AI in a box offline LLM solution leverages Rockchip RK3588S’ NPU (Crowdfunding)

offline open source LLM Radxa Rock 5A

Useful Sensors “AI in a box” LLM (large language model) solution works offline with complete privacy and leverages the NPU in Rockchip RK3588S processor for conversational AI similar to ChatGPT but without an internet connection or registration required. The AI box prototype currently relies on off-the-shelf hardware, specifically the Radxa ROCK 5A SBC with 8GB RAM, housed in a plastic enclosure, and the code relies on open-source models like Whisper speech-to-text model and Llama2 language models.   Besides conversational AI where you can interact with the box as if you were talking to a person, the AI in a Box can also be useful for other use cases: Live Captions – The box can display subtitles/closed captions for a live event or help in situations where people have trouble hearing a conversation using the audio input. Live Translation – It can also translate various languages close to real-time. Simply select […]

Amlogic S905X5 Armv9 TV Box SoC supports AV1, H.266, Ai-SR

Amlogic S905X5 TV Box

The Amlogic S905X5 may be first the Armv9 processor designed for set-top boxes (STB) and TV boxes, supports the latest AV1 and H.266 video codecs, and also comes with a 4 TOPS AI accelerator. We don’t have the full details yet, and the news comes from a tease from SEI Robotics which is about to unveil an Amlogic S905X5 AI-SR (AI Super Resolution) Android TV box at IBC 2023. Preliminary specifications for S905X5 SoC and SEI Robotics TV box: SoC – Amlogic S905X5 CPU – Quad-core Armv9 CPU (likely Cortex-A510 cores) delivering 40K+ DMIPS GPU – Arm Mali-G310 V5 entry-level GPU clocked at 1 GHz for 4K HDR user interfaces VPU – Up to 2x 4Kp60 10-bit H.266/VCC and AV1 video codecs, and the usual VP9, AVS3, H.265/HEVC, H.264 AI accelerator – 4 TOPS NPU enabling AI-SR Process – “Second generation” 6 nm process System Memory & Storage – TBD […]

Review of CM4 XGO Lite – A Raspberry Pi CM4 based smart robot dog with a robotic arm

XGO CM4 Lite Review Raspberry Pi Robot

The CM4 XGO Lite is a smart robot dog based on Raspberry Pi CM4 system-on-module and designed to learn to program using Blockly, Python, and ROS. This four-legged robot also happens to feature a 3-joint robot arm and a robot gripper installed on the back that can pick up light objects. The Raspberry Pi CM4 module drives the LCD screen and camera and performs AI and computer vision processing, while each joint is controlled with a servo motor, and a 6-axis tilt sensor ensures stable walking and movement. We’ve already discussed the capabilities of the CM4 XGO Lite, aka XGO Lite 2, when it was announced earlier this year, so we’re not going to go into details here, but some of the highlights include support for faster AI edge computing applications such as face detection and object classification, omnidirectional movement, six-dimensional posture control, posture stability, and multiple motion gaits. Robot […]

Cadence Neo NPU IP scales from 8 GOPS to 80 TOPS

Cadence Neo NPU

Cadence Neo NPU (Neural Processing Unit) IP delivers 8 GOPS to 80 TOPS in single core configuration and can be scaled to multicore configuration for hundreds of TOPS. The company says the Neo NPUs deliver high AI performance and energy efficiency for optimal PPA (Power, Performance, Area) and cost points for next-generation AI SoCs for intelligent sensors, IoT, audio/vision, hearables/wearables, mobile vision/voice AI, AR/VR and ADAS. Some highlights of the new Neo NPU IP include: Scalability – Single-core solution is scalable from 8 GOPS to 80 TOPS, with further extension to hundreds of TOPS with multicore Supports 256 to 32K MACs per cycle to allow SoC architects to meet power, performance, and area (PPA) tradeoffs Works with DSPs, general-purpose microcontrollers, and application processors Support for Int4, Int8, Int16, and FP16 data types for CNN, RNN and transformer-based networks. Up to 20x higher performance than the first-generation Cadence AI IP, with […]

Rockchip RK3588 AIoT gateway supports WiFi 6, 5G, RS232, RS485, LoRaWAN, BLE, and Ethernet

