Arm introduced their very first microNPU (Micro Neural Processing Unit) for microcontrollers at the beginning of the year with Arm Ethos-U55 designed for Cortex-M microcontrollers such as Cortex-M55, and delivering 64 to 512 GOPS of AI inference performance or up to a 480x increase in ML performance over Cortex-M CPU inference. The company has now unveiled an update with Arm Ethos-U65 microNPU that maintains the efficiency of Ethos-U55 but enables neural network acceleration in higher performance embedded devices powered by Arm Cortex-A and Arm Neoverse SoCs. Arm Ethos-U65 delivers up to 1 TOPS, and as seen in the diagram enables features that can not be done with Ethos-U55 including object classification and real-time classification. Compared to Ethos-N78 NPU, the new microNPU offers less AI performance, but a significantly higher efficiency although AFAIK no quantified by Arm. The company says the development workflow remains the same with the use of the TensorFlow Lite Micro (TFLmicro) runtime that runs on a Cortex-M […]
Google Coral SBC was the first development board with Google Edge TPU. The AI accelerator was combined with an NXP i.MX 8M quad-core Arm Cortex-A53 processor and 1GB RAM to provide an all-in-all AI edge computing platform. It launched for $175, and now still retails for $160 which may not be affordable to students and hobbyists. Google announced a new model called Coral Dev Board Mini last January, and the good news is that the board is now available for pre-order for just under $100 on Seeed Studio with shipping scheduled to start by the end of the month. Coral Dev Board Mini specifications haven’t changed much since the original announcement, but we know a few more details: SoC – MediaTek MT8167S quad-core Arm Cortex-A35 processor @ 1.3 GHz with Imagination PowerVR GE8300 GPU AI/ML accelerator – Google Edge TPU coprocessor with up to 4 TOPS as part of Coral Accelerator Module System Memory – 2GB LPDDR3 RAM Storage – […]
Earlier this year, we covered some video analytics solutions based on AAEON UP Xtreme Edge embedded computer combining an Intel Whiskey Lake processor with Intel Movidius Myriad X AI accelerator modules, as well as video management & analytics software solutions from Milestones & SAIMOS, or aotu.ai BrainFrame. iWave Systems has now introduced a similar solution with Corazon-AI gateway capable of handling up to 8 IP cameras in real-time, but instead of relying on AI accelerators, the company leverages Xilinx Zynq Ultrascale+ Arm Cortex-A53/R5 FPGA MPSoC for AI inference. Corazon-AI gateway specifications: SoC – Xilinz Zynq Ultrascale+ ZU2, ZU3, ZU4 or ZU5 MPSoC Processing System (PS) Quad/Dual Arm Cortex-A53 @ 1.5GHz, dual Cortex-R5 @ 600MHz Arm Mali-400MP2 GPU @ 677MHz H.264/H.265 Video Encoder/Decoder Programming Logic (PL) Up to 256K Logic cells PL GTH Transceivers x 4 @ 12.5 Gbps System Memory 64bit, 2GB DDR4 with ECC for PS (upgradable) 32bit, 1GB DDR4 for PL (upgradable) Storage – 8GB eMMC Flash (upgradable), […]
Last week we wrote about Hailo-8 M.2 card delivering up to 26 TOPS of AI performance, and comparing well against Google Edge TPU and Intel Movidius Myriad X both in terms of footprint, performance, and efficiency. I’ve since then had a conference call with Liran Bar, VP of Business Development for Hailo, where we had time to discuss more about Hailo’s AI solutions, and how to interpret & understand AI benchmarks that may be misleading in many instances. Hailo-8 Architecture In the first post, we noted the chip managed to get the extra performance and efficiency thanks to a “proprietary novel structure-driven Data Flow architecture instead of the usual Von Neumann architecture”. But that’s a bit abstract, so Liran told me one of the key reasons for the performance improvement is that RAM is self-contained without the need for external DRAM like other solutions. This decreases latency a lot and reduces power consumption. Internally, the Hailo-8 chip is comprised of […]
