AMD, in partnership with Optomotive, introduced an industrial smart camera built around the Zynq Ultrascale+ MPSoC-based Kria K26 System-on-Module (SoM). The SMILODON 10G EVO features a 25MP image sensor from Gpixel, supports up to 48 LVDS interfaces, and provides either 10 Gigabit SFP+ or 1 Gigabit RJ45 Ethernet connectivity.

The SMILODON 10G EVO features the Kria K26 SoM similarly to the Axiomtek RSC201 covered yesterday.

• Kria K26 SoM – Quad-core Arm Cortex-A53 MPCore (up to 1.5GHz); Dual-core Arm Cortex-R5F MPCore (up to 600MHz); Mali-400 MP2 (up to 667MHz); 1.36TOPS

The product brief indicates that this embedded camera supports GMAX25xx sensors with resolutions ranging from 5 to 25 Megapixels. These sensors feature a pixel size of 2.5 µm, and support frame rates up to 150 fps (10-bit) or 42 fps (12-bit), depending on configuration.

Kria K26 SoM block diagram
(click image to enlarge)

For processing, the SMILODON 10G EVO has 4GB DDR4 (up to 19.2 GB/s) memory.  Storage includes 64MB of QSPI Flash and up to 16GB of eMMC.

Other peripherals featured on this embedded camera include 1x opto-isolated input, 1x opto-isolated output, 1x 10G SFP+ Ethernet or 1x Gigabit RJ45 LAN port, 4x USB ports and a 10-pin Hirose HR10A connector.

   
SMILODON 10G EVO specs
(click images to enlarge)

Regarding product development, Optomotive claimed, “The reference design can be easily edited with standard Vivado tools. Optomotive’s custom IP cores seamlessly integrate inside the Vivado IDE toolchain. A large portion of the FPGA programmable logic (PL) is free for the programming and development of new algorithms or the implementation of additional IP cores.”

 
ECT-KRIA block diagram (left) and front view (right)
(click images to enlarge)

AMD also presented two additional Kria-based products from ODM partners, in this case an IIoT Edge Gateway from Ectron and an Edge AI Appliance from VVDN Technologies.

Ectron’s ECT-KRIA computers target industrial or vehicle applications. According to the product brief, the peripherals offered are 2x GbE LAN ports, 2x USB 3.0/2.0 ports, GPS, Digital I/O (ECT-KRIA-EC), 4G/LTE and 5G wireless support.

             
VVDN-EAIA specs
(click images to enlarge)

The last product presented is the VVDN-EAIA also based on the AMD Kria K26 SoM. This product was described as “an edge AI appliance optimized for high quality, real-time video analytics for smart retail, smart cities, and industrial use cases.”

This embedded device integrates up to 4x Kinara Ara-1 AI accelerators, 3x USB 3.0 ports, 16GB of eMMC and up to 2TB of optional SSD storage. The VVDN-EAIA is also capable of handling up to 8x RTSP (Real-Time Streaming Protocol) cameras.

VVDN mentioned that this embedded appliance has support for a Petalinux based SDK and Kinara SDK to speed up users’ AI models and video pipelines.

Further information

For more information, refer to the AMD’s product announcement which can be found here.