AM67A BeagleY®-AI Evaluation Board for Edge AI Applications

BeagleBoard.org's BeagleY-AI is based on Texas Instruments’ AM67A Arm®-based vision processor, which includes a quad-core 64-bit Arm Cortex®-A53 CPU subsystem at a speed of 1.4 GHz. The dual general-purpose digital signal processing (DSP) subsystem includes vision/deep learning accelerators and is capable of 4 trillion operations per second (TOPS) of deep-learning algorithm performance in an industry-standard form factor.

Features

• Processor
• Texas Instruments AM67A, 64-bit Arm Cortex-A53 @ 1.4 GHz, multiple cores including Arm/GPU processors, DSP, vision/deep learning accelerators, and Arm Cortex-R5 @ 800 MHz I/O co-processor
• RAM
• 4 GB LPDDR4
• USB ports
• 4 x USB 3.0 Type-A ports supporting 5 Gbps operation, 1 x USB 2.0 Type-C®, supports USB 2.0 device mode
• Ethernet
• Gigabit Ethernet with PoE+ support via add-on
• Camera/display
• 2-lane x 4-lane MIPI camera connector (one connector muxed with DSI)
• Display output
• 1 x HDMI display, 1 x OLDI display, 1 x DSI MIPI display (muxed with CSI)
• Real-time clock (RTC)
• Supports external coin-cell battery (2648-SC1163-ND) for power failure time retention
• Debug UART
• 1-pin x 3-pin JST-SH 1.0 mm debug UART port
• Power
• 5 V/3 A DC power via USB-C
• PCIe® interface
• PCI Express Gen3 x 1 interface for fast peripherals [requires separate M.2 HAT (2648-SC1166-ND) or other adapter]
• Expansion connector
• 40-pin expansion header compatible with HATs
• Storage
• microSD card slot with UHS-1 support
• Tag connect
• 1 x JTAG, 1 x external PMIC programming port