Dev Kit Weekly: AAEON de next-TGU8 Single Board Computer
September 02, 2022
Measuring just 86 mm x 55 mm, the AAEON de next-TGU8 SBC is the industry’s smallest form factor board hosting an Intel® Core™ i-level processor. And the Tiger Lake UP3 processors’ performance pays off in this small package, optimizing the de next-TGU8’s for tasks like local execution of AI workloads and real-time, deterministic decision-making required to build out the intelligent industrial edge.
The AAEON de next-TGU8 features quad-core 11th generation Intel® Core i3/i5/i7 (Tiger Lake-UP3) processors that clock in as fast as 4.4 GHz to deliver PassMark CPU Mark scores as high as 9980. The efficiency of these processors help the SBC achieve a TDP between 15W and 28W, with a configurable down TDP as low as 12W, which contributes to the overall board’s minimal 55W to 84W power profile. The benefits of that power efficiency continue into deployment, where AAEON SBC can operate in 0°C to 60°C temperature ranges and up to 90% humidity.
Intel’s multicore 11th gen Tiger Lake-UP3 processors provide up to four cores and eight threads that can be applied to the execution of AI models at the industrial edge. They also integrate UHD Graphics or Iris Xe graphics and two VDBOX multi-function codec engines that drive as many as four 4K60 HDR or two 8K60 SDR display outputs and support 1080p30 video streaming.
The UP3 processors also integrate Deep Learning Boost, an AI acceleration technology that speeds up AI inferencing by capitalizing on Intel’s Advanced Vector Instructions (AVX)-512 and Vector Neural Network Instruction (VNNI) instruction sets. It also supports the bfloat16 format to efficiently encode high-intensity, low-precision workloads.
And as if that wasn’t enough to empower industrial AI and machine learning designs, the de next-TGU8 is compatible with tools like the Intel OpenVINO Toolkit that optimizes neural network algorithms for fast inferencing on CPUs, GPUs, and other Intel processing units. OpenVINO works with models developed in frameworks like TensorFlow 2, Caffe, MXNet, and ONNX-compatible environments to streamline the development and deployment of edge AI systems. You can also work with it through the Intel Dev Cloud.
It even comes with an open Model Zoo that provides several pre-trained models and demos that can be downloaded from Github, including examples of time forecasting and object detection that may be particularly interesting to industrial systems engineers working on AI projects. If you’re one of those and are interested in measuring how an OpenVINO-enabled de next-TGU8 fares in your application, check out the benchmarking tool on docs.openvino.ai to see how the board’s supported processor SKUs faired against generic workloads or learn how to create an OpenVINO benchmark for your own AI application.
Despite all that performance in such a small form factor, the de next-TGU8 makes no sacrifices in terms of I/O. Here, it provides an RJ45 port backed by Intel i219LM and Intel i225LM Ethernet controllers, as well as two USB Type-A ports and, of course, an HDMI 1.4b display connector.
If all that isn’t enough, the de next TGU-9 is compatible with FPC expansion cards like this one that comes with a riser kit, and various adapter cards like this PER-T641 card that includes an 8-bit DIO port and two USB 2.0 interfaces as well as these M.2 E and B key modules that can be used to add on Wi-Fi connectivity and additional AI acceleration performance, respectively, as needed.
So now, you know what’s needed to increase machine intelligence at the industrial edge. It’s called the de next-TGU8 from AAEON. The single board computer is in pre-production, but if you’d like to purchase your own, you can request a quote through the product page on the AAEON website. Or, if you feel like playing games, you can always enter the raffle below to win this board here, along with all its wonderful expansion cards, for absolutely free. And if you are our lucky winner, we’ll even ship it to you anywhere in the world (also for free).
So, good luck in this week’s raffle, but all you AI developers out there should keep in mind that we have a strict no-artificially intelligent raffle winners policy here on Dev Kit Weekly so no tricky entry attempts, please.
As always, thanks for watching, good luck in the raffle, and we’ll see you next week on Dev Kit Weekly.