New BeagleY-AI brings open source hardware to machine learning applications

Machine Learning


[Partner Content] It seems like everyone has an AI solution for embedded system hardware. However, in most cases, when you introduce that hardware into a product, you need to use it as-is, leaving no room for cost optimization. However, things are different with his BeagleBoard, which also applies the open source hardware approach to his new BeagleY-AI board.

It seems like everyone has an AI solution for embedded system hardware. However, in most cases, when you introduce that hardware into a product, you need to use it as-is, leaving no room for cost optimization. However, things are different with his BeagleBoard, which also applies the open source hardware approach to his new BeagleY-AI board.

BeagleY-AI is built using the same form factor as other credit card-sized single-board computers (SBCs) and is built around Texas Instruments' (TI) powerful AM67A system-on-chip (SoC). I am. Alongside a quad-core 64-bit Arm Cortex-A53 (1.4 GHz) are his two general-purpose C7x DSPs with a matrix multiplication accelerator (MMA) capable of 4 TOPs. These are complemented by an Arm Cortex-R5 that handles the real-time interface, GPU, video and vision accelerators.

AI and machine learning (ML) applications can be developed using TensorFlow and easily up and running on Linux-based operating systems. Optimizing ML algorithms is simplified thanks to the Arm native toolchain for his TI DSP for users who want to squeeze the most out of their board's performance. It's also an easy path to optimized hardware. As an open-source platform, developers can reuse schematics and BOMs to create their own cost-optimized hardware. At embedded world 2024, CTO Jason Kridner also demonstrated how to port his entire image recognition application disk from his Raspberry Pi to his BeagleY-AI, highlighting the biggest challenges in porting applications between SBCs. overcame his one.

Learn more about BeagleY-AI here.





Source link

Leave a Reply

Your email address will not be published. Required fields are marked *