machine learning

Posted on by Debashis Das - No Comments on 1GHz Renesas RA8D2 and RA8M2 Cortex-M85 MCUs feature up to 1MB MRAM, 2MB SRAM

1GHz Renesas RA8D2 and RA8M2 Cortex-M85 MCUs feature up to 1MB MRAM, 2MB SRAM

Renesas RA8M2 and RA8D2 dual core Cortex M85, M33 MCU with MRAM storage

After introducing the 1 GHz RA8T2 MCU earlier this month, Renesas has recently expanded its RA8 lineup with the RA8D2 and RA8M2 groups of MCUs, where the RA8D2 is designed for graphics, HMI, and AI applications with support for display, camera, and audio interfaces, while the RA8M2 targets general-purpose, high-performance IoT and industrial control systems with networking and compute capabilities. Both MCU groups deliver a performance of up to 7,300 Coremarks, offer up to 1MB of MRAM and 2MB of SRAM, and support SiP options with up to 8MB flash. The RA8D2 adds HMI capabilities, including a 1280×800 graphics LCD controller, a 2D drawing engine, MIPI DSI and CSI-2 interfaces, and audio input support for voice and vision AI. Common features include dual Gigabit Ethernet with TSN, USB 2.0, CAN FD, I3C, and various analog peripherals. Integrated RSIP-E50D security, Helium acceleration for DSP/ML with various software support. Renesas RA8D2 and […]

Posted on by Debashis Das - 18 Comments on GIGABYTE AI TOP ATOM – An NVIDIA GB10 desktop AI supercomputer with 1 petaFLOP AI, 10GbE, 128GB RAM

GIGABYTE AI TOP ATOM – An NVIDIA GB10 desktop AI supercomputer with 1 petaFLOP AI, 10GbE, 128GB RAM

GIGABYTE AI TOP ATOM, NVIDIA Grace Blackwell GB10 Superchip

GIGABYTE has introduced the AI TOP ATOM, a compact desktop AI supercomputer with 1 petaFLOP of AI performance, which is very similar to NVIDIA DGX Spark. That’s because it’s also built around the NVIDIA Grace Blackwell GB10 Superchip. Housed in a 1-liter chassis, it’s designed for generative AI, large language models, and machine learning workloads directly on the desktop. The AI TOP ATOM features 128 GB of LPDDR5x unified memory, supports up to 4 TB PCIe Gen5 SSD storage. The 1,000 TOPS (1 petaFLOP) FP4 of AI compute enables it to handle models with up to 200 billion parameters, or 405 billion in a dual-system configuration. Connectivity includes 10GbE networking, Wi-Fi 7, Bluetooth 5.3, HDMI 2.1a, and multiple USB 3.2 Gen 2×2 Type-C ports. GIGABYTE AI TOP ATOM specifications: SoC – NVIDIA GB10 CPU – 20-core Armv9 processor with 10x Cortex-X925 cores and 10x Cortex-A725 cores Architecture – NVIDIA Grace […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 3 Comments on Google’s open-source, RISC-V-based Coral NPU is integrated into Synaptics SL2610 Edge AI SoCs

Google’s open-source, RISC-V-based Coral NPU is integrated into Synaptics SL2610 Edge AI SoCs

Google Coral NPU

Google has very recently introduced Coral NPU full-stack, open-source RISC-V-based platform for always-on AI on low-power edge devices and wearables. The first chip to integrate the Coral NPU is the upcoming Synaptics Astra SL2610 family. Google Coral NPU The Coral NPU aims to address the software fragmentation on entry-level AI accelerators that makes them difficult to program. By releasing an open-source NPU and associated source code, Google hopes its design will be adopted by silicon vendors, reduce software fragmentation over time, and help machine learning (ML) developers bring products to market faster. Building on the works on the Coral platform, the new, open-source Coral NPU is comprised of three main components: A scalar core – A lightweight, C-programmable RISC-V core that manages data flow to the back-end cores. It uses a simple “run-to-completion” model for ultra-low power consumption and traditional CPU functions. A vector execution unit – A single instruction […]

Posted on by Debashis Das - No Comments on Ambiq Apollo510B ultra-low-power Cortex-M55 Edge AI MCU adds Bluetooth LE 5.4

Ambiq Apollo510B ultra-low-power Cortex-M55 Edge AI MCU adds Bluetooth LE 5.4

Ambiq Apollo510B

After the release of Apollo510, Ambiq has released Apollo510B, an ultra-low power Edge AI MCU that adds a 48 MHz network coprocessor for Bluetooth 5.4 LE (BLE) support. The new SoC combines Cortex-M55 with Helium MVE for AI/ML acceleration, secureSPOT 3.0 security, and graphiqSPOT 2.0 graphics for connected wearables, healthcare devices, and industrial IoT applications. The Apollo510B features 3.75MB of system RAM, 4MB of non-volatile memory, and a 12-bit ADC. It supports MIPI DSI and Quad SPI interfaces for displays, and offers audio capabilities such as always-on low-power ADC, a PDM stereo microphone interface, and dual multichannel I²S ports with asynchronous sample rate conversion. Peripheral options extend to USB 2.0 HS/FS, dual SDIO/eMMC controllers, multiple SPI and I²C masters, UART interfaces with flow control, and various GPIOs. Ambiq Apollo510B specifications: MCU Core Arm Cortex-M55 core at up to 250 MHz with Helium (MVE) vector instructions, FPU, TrustZone, MPU Caches/TCM – […]

Posted on by Jean-Luc Aufranc (CNXSoft) - No Comments on Snapdragon W5+ and W5 Gen 2 wearable platforms gain NB-NTN satellite support

