Versal AI Core Series
Versal AI Core series delivers breakthrough AI inference and wireless acceleration with integrated AI engines that deliver over 100X greater compute performance than today’s server-class CPUs.
Portfolio's Highest Compute for Maximum AI and Workload Acceleration
Scalar Engines
Three scalar processor types are available in Versal™ adaptive SoCs to support diverse application needs. The application processing unit is ideal for complex applications supported by an OS, and the real-time processing unit is ideal for applications needing low latency, determinism, and real-time control. A separate platform management controller manages system boot, security, and debug.
Adaptable Engines
Programmable logic enables the development of custom computational blocks for ever-changing algorithms. The rearchitected logic in Versal adaptive SoCs provides 4X greater density per CLB, reducing the need for costly global routing. Loaded with a wide variety of memory elements and tightly coupled with programmable I/O, the adaptable engines allow users to create powerful accelerators for any application.
Intelligent Engines
AI Engines provide up to 5X higher compute density for vector-based algorithms. Optimized for real-time DSP and AI/ML computation, AI Engines provide deterministic performance.
Enhanced DSP engines provide support for new operations and data types, including single and half-precision floating point and complex 18x18 operations.
Next-Generation I/O
The Versal AI Core series combines PCIe® Gen4/Gen5 compliance, CCIX support, high performance GPIO, and multirate Ethernet MACs supporting various ethernet configurations to maximize connectivity and flexibility. Additionally, Versal AI Core series devices feature rearchitected low latency 32.75Gb/s transceivers.
Video Decoder Unit
Power-optimized hardened video decoder units (VDUs) comprise multiple video decoder engines (VDEs) supporting H.264 and H.265 from a single 4Kp60 stream to up to thirty-two 720p15 streams per engine. VDUs are ideal for many image processing applications where multiple video inputs feed a central hub that is both decoding and executing advanced ML algorithms.
Programmable Network on Chip
The programmable network on chip (NoC) provides an optimized multi-terabit interconnect between the different compute engines and integrated IP blocks present in the Versal adaptive SoC architecture, simplifying timing closure and saving logic resources. The NoC compiler provides a streamlined programming experience while allowing users to manage latency and QoS, ensuring that critical data paths are prioritized.
See how AMD FPGAs and adaptive SoCs enable AI Inference acceleration
AI InferenceCheck out the Versal AI Core series evaluation kit
Contact AMD sales representatives to join the early access program
Contact SalesVersal AI Core series devices are optimized for compute-intensive applications, specifically digital signal processing, artificial intelligence, and machine learning
5G Radio & Beamforming
Beamforming results in enormous signal processing requirements that need to be processed in real time. While adaptable engines are handling status control and data formatting, AI Engines and DSP Engines are processing fixed & floating point signal processing workloads for high performance parallel processing.
Data Center Compute
As convolutional neural networks continue to evolve, the challenge to keep up with increasing computational density requirements can only be addressed by the Versal™ AI Core series devices, which have AI Engines optimized to efficiently deliver computational density, both cost effectively and power efficiently.
Video Processing for Smart Cities
Versal AI Core series devices enable the plethora of new video processing applications being deployed at the edge for Smart Cities, with AI Engines driving real-time license plate, facial recognition, or object classification, video decoder units for video transcoding, and adaptable engines for video scaling, compression, and customization.
Medical Image Processing
Versal AI Core series accelerates parallel beamforming and real-time image processing to create higher quality images and machine-based image analysis to enable physicians and radiologists to make faster and more accurate diagnoses.
Radar Processing
Merging powerful vector-based DSP engines with AI engines in a small form factor enables advanced radars, such as active electronically scanned arrays. AI Core series devices provide terabits per second of antenna bandwidth in a single package.
Wireless Test Equipment
Real-time DSP is used extensively in wireless communications test equipment. Versal AI Engine architecture is well-suited to handle all types of protocol implementations, including 5G from the digital front end to beamforming and baseband.
