Important Note: More Flexible AMD Vivado™ Licensing in 2026.1

Starting with the 2026.1 release, AMD Vivado™ Design Suite introduces new tiered licensing options, allowing customers to pay only for the device families and features they need, while Vivado Enterprise remains unchanged.

AMD Vitis™ Unified Software Platform Licensing Structure

Vitis Embedded Software

No license required for standard embedded software development.

Vitis HLS

No license required for C synthesis/simulation; compilation of generated RTL requires a valid Vivado Design Suite license.

AI Engine Development for High-Performance DSP (AIE / AIE-ML / AIE-ML v2)

AI Engine tools license (free) & Vivado Enterprise (perpetual) or Vivado Pro (subscription) license 

Vitis System Design Flows

Hardware linking and implementation requires a valid Vivado Design Suite license.
AI Engine-based devices require Vivado Enterprise (perpetual) or Vivado Pro (subscription) license. 

Vitis Model Composer

Requires a Vitis Model Composer license and valid Vivado Design Suite license.
These updates provide greater flexibility across embedded, AI Engine, and system-level development workflows.

The Vitis Software Platform Development Environment

Design and Simulation Flows

Adam Taylor Presents: Step-by-Step System Design with Vitis Unified Platform

Learn to create an embedded system solution using the Vitis Unified heterogeneous system flow.

Vitis Embedded Software Development Flow

(Traditionally called Embedded SDK for previous FPGA families)

Export hardware from Vivado as a platform file
Arrow
Develop application code
 
Arrow
Debug and generate boot image

Designers who are developing C/C++ code for the Arm® embedded processor subsystem in AMD adaptive SoCs will typically use this flow.

  • Hardware engineers design programmable logic and export the hardware as a Xilinx Support Archive (XSA) file using AMD Vivado™ Design Suite. 
  • Software engineers incorporate this hardware design information in their target platform and use the Vitis Embedded software to develop their application code.

Developers can perform all system-level verification within the Vitis Embedded software and generate boot images to launch the application. 

Learn more in the Vitis Tools for Embedded Software Development section in UG1400 >

Vitis System Design Flow

(Hardware and Software)

Vitis System Design Flow Chart

System designers who are integrating both the software and hardware portions of their design in AMD adaptive SoCs will typically use this flow.

This flow is used to develop heterogeneous embedded system designs comprising of software applications running on Arm® embedded processors and compute kernels running on programmable logic (PL) and/or Versal™ AI Engine arrays.

This flow comprises:

  • A software host application written in C/C++ and typically run on the embedded Arm processor subsystem. It uses the native API implemented by the AMD Vitis Runtime Library to interact with hardware kernels within the AMD device. 
  • Hardware kernels that can be generated from C++ using the AMD Vitis™ HLS tool or described directly in RTL using AMD Vivado™ Design Suite.

Learn more in the Vitis Tools for Heterogeneous System Design section in UG1393 >

AMD Alveo™ Data Center accelerator cards employ the same system design flow—the software program runs on an x86 host, and the kernels run in the FPGA on a PCIe®-attached acceleration card.

Learn more in Vitis tools for Data Center Acceleration section in UG1393 >

Vitis Heterogeneous Simulation Flow

Simulate in Your Preferred Tools
 • Use existing MATLAB®, Python™, C++, or HDL testbenches
 • Avoid rewriting testbenches or learning new workflows
 • Accelerate algorithm-to-hardware iteration

Unified AI Engine + PL Simulation (Vitis Subsystem)
 • Simulate AI Engine and PL together
 • Replace fragmented flows with one consistent methodology
 • Detect integration issues early

AMD Vitis Hardware in the Loop (Vitis HIL) Validation
 • Shorten system-level validation time
 • Stream real I/O through silicon for faster debugging
 • Verify end-to-end throughput before final hardware

Tools and Libraries

What’s New in AMD Vitis™ Unified Software Platform 2026.2

Enhanced Design Flow with AMD Versal™ AI Engines
  • New Cholesky solver and resource-efficient packet-switching DSP IP
  • ML-based DSP configuration helper (Early Access)
  • Production FP8 support and new floating-point accumulator for AIE-MLv2
  • Per-kernel compiler tuning, independent partitions, and runtime partition switching
  • Rotating-margin API for reduced memory use on Versal AI Edge Series Gen 2
Easier Verification of Versal Adaptive SoC Designs
  • Encrypted-source x86sim and Analyzer code coverage visualization (Early Access)
  • Expanded Vitis™ Functional Simulation: AXI side channels, packet routing, AP float, coefficient updates, and output-only graphs
  • Versal RF LDPC, channelizer, and resampler models (Early Access); new spectrum analyzer tutorial
  • HIL automatic server startup and production VRK160/VEK385 support
Enhancements to Vitis Model Composer for AI Engine for DSP Designs
  • Simplified SSR connectivity with automatic lane replication and reconnection
  • New QRD, SVD, and Cholesky solver blocks
  • Enhanced FIR, DSPFP32, INT8 dot-product, and vector xFFT support
  • Faster Versal RF FFT simulation using SystemC/TLM models (Early Access)
  • HLS Library deprecation in 2027.1; migration to HLS Kernel block
Updates to Vitis IDE for Embedded Development
  • AMD Ross™ AI assistant with expanded model, agent, context, and documentation-search support
  • VS Code target debugging and improved Vitis variable inspection/watchpoints
  • Integrated Python™ console and enhanced build/platform query APIs
  • Unified DFX packaging (Early Access; hardware-only validation)
  • Persistent, concurrent QEMU sessions and attach-debug support
  • Expanded hardware support for Zephyr , device tree generation, and debugging
  • Updated embedded, data-center, and AWS F2 examples

Resources