Arm Helium Delivers up to 15x Performance Uplift for Machine Learning on Cortex-M MCUs

Arm Helium

Arm has just unveiled Armv8.1-M architecture that adds Arm Helium technology, the M-Profile Vector Extension (MVE) for the Arm Cortex-M cores that will improve the compute performance of Cortex-M based microcontrollers. Helium will deliver up to 15 times more machine learning (ML) performance and up to 5 times uplift to signal processing allowing local decision-making on low-power embedded devices. Helium instructions will enable new applications for Arm Cortex-M microcontrollers in audio devices, sensor hubs, keyword spotting, voice command control, power electronics, communications and still image processing. Helium and Neon (the Advanced SIMD technology for Arm Cortex-A processors) are similarities but Helium has been designed for efficient signal processing performance in small processors. One different illustrated below is that while NEON loads 128-bit instructions (e.g. VLDR, VLMA), Helium will split up 128‑bit wide instruction into four equally sized chunks, called “beats” (labelled A to D) due to difference between Cortex-M and Cortex-A cores as explained in details in an Arm Research …

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

Ubuntu 18.04 Now Boots on Some Snapdragon 835 Arm Laptops

Snapdragon 835 Ubuntu Arm Laptop

The first “proper” Windows 10 Arm laptops were unveiled at the end of 2017 and beginning of 2018, all based on Qualcomm Snapdragon 835 processor with always-on LTE connectivity, 20+ hour battery life, a fairly expensive price tag, and somewhat underwhelming performance. Qualcomm was not interested in supporting Linux, but there was interest from the community, and now it seems Ubuntu 18.04 images are available for Lenovo Miix 630, HP Envy x2, and ASUS Novago TP370 thanks to Aarch64-laptop project currently hosted on Github. Now the prebuilt images are not really ready for end users since UFS storage and WiFi are not working on any laptop yet, the touchpad is not working on the ASUS laptop, and accelerated graphics needs to be implemented. Interestingly WiFi is related to UFS on those laptops, and Marc Gonzalez is said to be being actively worked on UFS upstream support, which should enable for internal storage and WiFi. That means now you’d need to …

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

Autoware is an “All-in-One” Open-source Software for Autonomous Driving

Autoware.AI Software Architecture

All major automotive companies, and some technology companies are all working on autonomous driving with the ultimate goal of achieving level 5 autonomous driving meaning no human intervention is needed at any stage. Development will take some more time, and companies are now competing with closed source software and hardware. But as I browsed through Linaro Connect Bangkok 2019 schedule, I found out there’s an open source autonomous driving software called Autoware.AI. Several “Autoware” projects are managed by the newly founded Autoware Foundation, a non-profit organization created to develop a synergy between corporate development and academic research in order to provide access to autonomous driving technology for everyone: Autoware.AI is the first version built on ROS 1, and Linux, and has been developed as a research and development platform Autoware.auto is the second version built on ROS 2, and Linux, with a complete redesign. Autoware.IO is an interface project for Autoware to be extended with proprietary software and third-party libraries for …

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

OpenISA VEGAboard Combines RISC-V and ARM Cortex-M Cores

VEGAboard

OpenISA has launched an Arduino compatible RISC-V development called VEGAboard that features RV32M1 wireless microcontroller with a RISC-V RI5CY core, a RISC-V ZERO-RISCY core as well as Arm Cortex-M4F and Cortex-M0 cores, and a radio operating in the 2.36 GHz to 2.48 GHz range. An external NXP Kinetis K26 Arm Cortex-M4 MCU is added to the board for OpenSDA (Open-Standard Serial and Debug Adapter) debugging over a single USB cable. The board was offered for free, I’m just not sure when, but they are already out of stock. Hopefully, they’ll start selling the board soon enough. VEGAboard (RM32M1-VEGA) board key features and specifications: Ultra-low-power RV32M1 Wireless MCU supporting BLE, Generic FSK, and IEEE Std 802.15.4 (Thread) platforms IEEE Std. 802.15.4-2006 compliant transceiver supporting 250 kbps O-QPSK data in 5.0 MHz channels, and full spread-spectrum encoding and decoding Fully compliant Bluetooth v4.2 Low Energy (BLE) Reference design area with small-footprint, low-cost RF node: Single-ended input/output port Low count of external components …

