AEWIN, Blaize partner on edge AI inference solutions
AEWIN has partnered with Blaize to integrate its multi-access edge computing platform with Blaize PCIe edge AI accelerator. Enterprise customers can benefit from the Blaize AI software suite to deliver high performance with system reliability, lower power consumption, and faster time to market for edge AI applications, the companies said.
“AEWIN’s collaboration with Blaize enables us to serve the demands of real-time traffic inference/analysis to accomplish AI applications at edge efficiently,” said Charles Lin, CEO of AEWIN. “As the global MEC markets continue to grow by leaps and bounds, AEWIN-Blaize Edge server with utmost performance and efficiency is our response for this demand.”
As part of the collaboration, AEWIN’s SCB-1932 MEC network appliance will be equipped with the Blaize Xplorer X1600P-Q PCIe accelerator for edge AI applications. The product is a 2U rackmount platform with a dual-core Intel Xeon Scalable processor LGA4189, and also supports up to 6TB DDR4 memory which is exceptionally over the demand from complex edge AI applications.
The SBC-1932 will utilize the Blaize X1600P-Q AI accelerator to provide an enterprise solution that is designed for edge and AI inference applications, enabling industrial PCs, and servers to use integrated AI inference.
The Blaize Xplorer X1600P-Q accelerator is based on the Blaize Graph Streaming Processor (GSP) architecture which is designed to run efficiently and is programmable using the Blaize.Picasso SDK and Blaize AI Studio software. The Blaize Graph Streaming Processor architecture leverages task-level parallelism and streaming execution processor to provide the flexibility of low-energy consumption, high performance, and improved scalability. Through this, multiple AI workloads can operate without any performance tradeoff with energy consumption.
At the heart of the Blaize Xplorer X1600P-Q are four Blaize 1600 system-on-chips with each 16 GSP cores providing 64 to 80 TOPs performance. For the enterprise version, the maximum power consumption is 75W is far less than many other commercially available chips fo rinferencing. The support for Blaize Picasso SDK accelerates the AI development process through the support for machine learning frameworks, such as TensorFlow, PyTorch, Caffe2, and ONNX.
“Our partnership with AEWIN Technologies provides a customizable solution for smart city utilizations in traffic management, pedestrian safety, and multi-object detection,” said Dinakar Munagala, co-founder and CEO, of Blaize. “Easily programmable with the Blaize Picasso SDK and AI Studio software, the X1600P-Q PCIe software-defined accelerator runs efficiently in a complete streaming fashion, critical to providing secure and reliable systems to the Global marketplace.”
Blaize recently partnered with Innovatrics to build facial recognition applications for security and surveillance applications. In the rising demand for computer vision applications, UK chip-designer Arm has also announced the launch of a new image signal processor, the Mali-C55, to provide improved image quality and better performance for complex IoT vision systems.
Article Topics
AEWIN | Blaize | edge AI | enterprise | inference
Comments