Versal™ AI Core Series Features Overview
Scalar Engines Features
| VC1502 | VC1702 | VC1802 | VC1902 | VC2602 | VC2802 | |
|---|---|---|---|---|---|---|
| Application Processing Unit | Dual-core Arm® Cortex®-A72, 48KB/32KB L1 Cache w/ parity & ECC; 1MB L2 Cache w/ ECC | |||||
| Real-time Processing Unit | Dual-core Arm Cortex-R5F, 32KB/32KB L1 Cache, and 256KB TCM w/ECC | |||||
| Memory | 256KB On-Chip Memory w/ECC | |||||
| Connectivity | Ethernet (x2); USB 2.0 (x1); UART (x2); SPI (x2); I2C (x2); CAN-FD (x2) | |||||
Intelligent Engines Features
| VC1502 | VC1702 | VC1802 | VC1902 | VC2602 | VC2802 | |
|---|---|---|---|---|---|---|
| AI Engines | 198 | 304 | 300 | 400 | 0 | 0 |
| AI Engines-ML | 0 | 0 | 0 | 0 | 152 | 304 |
| DSP Engines | 1,032 | 1,312 | 1,600 | 1,968 | 984 | 1,312 |
Adaptable Engines Features
| VC1502 | VC1702 | VC1802 | VC1902 | VC2602 | VC2802 | |
|---|---|---|---|---|---|---|
| System Logic Cells (K) | 815 | 981 | 1,586 | 1,968 | 820 | 1,139 |
| LUTs | 372,352 | 448,512 | 725,000 | 899,840 | 375,000 | 520,704 |
Platform Features
| VC1502 | VC1702 | VC1802 | VC1902 | VC2602 | VC2802 | |
|---|---|---|---|---|---|---|
| NoC Master / NoC Slave Ports | 21 | 21 |
28 | 28 | 21 | 21 |
| DDR Memory Controllers | 3 | 3 |
4 | 4 | 3 | 3 |
| CCIX & PCIe® w/ DMA (CPM) | 1 x Gen4x16, CCIX |
1 x Gen4x16, CCIX |
1 x Gen4x16, CCIX |
1 x Gen4x16, CCIX |
2 x Gen5x8, CCIX |
2 x Gen5x8, CCIX |
| PCI Express® | 4 x Gen4x8 | 4 x Gen4x8 | 4 x Gen4x8 | 4 x Gen4x8 | 4 x Gen5x4 | 4 x Gen5x4 |
| 100G Multirate Ethernet MAC | 3 | 4 | 4 | 4 | 2 | 2 |
| Video Decoder Engines (VDEs) | - | - | - | - | 2 | 4 |
| GTY Transceivers | 32 | 44 | 44 | 44 | 0 | 0 |
| GTYP Transceivers | 0 | 0 | 0 | 0 | 321 | 321 |
1. 16 GTYP transceivers are dedicated to CPM5 for PCI Express use.
Documentation
Versal Design Guidance and Documentation
AMD provides a breadth of documentation, resources, and methodologies to accelerate your development on the Versal™ architecture. If you’re not sure where to begin with Versal adaptive SoCs, the Design Flow Assistant is an interactive guide to help you create a development strategy, while the Design Process Hubs are a visual and streamlined reference to all Versal documentation by design process.
Integrated Software and Hardware Platform for All Developers
With an inherently software programmable silicon infrastructure, the Versal™ adaptive SoC is designed from the ground up to be software-centric. The enhanced Vivado™ ML Editions introduces a new system design methodology and development environments such as traffic analyzer, NoC compiler, data flow modeling, and more. A high-speed, unified, cohesive debug environment accelerates debug and trace across Scalar, Adaptable, and Intelligent engines.
Download Vivado ML Editions >
The Vitis™ unified software platform provides comprehensive core development kits, libraries that use hardware-acceleration technology. The platform provides an efficient, convenient, and unified software environment from the cloud to the edge. As a proud member of the open source community, the Vitis unified software platform is entirely free and open source.
Download Vitis Unified Software Platform >
AMD Vitis Model Composer is a model-based design tool that enables rapid design exploration within the Simulink® and MATLAB® environments. It facilitates AI Engine ADF graph development and testing at the system level, allowing the user to incorporate RTL and HLS blocks with AI Engine kernels and/or graphs in the same simulation. To learn how to use Versal AI Engines with Vitis Model Composer, visit the AI Engine resource page.
Versal Adaptive SoC Evaluation Kits
Be the first to experience the Versal adaptive SoC architecture, built from the ground-up to be natively software programmable. Through a host of tools, software, libraries, IP, middleware, and frameworks, adaptive SoCs enable dynamically customizable accelerated computing solutions through industry-standard design flows. The Versal AI Core series VCK190 evaluation kit has everything you need to jump-start your designs.
Learn more about the Versal AI Core series VCK190 evaluation kit >
Training Courses
AMD training and learning resources provide the practical skills and fundamental knowledge you need to be fully productive in your next development project.
Designing with the Versal ACAP: Architecture and Methodology
Designing with the Versal ACAP: Network on Chip
Designing with the Versal ACAP: Power and Board Design
Designing with Versal AI Engine: Architecture and Design Flow - 1
Designing with Versal AI Engine: Graph Programming with AI Engine Kernels - 2
Designing with Versal AI Engine: Kernel Programming and Optimization - 3
Versal Design Guidance and Documentation
AMD provides a breadth of documentation, resources, and methodologies to accelerate your development on the Versal™ architecture. If you’re not sure where to begin with Versal adaptive SoCs, the Design Flow Assistant is an interactive guide to help you create a development strategy, while the Design Process Hubs are a visual and streamlined reference to all Versal documentation by design process.