Dusun Pi 5 SDGW-380

Dusun DSGW-380, also called Dusun Pi 5, is an industrial AIoT gateway powered by the Rockchip RK3588 octa-core processor with a 6 TOPS AI accelerator and supporting a wide range of connectivity options. The gateway comes with 8GB LPDDR4 memory, and up to 128GB eMMC flash, and operates in a wide -25 to +75°C temperature range. It supports dual gigabit Ethernet, RS232, and R485 wired connectivity and various wireless protocols including WiFi 6, Bluetooth LE, 5G, and LoRaWAN. Dusun DSGW-380 specifications: SoC – Rockchip RK3588 octa-core processor with CPU – 4x Cortex‑A76  cores @ up to 2.4 GHz, 4x Cortex‑A55 core @ 1.8 GHz GPU – Arm Mali-G610 MP4 “Odin” GPU @ 1.0 GHz Video decoder – 8Kp60 H.265, VP9, AVS2, 8Kp30 H.264 AVC/MVC, 4Kp60 AV1, 1080p60 MPEG-2/-1, VC-1, VP8 Video encoder – 8Kp30 H.265/H.264 video encoder AI accelerator – 6 TOPS NPU System Memory – 8GB RAM Storage – […]

Andes launches AX45MPV RISC-V CPU core with Vector Extension 1.0

AX45MPV RISC core vector extension

Andes Technology has recently announced the general availability of the AndesCore AX45MPV RISC-V CPU which builds upon the AX45MP multicore processor and adds RISC-V Vector Extension 1.0. Equipped with RISC-V vector processing and parallel execution capability, the new RISC-V CPU core targets SoCs processing large amounts of data for applications such as ADAS, AI inference and training, AR/VR, multimedia, robotics, and signal processing. AX45MPV key features and specifications: 64-bit in-order dual-issue 8-stage CPU core with up to 1024-bit Vector Processing Unit (VPU) – compliant with RISC-V V-extension (RVV) 1.0 + custom extensions Supports clusters of up to 8 cores L2 cache and coherence support High bandwidth vector local memory (HVM) AndeStar V5 Instruction Set Architecture (ISA) Compliant with RISC-V GCBPV extensions Andes performance extension Andes CoDense extension for further compaction of code size Separately licensable Andes Custom Extension (ACE) for customized scalar and vector instruction 64-bit architecture for memory space […]

Gateworks GW7400 networking SBC features 6 Gigabit Ethernet ports, M.2 & mini PCIe sockets for wireless connectivity

NXP I.MX 8M Networking SBC 6 Ethernet ports

Gateworks Venice GW7400 is a rugged industrial single board computer for networking and gateway applications with six Gigabit Ethernet ports, and one M.2 and three mini PCIe sockets for WiFi, cellular, and other wireless options. We first mentioned the GW7400 board around three years ago, when the company introduced the Venice family powered by NXP i.MX 8M processor. But the board launched today has a different design with more Ethernet, an NXP i.MX8M Plus AI processor, up to 4GB RAM, up to 64GB eMMC flash, and various interfaces such as USB 3.0, dual CAN Bus, DIO, I2C, SPI, and UART. Gateworks GW7400 specifications: SoC – NXP i.MX 8M Plus CPU – Quad-core Cortex-A53 @ up to 1.6 GHz (industrial-grade) Real-time core – Arm Cortex-M7 @ up to 800 MHz GPU – Vivante GC7000UL 3D GPU, Vivante GC520L 2D GPU VPU – Up to 1080p60 video encoding and decoding AI accelerator […]

Allwinner 2023-2024 roadmap reveals A736/A737 Arm Cortex-A78/A76 processors

Allwinner Roadmap 2023 2024 automotive industrial SoCs

Allwinner should launch new Cortex-A76/A55 and Cortex-A78/A55 processors in 2024 according to the company’s roadmap including the Allwinner A736/A737 for tablets and the T736/T737 designed for automotive and industrial applications. In recent years, we’ve seen Rockchip and Amlogic introduce more powerful processors with the Rockchip RK3588 octa-core Cortex-A76/A55 processor and Amlogic A311D2 octa-core Cortex-A73/A53 or the more recent Amlogic S928X Cortex-A76/A55 for 8K TV boxes. But we’re still seeing some recent boards based on Allwinner Cortex-A7 32-bit processors, although recently we covered the Allwinner A523 octa-core Cortex-A55 processor for tablets. So today, I decided to go on a quest to find out whether Allwinner plans to use 64-bit Arm “big” cores in their future design. I first ended up on the linux-sunxi website where they list the Allwinner T736 octa-core “sun60i” processor with two Cortex-A76 cores and six Cortex-A55 cores, but no other details. This leads me to some “notes” […]

Exit mobile version