We previously covered Charmed Labs PIXY2 computer vision camera based on an NXP LPC4330 microcontrollers that worked with Arduino, Raspberry Pi, and other development boards. The company is now back with a fully integrated more powerful solution with Vizy AI camera featuring a Raspberry Pi 4 SBC with up to 8GB RAM. Vizy AI camera key features and specifications: SBC – Raspberry Pi 4 with Broadcom BCM2711 quad-core Arm Cortex-72 processor, up to 8 GB RAM Camera – High-resolution camera based on Sony iMX477 12.3 MP sensor that can capture at over 300 frames/second and support both daytime and nighttime viewing; Both M12 and C/CS lenses are supported Video Output – 2x micro HDMI ports Audio – Analog stereo audio port Networking – Gigabit Ethernet, dual-band WiFi 5, and Bluetooth 5.0 USB – 2x USB 3.0 ports, 2x USB 2.0, 1x USB Type-C port from Raspberry Pi 4 (But not sure it’s actually used) Expansion – 8-pin terminal block for […]
VIA Technologies has launched the VIA Pixetto vision sensor aims to teach artificial intelligence (AI) and machine learning (ML) to students aged twelve and above. The Linux board features an Hisilicon Hi3518E V300 Cortex-A7 HD camera SoC combined with 64MB RAM, 128MB SPI flash, a MicroSD slot, as well as GalaxyCore GC2053 2M CMOS sensor offering 1920 x 1080 resolution and a 130° field of view. VIA Pixetto hardware specifications: SoC – Hisilicon Hi3518E V300 Cortex-A7 HD camera SoC @ 900 MHz with H.264, H.265, MJPEG/JPEG encoder System Memory – 64MB DDR, MicroSD card slot Storage – 128MB SPI flash Camera – GalaxyCore GC2053 2M CMOS sensor with 1920 x 1080 resolution, 130° FoV Audio – Microphone USB – 1x Micro USB 2.0 port Connectivity – WiFi 4 via Hisilicon Hi1131 low-power Wi-Fi chip and ceramic antenna Expansion – Grove connector Misc – 3x LEDs (Power, OS boot, active status), 1x reset button Power Supply 5V/500mA via Micro USB port […]
Yesterday, I wrote about what I felt what a pretty unique board: Evo M51 board following Adafruit Feather form factor, and equipped with an Atmel SAMD51 Cortex-M4F MCU and an Intel MAX 10 FPGA. But less than 24 hours later, I’ve come across another Adafruit Feather-sized Cortex-M4F board with FPGA fabric. But instead of using a two-chip solution, QuickLogic QuickFeather board leverages the company’s EOS S3 SoC with a low-power Cortex-M4F core and embedded FPGA fabric. QuickFeather specifications: SoC – QuickLogic EOS S3 with Arm Cortex-M4F Microcontroller @ up to 80 MHz and 512 Kb SRAM, plus an embedded FPGA (eFPGA) with 2400 effective logic cells and 64Kb RAM Storage – 16Mbit SPI NOR flash USB – Micro USB port with data signals tied to eFPGA programmable logic Sensors – Accelerometer, pressure sensor, built-in PDM microphone Expansion I/Os – Breadboard-compatible 0.1″ (2.54 mm) pitch headers including 20 Feather-defined GPIO + 13 additional GPIO with UART, I2C, I2S and SPI Debugging […]
Rockchip RK1808 neural network processing unit was initially an IP Block inside RK3399Pro, but the company eventually launched RK1808 Cortex-A35 processor as a standalone solution now providing up to 3.0 TOPS for AI inferencing in modules, USB sticks, and development kits. Boardcon offers another option with EM1808, a Rockchip RK1808 SBC equipped with the processor. The board should be suitable for two main types of AI applications, namely smart audio applications thanks to four audio ports, speaker header, & an onboard 4-mic array, and computer vision with MIPI CSI & DSI interfaces. Boardcon EM1808 board is comprised of a baseboard and CPU module with the following overall specifications: SoC – Rockchip RK1808 dual Cortex-A35 processor up to 1.6GHz with 3.0 TOPS (for INT8) NPU, VPU supporting H.264 1080p60 decode, 1080p30 encode System Memory- 2GB LPDDR3 Storage – 8GB eMMC flash, MicroSD slot, M.2 NVMe SSD interface Display I/F – 26-pin MIPI DSI header Camera I/F – 26-pin MIPI CSI header […]
Privacy & Cookies Policy
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.