MIPS Based TritonAI 64 AI IP Platform to Enable Inferencing & Training at the Edge

TritonAI 64 Block Diagram

After announcing their first MIPS Open release a few weeks ago, Wave Computing is back in the news with the announcement of TritonAI 64, an artificial intelligence IP platform combining MIPS 64-bit + SIMD open instruction set architecture with the company’s WaveTensor subsystem for the execution of convolutional neural network (CNN) algorithms, and WaveFlow flexible, scalable fabric for more complex AI algorithms. TritonAI 64 can scale up to 8 TOPS/Watt, over 10 TOPS/mm2 using a standard 7nm process node, and eventually would allow both inference and training at the edge. The platform supports 1 to 6 cores with MIPS64r6 ISA boasting the following features: 128-bit SIMD/FPU 8/16/32/int, 32/64 FP datatype support Virtualization extensions Superscalar 9-stage pipeline w/SMT Caches (32KB-64KB), DSPRAM (0-64KB) Advanced branch predict and MMU Integrated L2 cache (0-8MB, opt ECC) Power management (F/V gating, per CPU) Interrupt control with virtualization 256b native AXI4 or ACE interface Here’s the description provided by the company for their WaveTensor and WaveFlow …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Novasom M7 SBC aims to be a Drop-In Replacement for Raspberry Pi 3 in Industrial Projects

Novasom M7 Raspberry Pi Rockchip RK3328

Raspberry Pi boards are great for education and hobbyist projects, and while they are also found in industrial projects, they may not be the ideal solution for such commercial projects because of potential availability issues, stability issues in demanding environments, lack of certifications, and lack of commercial support. Novasom RASPMOOD family of SBCs aims to provide a drop-in replacement for Raspberry Pi based designs by providing mechanically and electrically compatible boards, as well as a software layer that allows the RASPMOOD boards to leverage the software you’ve already developed for your Raspberry Pi 3 design. Today we’ll specifically look at Novasom M7 “RASPMOOD” board – aka SBC-M7 – powered by a Rockchip RK3328 processor. Specifications: SoC – Rockchip RK3328 quad core Cortex-A53 processor with Mali-450MP4 GPU System Memory – Up to 4 GB DDR3 RAM Storage – Up to 256 GB eMMC Flash + uSD slot Video Output – HDMI 2.0 to to 4K @ 60 Hz, LVDS via external …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Android Q Beta and Preview SDK Released

Android Q ART Performance

Google has just announced the first public release of Android Q dubbed “Beta 1”- for Pixel devices only – as well as the corresponding preview SDK for developers. Android Q is said to bring  additional privacy and security features for users, enhancements for foldables, new APIs for connectivity, new media codecs and camera capabilities, NNAPI extensions, Vulkan 1.1 support, faster app startup, and more. What’s new in Android Q Privacy protections improvements More control over location with multiple options to “allow all the time”, “allow only while the app is in use” (e.g. not while running in the background), and “deny” Users will be able to control apps’ access to the Photos and Videos or the Audio collections via new runtime permissions Android Q will prevent apps from launching an Activity while in the background, relying on high-priority notifications instead Access to non-resettable device identifiers, including device IMEI, serial number, and similar identifiers will be limited Support for foldables and …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

HiSilicon Hi3559A Board Enables Smart 8K Camera Development

Hi3559A 8K camera board

HiSilicon Hi3559A 5-core Arm Cortex A73/A53 camera processor was announced in 2017 with support for 8K 30fps or 4K 120fps video recording, as well as an integrated dual core neural network accelerator (NNA). It should not be confused with the earlier Hi3559 4K 30 fps processor, or the newer Hi3559C that looks to be the same as Hi3559A without the NNA. Nevertheless, today I was lead to check about Hi3559A processor once again, and I could not find any camera that you can just buy and use today. I still found some upcoming Hi3559A products with OBSBOT Tail AI camera offered for as low as $509 on Kickstarter (delivery scheduled for April 2019), one 4K  120fps OEM? camera from GKuvision, as well as an HiSilicon Hi3559A development board for 8K camera applications, which I’ll look into more details in this blog post. HI3559AV100DMEB specifications: SoC – Hisilicon Hi3559A V100 with 2x Arm Cortex A73 cores @ 1.8 GHz, 2x Arm …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Orange Pi AI Stick 2801 Neural Compute Stick SDK & Documentation Released

