AMD recently introduced a set of new automotive-qualified FPGAs to the Artix family. The new devices are small and cost-optimized—two factors that are becoming increasingly important in automotive design.
AMD Artix FPGA family.
Advanced driver assistance system (ADAS) features are now standard in virtually all new automobiles. ADAS includes many sensors, processing electronics, feedback systems, and control response electronics. While some parts of ADAS are centralized, much intelligence is being sent to the ADAS peripherals, where size and cost are first-order factors.
The smallest Artix Ultrascale+ XA AU7P comes in a 9 mm x 9 mm package, making it compact enough for peripheral devices such as cameras, sensors, and in-vehicle infotainment (IVI) systems. The part is the smallest 16 nm process FPGA or SoC AMD offers.
Key Features of the Artix Ultrascale XA Family
FPGAs have become a common item in the automotive designer’s tool set. Being field programmable, they can be designed, upgraded, or modified to accommodate bug fixes, security updates, or functional enhancements. The new lower cost and smaller form factor Ultrascale+ parts (datasheet linked) allow this capability to be installed for more applications than was previously possible.
FPGA with columnar resources
AMD designed the Artix Ultrascale XA family for automotive applications from the start. This gives the parts high reliability and long service lives. The parts are AEC-Q100 qualified and certified to ISO26262 ASIL-B. AMD designed them for a 15+ year product lifespan. AMD promises support for the AMD 7 series FPGAs and adaptive SoCs through 2040 and AMD Ultrascale+ FPGAs and adaptive SoCs through 2045.
The family includes multiple high-speed (16 Gb/s) serial transceivers with power optimization for signal integrity. They also provide DSP functionality with fixed and floating-point math for improved local image and video processing. For advanced security and anti-tamper functionality, the FPGAs include RSA-2048 authentication, NIST-certified AES-CGM decryption, and DPA countermeasures.
|
XA AU7P |
XA AU10P |
XA AU15P |
|
|
System Logic Cells (K) |
81 | 96 | 170 |
|
CLB LUTs (K) |
37 | 44 | 77 |
|
Memory (Mb) |
3.8 | 3.5 | 5.1 |
|
DSP Slices |
216 | 400 | 576 |
|
Maximum I/O Pins |
248 | 228 | 228 |
ADAS and IVI are highly data-intensive systems, and data must move in and out of the new parts quickly. The family’s architecture can transfer data at up to 58.0 Gb/s to allow 25G+ backplane designs. Power optimization delivers improved power per bit performance over prior part versions. The chip supports 8.0GT/s (Gen3) and 16.0GT/s (Gen4) in its integrated PCIe interfaces and 100G Ethernet and 150G Interlaken.
Applications in the Automotive Edge
Automotive computing systems have become highly complex networks with high-powered central processing and control systems connected to a broad array of edge sensors and actuators. Rapid response for non-critical systems and real-time response for critical systems is an absolute requirement for modern IVI and ADAS. The Artix Ultrascale XA family is designed to support this need by allowing initial processing in the sensor. This reduces workload in the central ADAS processing systems, allowing for real-time response even while increasing the number of sensor inputs.
The size of the AMD Artix Ultrascale FPGA compared with a keyfob.
Cameras for ADAS vision systems are one of the primary target applications for the new chips. The small size, low cost, and DSP functionality will allow automobile manufacturers and their tier-1 suppliers to create smart cameras that can perform much of the initial image processing in-camera.
The new Artix chip complements the AMD automotive line, which includes products in the Spartan 7, Zynq 7000, and Zynq UltraScale+ product lines.
All images used courtesy of AMD.
