AMD XILINX VERSAL

VERSAL

New Class of Adaptive SoCs

Versal® ACAP integrates capabilities that far exceed those of conventional CPUs, GPUs, and FPGAs, enabling adaptive, domain specific architectures to accelerate whole applications

An Architecture Ahead

Surpassing the limits of Moore’s Law and traditional FPGA architectures, Versal ACAP uniquely delivers unprecedented hard IP integration and SW programmable silicon infrastructure

System-Level Value

Key innovations deliver unparalleled application- and system-level value for cloud, network, and edge applications at the forefront of their markets


Featured White paper

This white paper provides both a qualitative and quantitative analysis of Versal ACAP system-level capabilities for a host of markets ranging from cloud to wired networking and 5G wireless infrastructure.


Learn how the Versal architecture delivers best-in-class performance/watt leadership over competing 10nm FPGA architectures in end-applications such as AI compute accelerator, 5G Massive MIMO, network accelerator, smart SSDs, and multi-terabit SmartPHY—supported with data that can be validated with public tools.

 

alt-text

System-Level Value for Five Real-World Applications

AI Inference

Solution Brief >

Network

Acceleration

Solution Brief >

Storage

Acceleration

Solution Brief >

SmartPHY for DCI

and Transport

Solution Brief >

5G

Beamforming

Solution Brief >


AN ARCHITECTURE AHEAD

SW Programmable Silicon Infrastructure

ACAPs have been built from the ground-up to be natively software programmable without needing to know RTL (register-transfer level) languages or FPGA architectures. The platform is readily accessible to software developers and data scientists using their preferred tool flows, while providing an accelerated development path for hardware designers.


Key to the platform’s ease of programming is the integrated shell—consisting of a platform management controller, hardened host interface (integrated PCIe® w/DMA), memory controllers, and network interfaces interconnected with a multi-terabit, memory mapped programmable network on chip (NoC) for massive data flow and instant connectivity to compute and network infrastructure. The integrated shell delivers pre-engineered timing closure and logic resource savings so developers can focus on their key differentiation.


Unprecedented Integration of Hardened, Domain-Specific IP

 

 

 

 

 

 

 

 

With 3X more hardened IP than competing 10nm FPGAs, Versal ACAPs feature a common hardened infrastructure across all devices for ease of programming and code portability, as well as domain-specific hardened IP for diverse markets.

 

The balance of adaptable hardware and hardened cores delivers performance/watt leadership for leading applications in cloud, network, and edge, all while retaining hardware flexibility to adapt to changing requirements and market dynamics.


TRY YOURSELF

Run Benchmarks of Versal ACAPs vs. Competing FPGAs as Cited in White Paper


Fabric Performance Benchmarks

Register to access OpenCores designs used for fabric benchmarks referenced in the white paper. Requires Vivado ML 2021.1 and Quartus tools.

 

Register to Access Design Suite >

System-Level Power Benchmarks

Access Xilinx power estimator (XPE) scripts to run end-application power comparisons referenced in the white paper using XPE 2021.2 and Quartus power & calculator (PTC) 2021.2.

 

Register to access XPE  Scripts >


ADDITIONAL RESOURCES

Versal AI Edge Seris
White Paper

Beamforming on AI Engine
 

Versal Premium Seris
White Paper

AI Engine White Paper
 


GET STARTED WITH THE PORTFOLIO

Prime Series

The foundational Versal ACAP series with broad applicability across markets, Versal Prime series is in full volume production. Developers can download tools or purchase evaluation kits to get started now.

View Prime Series >

AI Core Series

Delivering breakthrough AI inference and wireless acceleration, the Versal AI Core series is now in full volume production. Developers can download tools or purchase evaluation kits to get started now.

View AI Core Series >

 

Premium Series

Featuring unprecedented integration of networked cores for the most challenging compute and networking application, the Versal Premium series is sampling now. Contact your local sales representative to apply for the Early Access program or visit the Contact Sales page.

View Premium Series >

 

AI Edge Series

Delivering 4X AI performance/watt vs. GPUs for power-and thermally-constrained edge applications, the Versal AI Edge series is now in Early Access. Contact your local sales representative to apply for the Early Access program or visit the Contact Sales page.

View Edge Series >

 

HBM Series

Delivering hyper integration of fast memory, secure data, and adaptive compute for compute intensive, high bandwidth applications, the Versal HBM series is now in Early Access. Contact your local sales representative to apply for the Early Access program or visit the Contact Sales page.

View HBM Series >

 

Versal AI RF Series Coming Soon