How to use Orange Pi AI Stick SDK

We covered the launch of Orange Pi AI Stick 2801 neural compute stick a few days ago, the first easily available Gyralcon Lightspeeur based USB stick, and noted that while the hardware was there, we had no details about software development kit and documentation at the time. I’ve got some more information now. First, the company release an English presentation about the neural stick, and while it’s not technical documentation, it provides insights into what it is capable of, and an overview about the workflow. More importantly, Shenzhen Xunlong Software also released Orange Pi AI Stick SDK and user manual for x86 PC, as well as cat/dog data and pre-trained models to get started. However, if you want to perform training with your own data, you’ll need to purchase the stick plus PLAI model transformation and training tools sold on Aliexpress for $218 plus shipping, that’s $69 for the hardware, and $149 for the software tools. The company further explains …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

MAIX Development Boards with Sipeed M1 RISC-V AI Module Launched for $5 and Up (Crowdfunding)

MAIX GO Board

Sipeed M1 (aka MAIX-I)is a compact module based on Kendryte K210 dual core RISC-V processor designed for low power artificial intelligence workloads at the edge, such as face detection, object recognition, or audio processing. The module and some development boards are available on Taobao for the Chinese market, as well as YOYCart for the rest of the world, but the company has now launched several MAIX boards based on M1 module through an Indiegogo campaign with price starting at just $5 a piece. Sipeed MAIX Bit (aka MAIX Micro) is the cheapest one at $5 (early bird) / $6 with the following specifications: SoC – Kendryte K210 dual core 64-bit RISC-V processor @ 400 MHz (overclockable up to 800 MHz) with KPU CNN hardware accelerator APU audio hardware accelerator with support for up to 8 mics, up to 192 KHz sample rate FPIOA (Field Programmable IO Array) mapping 255 functions to all 48 GPIOs on the chip. 8 MB general purpose …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

GAPUINO GAP8 is a $229 RISC-V MCU Developer Kit for A.I. Applications

GAPuino Board

GreenWaves GAP8 is a low power RISC-V “MCU class” processor with eight compute cores optimized for artificial intelligence applications, and its main selling point is the ability to do tasks like computer vision or audio processing at very low power, even good enough to run on batteries. When we first covered GAP8 RISC-V processor at the beginning of the year, the company also mentioned a development kit comprised of GAPDUINO Arduino compatible board, a sensor board, and a QVGA camera module to experiment with the solution.  The board and development kit are now easier to purchase as the devkit is sold on SeeedStudio for $229. GAPuino board specifications: SoC – GAP8 IoT Application Processor with 8x RISC-V  compute cores, 1x RISC-V fabric controller core delivering up to 200 MOPS at 1mW and  >8 GOPS at a few tens of mW Memory / Storage –  HyperBus combo DRAM/Flash with 512 Mbit Flash + 64 Mbit DRAM; 256 Mbit Quad SPI flash …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Firefly Baidu Face Recognition Kit Comes with Monocular, Binocular, or Structured Light Camera

Baidu Face Recognition Kit

Firefly AIO-3399J industrial board comprised of a Rockchip RK3399 SoM and a baseboard offering plenty of I/O & connectivity options has been bundled with various  other accessories including a 10.1″ touchscreen display, and cameras to create a development platform for Baidu AI offline (aka “at the edge”) face recognition technology. The development kit is available in different variants with either a monocular camera, a binocular camera, or a structured-light camera. There’s also a deluxe kit with WiFi and a stand. Baidu Face Recognition Kit key specifications: SoC – Rockchip RK3399 hexa-core big.LITTLE processor with dual core ARM Cortex A72 up to 2.0 GHz and quad core Cortex A53 processor, ARM Mali-T860 MP4 GPU with OpenGL 1.1 to 3.1 support, OpenVG1.1, OpenCL and DX 11 support System Memory – 2GB DDR3 RAM Storage – 16GB eMMC 5.1 flash, micro SD card slot Display – 10.1″ 1280×800 capacitive touch display Connectivity – Gigabit Ethernet, Dual band WiFi, Bluetooth 4.1 LE, and mPCIe …

Support CNX Software – Donate via PayPal or become a Patron on Patreon