AMD Zynq™ UltraScale+™ XA MPSoCs
Enabling next-generation ADAS and AD Systems and ideally suited for various automotive customer platforms by delivering the right performance/watt while integrating functional safety and security features.
Overview
Many new functions of next-generation vehicles, like advanced driver assistance, require a greater need for the ability to sense and perceive what’s happening around the vehicle in all driving conditions. The AMD Automotive (XA) portfolio is currently deployed in an array of production-ready advanced sensing technologies including cameras, radar, and LiDAR.
Market Projections for ADAS
325M
325M global unit demand for automotive cameras by 2026
(IHS)
130M
130M forward-camera units forecasted by 2026
(TechInsights)
241M
241M short-, mid-, and long-range radar units produced by 2026
(TechInsights)
Solutions
Forward cameras are a critical element of ADAS and automated driving (AD) because they provide the advanced sensing capabilities required for vehicle perception, control, and path planning. Image processing within forward camera, whether based on computer vision (CV) or neural networks (NN), can all be processed by our scalable AMD Zynq™ UltraScale+™ XA MPSoC platform. Key product features include:
- Custom, differentiated hardware accelerators tightly-coupled with application software in a heterogeneous processing environment
- Comprehensive functional safety to meet OEM requirements
- Scalable Deep Learning Processing Unit (DPU) for advanced AI perception with ultra-low latency processing
Application Processor
- 32-bit
- Dual/Quad-Core
Memory Subsystem
- High Bandwidth
- Low Latency
Video Codec
- 8K4K (15fps)
- 4K2K (60fps)
Real-Time Processors
- 32-bit Dual-Core
Graphics Processor
- ARM Mali-400MP2
Platform & Power Management
- Granular Power Control
- Functional Safety
Fabric Acceleration
- Customizable Engines
- High-Speed Connectivity
High-Speed Peripherals
- Key Interfaces
Configuration & Security Unit
- Anti-Tamper & Trust
- Industry Standards
Automated Driving Domain Controller with Versal AI Edge Series Gen 2
Versal™ AI Edge Series Gen 2 adaptive SoCs include an enhanced AI Engine, offering higher efficiency and up to 3X performance per watt compared to the previous generation.1 The heterogeneous architecture of the AI Edge Series Gen 2 devices is ideal for multi-sensor vision perception systems requiring sensor fusion and real-time decision making.
- Offering a comprehensive processing platform contained within a single adaptive SoC device
- Providing performance leadership with extensive I/O capability, a virtualized software stack, and robust ecosystem
- Includes next-generation AI Engine technology offering up to 3X performance per watt improvement1
3D Surround View Camera
Automatic park assist (APA) and automated valet parking (AVP) features require surround view capabilities for low-speed driving situations. The XA portfolio powers multiple generations of surround view systems, and continues to address the industry’s integration of sensor configurations through our scalable portfolio. AMD – along with our ecosystem partners – creates silicon and IP solutions such as hardware-based video warping acceleration which enables efficient, low latency distortion correction for fisheye lenses, perspective projection, and the simultaneous stitching of multiple video frames.
Radar
Next-generation imaging radar sensors target the integration of short/mid/long range performance into a single sensor module to support a variety of ADAS and AD features. To support this radar integration, the AMD Zynq UltraScale+ XA MPSoC platform is highly scalable and offers functional integration features including sensor fusion and AI perception. Additional advantages of the platform include:
- Digital Signal Processing (DSP) slices which enable complex signal processing of radar data
- Parallel processing within the Programmable Logic that delivers optimal performance and enables the fully independent, yet simultaneous processing pipelines
- Adaptable hardware architecture that enables end product differentiation with complete IP ownership, enabling system designers to stay aligned with evolving ADAS/AD feature requirements
LiDAR
AMD devices bring tremendous value to LiDAR sensor architectures because of our capability to enhance advanced signal processing, point cloud pre-processing, and point cloud machine learning acceleration. Our devices are used in an array of LiDAR architectures – explore our LiDAR customer ecosystem below for more details.
Also, download the new white paper from ABI Research, which discusses the future of the automotive industry from ADAS and active safety to supervised and unsupervised autonomous driving.
