data Types
Integer
Floating Point
Complex
Fixed-point (fi, cfi)
Computing Byte Width
Microscaling (mx4, mx6, mx9)
Creating Arrays
From lists or tuples: dtype is required
From a NumPy array: dtype is optional
From a Scalar
Array Properties
Array Operations
Indexing and Slicing
Reshape
Arithmetic and Math operations (Not Supported)
astype(): Convert between Types
view(): Reinterpret the Bytes
NumPy Interoperability
Converting to NumPy
Supported NumPy functions
Comparison Operations
Complex Arrays
Fixed-Point, Microscaling, and Half Types
Fixed-point Quantization
Microscaling Blocks
HLS half vs. float16
Troubleshooting
API Reference
varray.array
Type factories
Helpers
Related Information
Chapter 5: VFS APIs
Which Language Should I Choose?
MATLAB API
Setup
Execution Modes: release and debug
Component Lifecycle
vfs.hlsKernel
Constructor
What vfs_build_dir must contain
Examples
Running
Output Shape
Inspecting Ports
Zero-input Kernels (setNumKernelRuns)
vfs.aieGraph
Constructor
What vfs_build_dir must contain
Examples
Running
Port Dump
Combined AIE and HLS Workflows
Utility Functions
API Reference
vfs.hlsKernel
vfs.aieGraph
Troubleshooting
Related Information
Python API
Setup
Component Lifecycle
vfs.hlsKernel
Constructor
What vfs_build_dir must contain
Examples
Running
Output Shape
Inspecting Ports
Zero-input Kernels (setNumKernelRuns)
vfs.aieGraph
Constructor
What vfs_build_dir must contain
Examples
Running
Port Dump
Combined AIE and HLS Workflows
API Reference
vfs.hlsKernel
vfs.aieGraph
vfs.readDumpFile.read_dump_file(path)
shape= Values
Troubleshooting
Related Information
Chapter 6 — Component Interfaces & Behavior
AIE Stream and Buffer Interfaces
Interface Types
Stream Interface
Buffer / Window Interface
Data Processing Behavior
Stream-Based Kernels
Buffer / Window-Based Kernels
When To Use Each Interface
Mixed Inputs: Synchronous and Asynchronous Parameters
Case 1: Synchronous Parameter
Case 2: Asynchronous Parameter
Summary of Cases 1 and 2
Summary
Related Information
HLS Stream and Buffer Interfaces
Interface Types
Stream Interface
Buffer / Array Interface
Data Processing Behavior
Stream-based Kernels
Buffer / Array-based Kernels
Mixed Interface Types (Stream and Scalar)
When to Use Each Interface
Summary
Related pages
Additional Resources and Legal Notices
Support Resources
Revision History
Please Read: Important Legal Notices
AUTOMOTIVE APPLICATIONS DISCLAIMER
Copyright