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

MXCHIP EMW3060 WiFi IoT module sells for less than $2

MXCHIP EMW3060

Over the last few years, we’ve seen several Arm based WiFi IoT modules trying to compete with Tensilica based ESP8266 including Pine64 PADI IoT Stamp and Air602 WiFi IoT modules with similar functionalities and price (around $2) as the Espressif modules. I have not seen many projects with any of those Arm based wireless modules, likely because Espressif Systems software ecosystem is mature, and it’s hard to compete for new entrants. Nevertheless there’s a new Arm WiFi IoT module, not based on Cortex-M core, but instead, MXCHIP EMW3060 is equipped with an older ARM9 core, 256KB RAM, and 2MB flash, and sells for just $1.79 on Seeed Studio in single quantity. The price further drops to $1.55 per unit if you order 50 pieces or more. MXCHIP EMW3060 WiFi IoT module specifications: SoC ARM9 core clocked at 120 MHz System Memory – 256KB SRAM Storage – 2MB flash Connectivity – 2.4GHz radio, 802.11 b/g/n WiFI, HT/40, support Station, Soft AP, …

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

Huaxintong StarDragon 4800 Server SoC is Based on Qualcomm Centriq 2400 Processor

Qualcomm started shipping samples of their Arm based Centriq 2400 server processors in 2016, before launching mass production the next year with three parts including Qualcomm Centriq 2460 48-core processor. Development seemed to go along nicely until Qualcomm allegedly decided to exit the server market in the middle of last year. The story got confusing when GIGABYTE still decided to launch their H221-Q20 server powered by Qualcomm Centriq 2400 processor last November, which would not make sense if Qualcomm is really existing the server market, and parts would not become unavailable after a short. But today, as I read the slides of GIGABYTE presentation at the HPC Asia workshop that took place on January 14-16, I realized Centriq 2460 is still alive but just changed owners… So the GIGABYTE H221-Q20 server is compatible with StarDragon 4800… What is that? StarDragon sounds familiar, a bit like Qualcomm Snapdragon. It turns out Qualcomm collaborated with the government of Guizhou in China to …

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

V-Raptor is a 24-Core Arm Server based on SocioNext SC2A11 SoC

SocioNext SC2A11 Module

Socionext SC2A11  24-core Arm Cortex-A53 processor has gotten into more hardware recently. We recently showcased a demo of an upcoming Banana Pi 24-core Arm server, but they’ll be joined by South Korea based XSLAB which prepare to launch their own V-Raptor 24-core microserver in February 2019. The server is based on SC2A11 processor coupled with their own BMC (Baseboard Management Controller). A 24-core micro server node is shown below with SC2A11 processor covered by a black heatsink, as well as a RAM stick. As I understand it, one or several V-Raptor microserver can then be inserted into a PCIe slot of the mainboard which supports up to 32 nodes on 2U rack size, so that would be a total of 24 x 32 = 768 Arm Cortex A53 cores. The detailed specifications of the server are not shown yet on the almost empty company website. The company may also have made a separate server, as shown in the photo below, where …

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

Huawei Kunpeng 920 is a 7nm 64-core Armv8 Server Processor

Huawei Kunpeng 920

We’ve seen several Armv8 processors for the datacenter launched in recent years from companies like Cavium – now part of Marvell – and Ampere. Another company now joins the fray with Huawei having just introduced their Kunpeng 920 64-core Armv8 processor manufactured with a 7nm process, and offering 8-channel DDR4 RAM, and fast interfaces such as 10GbE and PCIe 4.0. Kunpeng 920 features and specifications: CPU – 64x Armv8 cores clocked at up to 2.6 Ghz delivering a 930 SPECint score Memory I/F – 8x DDR4 @ 2933 MHz for 6 to 8 channels memory Storage – 16x SAS/SATA interfaces High Speed Interfaces – 40x PCIe 4.0 including 16 that can be used for CCIX (Cache Coherent Interconnect for Accelerators); 640 Gbps total bandwidth Networking – 2x 100G RoCE (RDMA over Converged Ethernet) Process – 7nm process The company customized their ARMv8 cores in order to improve performance by optimizing branch prediction algorithms, increasing the number of OP units, and …

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