LiDAR Partners
Ouster
Headquarters: USA
Type of LiDAR: Scanning, Solid-state, Multi-beam flash
AMD Product Family: Zynq 7000
Differentiators:
- Ouster uses an all-semiconductor design, dramatically simplifying the sensor’s architecture to make it cheaper, smaller, and more durable - without sacrificing performance
- Can capture ambient light, allowing it to capture 2D images (like a camera) using only the LiDAR sensor
Innovusion
Headquarters: USA
Type of LiDAR: Scanning, or Hybrid Solid-state
AMD Product Family: Zynq 7000
Differentiators:
- Solid hybrid design to achieve balance of visibility, Field of View (FoV), and frame rate, while achieving automotive grade reliability
- Strong believer in sensor fusion to achieve L4 and beyond
PointCloud
Headquarters: USA
Type of LiDAR: FMCW 3D imaging, Solid-state, Scanning
AMD Product Family: Zynq Ultrascale+ RFSoC
Differentiators:
- Silicon photonics chip level integrated optical engine built from the ground up for 3D imaging, complemented by FMCW ranging
- Enables high performance at low cost, power, and form factor
Phantom Intelligence
Headquarters: Canada
Type of LiDAR: Solid-state flash
AMD Product Family: Zynq Ultrascale+ MPSoC
Differentiators:
- Phantom Intelligence digital processing technologies greatly improve the detection, reliability, and range of conventional analog LiDAR for more confidence in threat assessment
- Unique IP licensing model for automotive LiDAR and sensor fusion
Baraja
Headquarters: Australia
Type of LiDAR: Spectrum-Scan™
AMD Product Family: Zynq 7000
Differentiators:
- Spectrum-Scan is a software-defined LiDAR, giving the user full programmatic control over how the vehicle sees by changing the LiDAR resolution, FoV and focus of attention on-the-fly
- Spectrum-Scan LiDAR provides inherent protection against interference through the unique use of multiple light wavelengths. This creates multiple filters for rejecting interference from sunlight or other LiDAR
RoboSense
Headquarters: China
Type of LiDAR: Mechanical, Solid-state MEMS & Flash
AMD Product Family: Zynq Ultrascale+ MPSoC
Differentiators:
- High resolution Solid-state LiDAR for automotive usages
- Smart LiDAR system with the LiDAR HW and AI perception algorithm SW
Hesai
Headquarters: China
Type of LiDAR: Solid-state Scanning
AMD Product Family: Artix™ 7 and Zynq 7000
Differentiators:
- Successfully achieved full interference rejection, and the technology has become a standard production feature since August 2018
- Of the 62 companies in California with self-driving car test permits, over 1/3 are Hesai’s customers
Zvision
Headquarters: China
Type of LiDAR: Solid-state
AMD Product Family: Zynq Ultrascale+ MPSoC
Differentiators:
- Designed for real, autonomous-grade reliability
- The most competitive performance thus far in MEMS LiDAR
Benewake
Headquarters: China
Type of LiDAR: Solid-state, LiDAR Distance Sensor, Long Range Solid-state
AMD Product Family: Spartan™ 6
SureStar
Headquarters: China
Type of LiDAR: Navigation, Mapping
AMD Product Family: Spartan 6, Kintex™ 7
Tanway
Headquarters: China
Type of LiDAR: Mixed Solid-State LiDAR (TDC + APD)
AMD Product Family: Zynq 7000
Differentiators:
- Hardware level image fusion in LiDAR module, pre-fusion of LiDAR point cloud + Camera image data.
- Laser measurement points correspond to multi-pixel units, and multi-dimensional fusion information data stream is packaged and output.
Blickfeld
Headquarters: Germany
Type of LiDAR: Solid-state, Scanning, MEMS, 905nm
AMD Product Family: Zynq Ultrascale+ MPSoC
Differentiators:
- Proprietary silicon MEMS mirror technology specifically developed for LiDAR applications to provide high resolution, long detection range, and a wide field of view
- Targeted at automotive-grade performance with the cost and size needed for the mass market
Opsys
Headquarters: Israel
Type of LiDAR: Semiconductor-based Solid-state Scanning
AMD Product Family: Zynq Ultrascale+ RFSoC
Differentiators:
- True solid-state with no moving parts; semiconductor-based
- Ultrafast scanning with 1000 FPS raw data
Portfolio
Case Studies
Resources
Get Started
Find devices and boards to fit your hardware needs, and leverage software libraries to develop your own applications
Stay Informed
Sign Up for Automotive Updates
Inquiries
Contact Sales for More Information
Footnotes
- Based on AMD internal performance and power projections for the AIE-ML v2 compute tile architecture in the Versal AI Edge Series Gen 2 using the MX6 data type, compared to performance specifications and AMD Power Design Manager power results for the AIE-ML compute tile architecture featured in the first-generation Versal AI Edge Series using INT8 data type. Assumptions: 2 row, 8 column sub-arrays. Operating conditions: 1 GHz FMAX, 0.7V AIE operating voltage, 100°C junction temperature, typical process, 60% vector load, % activations = 0 < 10%. Actual performance will vary when final products are released in market. Performance projections as of March 2024. (VER-023)
Footnotes
- Based on AMD internal performance and power projections for the AIE-ML v2 compute tile architecture in the Versal AI Edge Series Gen 2 using the MX6 data type, compared to performance specifications and AMD Power Design Manager power results for the AIE-ML compute tile architecture featured in the first-generation Versal AI Edge Series using INT8 data type. Assumptions: 2 row, 8 column sub-arrays. Operating conditions: 1 GHz FMAX, 0.7V AIE operating voltage, 100°C junction temperature, typical process, 60% vector load, % activations = 0 < 10%. Actual performance will vary when final products are released in market. Performance projections as of March 2024. (VER-023)