This week we review the Thundercomm TurboX AI Kit, which is based on the Thundercomm TurboX module, which is based on the Qualcomm SDA845 heterogeneous SoC, which is based on... just kidding. That rabbit hole matters because the SDA845 SoC packs in a ton of processing performance for different kinds of AI workloads, including:
8 Kyro CPUs
An Adreno 630 GPU
A Hexagon 685 DSP
A dual 14-bit Qualcomm Spectra 280 image signal processor accompanied by HDR10, HLG, and HEVC hardware encoders/decoders
All of that compute delivers 450 GigaFlops of computing performance, and is supported by 8GB of low-power DDR4 RAM, 64 GB of universal flash storage, and a MicroSD card interface. For those struggling with moving their facial recognition, object detection, or speech processing applications to the edge, this abundance of compute power, high-bandwidth I/O, and memory can surely help.
But what really makes the TurboX board special for AI developers is the software support that comes with it. This starts with the Qualcomm AI Engine, a software package comprised of the Qualcomm Neural Processing SDK and the Android NN API that really fast track the development of facial recognition and object detection applications. The AI engine comes equipped with pre-optimized algorithms and sample source code to help get the prototyping process started.
Developers can train models in AI frameworks like TensorFlow, Caffe, or other development frameworks that are compatible with the ONNX interchange format.
Brandon Lewis, Editor-in-Chief of Embedded Computing Design, is responsible for guiding the property's content strategy, editorial direction, and engineering community engagement, which includes IoT Design, Automotive Embedded Systems, the Power Page, Industrial AI & Machine Learning, and other publications. As an experienced technical journalist, editor, and reporter with an aptitude for identifying key technologies, products, and market trends in the embedded technology sector, he enjoys covering topics that range from development kits and tools to cyber security and technology business models. Brandon received a BA in English Literature from Arizona State University, where he graduated cum laude.
He can be reached by email at brandon.lewis@opensysmedia.com.
OmniVision Announces 1080p Image Sensor With Nyxel NIR and PureCel Plus Low Light Technologies
2.9 Micron Sensor for High Volume Surveillance Cameras Combines Low Power Consumption With High Performance, Selective Conversion Gain, Staggered HDR, and Near-Infrared and Low Light Capabilities
Build your dream ODM products with Shenzhen Wesion Technology Co., Ltd., an original design and manufacturing (ODM) company based out of Shenzhen China.
Maxim Integrated at Virtual CES 2021: Smaller, Smarter, More Secure
To understand how these ICs and technologies can function within your latest design, visit the Maxim Integrated virtual booth at CES 2021, starting on January 11, 2021.
OmniVision Announces Release of the OAX8000 AI-enabled ASIC
OmniVision announced the OAX8000 AI-enabled, automotive-specific integrated circuit (ASIC). The new ASIC is optimized for entry-level, stand-alone driver monitoring system (DMS).
Bigger Is Not Always Better: Exploiting Parallelism in NN Hardware
When considering hardware for high-performance neural networks, automotive engineers determine compute needs by simply adding NN requirements, which defines total system performance. Or does it?
Case Study: Providing Smart Hygiene Control in Food and Pharmaceutical Processing Plants
One part of the puzzle is to equip the food or pharmaceutical processing facility with the ability to provide smart hygiene control using special panel computers or box computers.
CoreAVI Releases New Safe AI and Computer Vision Solution
CoreAVI announced the release of a new Safe AI and Computer Vision Solution that includes the VkCoreVX™ SC, a safety critical implementation of Khronos’ OpenVX™1.3 API.
Transform Smart Building Data Into KPIs That Matter
Capturing IoT data about facilities performance is just the first step. The application of effective algorithms is key to implementing and capitalizing on a data analytics strategy.
Cloud Native Brings Computer Vision to the Critical Edge
Video analytics for critical operations at the edge presents numerous challenges, including computing resources, server deployment, and systems integration.
By leveraging powerful algorithms, deep learning technology, and intelligent analytics, the power of video has moved beyond security, providing a range of data that can help streamline operations.
2,088 Gbps/in2 – COM-HPC Connectors Increase Speed and Density
So, how do COM-HPC connectors enable 2,088 Gpbs/in2? To answer that question, we have to understand the basic design features of COM-HPC connectors and how they contribute to that metric.
COM-HPC Scales Heterogeneous Embedded Hardware into High-Performance Edge Computing
High-speed interfaces and provisions that allow modules to host PCIe targets mean that COM-HPC can support compute architectures ranging from Arm to GPUs to FPGAs and more.
If you sit down and make a list of promising technologies, the two names that would most likely top it are Artificial Intelligence and the Internet of Things.