Snapdragon W5+ and W5 Gen 2 wearable platforms gain NB-NTN satellite support

Snapdragon W5 Gen2 features NB-NTN satellite

Qualcomm Snapdragon W5+ and W5 Gen 2 are new wearable platforms adding NB-NTN satellite support for emergencies to the Snapdragon W5+/W5 wearable platforms introduced in 2022, and found in various WearOS smartwatches and the Beacon W5 SoM. The W5+/W5 Gen 2 platforms still feature a 1.7 GHz quad-core Cortex-A53 processor, an Adreno A702 GPU, an AON QCC5100 ML co-processor (W5+ Gen 2 only), MIPI DSI and MIPI CSI interfaces, cellular connectivity, dual-band WiFi, Bluetooth 5.3, GNSS, and optional NFC support. The main changes are NB-NTN support, an upgrade to 3GPP Rel 17 with Cat 1Bis, a 20% (or 30% depending on where you read) smaller RF frontend, and 50% more accurate GPS in urban or deep canyon areas. Snapdragon W5+/W5 Gen 2 specifications: SWS5100 SoC CPU – Quad-core Cortex-A53 processor @ up to 1.7 GHz GPU – Qualcomm Adreno A702 @ up to 1 GHz with OpenGL ES 3.1 API […]

Posted on by Debashis Das - 2 Comments on Banana Pi BPI-F4 – An industrial Edge AI SBC powered by Sunplus SP7350 SoC with 4.1 TOPS NPU

Banana Pi BPI-F4 – An industrial Edge AI SBC powered by Sunplus SP7350 SoC with 4.1 TOPS NPU

Banana Pi BPI F4 industrial Edge AI SBC

The Banana Pi BPI-F4 is a compact industrial-grade Edge AI development board built around the Sunplus SP7350 quad-core Cortex-A55 SoC with a 4.1 TOPS NPU. Compared to the SunPlus SP7021-based BPI-F2S SBC, the new BPI-F4 supports additional interfaces through terminal blocks and is suitable for AI vision, robotics, and control systems. The board supports booting from microSD or onboard eMMC and provides USB 3.0/2.0 ports, HDMI video output, Ethernet, and a MIPI CSI input for an OV5647 camera. Connectivity options include Gigabit Ethernet, Wi-Fi, and Bluetooth. Other features include seven terminal blocks for GPIO, ADC, SPI, I²C, UART, and PWM, jumper-based power and USB mode selection, and automatic boot. Banana Pi BPI-F4 specifications: BPI-F4-Core board SoC – Sunplus SP7350 (C3V) CPU Quad-core Cortex-A55 @ 2.1 GHz Cortex-M4 MCU @ up to 400MHz for always-on sub-system VPU – H.264 video decoding/encoding AI Accelerator – 4.1 TOPS NPU @ 900 MHz (Sometimes […]

Posted on by Jean-Luc Aufranc (CNXSoft) - No Comments on XIAOML Kit with ESP32-S3, camera, microphone, and IMU complements a free Machine Learning Systems book

XIAOML Kit with ESP32-S3, camera, microphone, and IMU complements a free Machine Learning Systems book

XIAOML Kit Machine Learning Systems Book

The XIAOML Kit is one of the devkits that complements Harvard University Professor Vijay Janapa Reddi’s book “Introduction to Machine Learning Systems“, available for free as a 2050-page PDF file. Made by Seeed Studio, the XIAOML Kit is composed of the XIAO ESP32S3 Sense with an ESP32-S3 WiFI and Bluetooth SoC, a microSD card slot, a built-in OV3660 camera and microphone, and the “IMU Breakout board” featuring a 6-axis IMU and 0.42-inch OLED display. The kit enables students, educators, and developers to build vision, sound, and motion applications through tinyML lab sessions developed with Marcelo Rovai (UNIFEI). XIAOML Kit specifications: Main Board – XIAO ESP32S3 Sense SoC – Espressif Systems ESP32-S3R8 dual-core Tensilica LX7 microcontroller @ 240 MHz with 512KB SRAM, 8MB PSRAM, Wi-Fi 4 & Bluetooth 5.0 dual-mode (Classic + BLE) connectivity Storage – 8MB flash, microSD card slot Sensors – OV3660 camera, digital microphone USB – USB-C port […]

Posted on by Debashis Das - No Comments on VS Code gets AutoML Embedded plugin for automated model tuning, deployment, and benchmarking

VS Code gets AutoML Embedded plugin for automated model tuning, deployment, and benchmarking

VS Code AutoML Embedded

AutoML for Embedded, developed by Analog Devices (ADI) and Antmicro, is an open-source plugin for Visual Studio Code that works alongside ADI’s CodeFusion Studio plugin. Built on the Kenning framework, it automates the full ML pipeline, including model search, hyperparameter tuning, optimization, compression, and deployment, making edge AI development easy on resource-constrained devices. The company mentions that it supports the ADI MAX78002 AI accelerator MCU, the MAX32690 MCU, Renode-based simulation, and Zephyr RTOS. It uses SMAC and Hyperband algorithms for automated model search and hyperparameter tuning, along with model compression and quantization to meet strict memory and compute limits. The plugin offers built-in benchmarking for inference speed, memory footprint, and accuracy, while performing RAM and compute compatibility checks. All these features make it useful for applications like image classification, anomaly detection, predictive maintenance, NLP, and action recognition on low-power IoT and embedded systems. AutoML Embedded overview: Type – Open-source AutoML plugin […]

Exit mobile version