Advanced technology boosts AI-powered intelligent edge solutions and expands neural network capabilities.
Microchip Technology announced the acquisition of Neuronix AI Labs, a move set to enhance its range of AI-enabled edge solutions on field programmable gate arrays (FPGAs). Neuronix AI Labs specializes in neural network sparsity optimization technology, which reduces power requirements, physical size, and computational load for functions such as image classification, object detection, and semantic segmentation while preserving accuracy.
The company’s PolarFire FPGAs and SoCs are known for their low power consumption, reliability, and security. The integration of Neuronix’s technology enables the company to provide more cost-effective, scalable solutions for computer vision applications in systems with strict cost, size, and power constraints. This acquisition is targeted at significantly enhancing AI/ML processing capabilities on both low and mid-range FPGAs.
This technology acquisition also democratises access to advanced parallel processing capabilities for non-FPGA designers. They can now use standard AI frameworks without needing detailed knowledge of FPGA design processes. Integrating Neuronix AI intellectual property with the company’s existing compilers and software design kits allows AI/ML algorithms to be implemented directly onto customisable FPGA logic without deep familiarity with register-transition level (RTL) or the intricacies of FPGA architecture. The technology facilitates on-the-fly updates and upgrades to convolutional neural networks (CNNs) without the need for hardware reprogramming.
Yaron Raz, CEO of Neuronix AI Labs, highlighted their focus on developing superior neural network acceleration technologies, aiming to redefine expectations regarding size, power, and cost. “Joining forces with Microchip enables us to expand and align with an FPGA portfolio renowned for setting benchmarks in power efficiency,” Raz commented.
Bruce Weyer, corporate vice president of Microchip’s FPGA business unit, emphasized the impact of this acquisition, stating, “The inclusion of Neuronix AI Labs’ technology enhances our FPGA and SoC offerings in intelligent edge systems employing AI/ML algorithms.” He noted that the merger of Neuronix technology with Microchip’s VectorBlox design flow will increase neural network efficiency and optimize GOPS/watt performance in their low-power devices. System designers will now have the tools to create compact hardware configurations that were previously challenging due to size, thermal, or power limitations.
