Servers are powered by AI for HPC

At this year’s ISC in Frankfurt, Tyan, is showcasing its second generation Intel Xeon Scalable processor-based server platforms. According to the company, they are optimised for predictive analytics, machine learning (ML) and high performance computing (HPC) applications.

“The most demanding workloads including HPC, AI deep learning, data analytics, and multi-cloud infrastructure require maximum performance across compute, memory, storage, and network resources, “said Danny Hsu, vice president of MiTAC Computing Technology’s Tyan business unit

Powered by the newest generation of Intel Xeon Scalable processors, Tyan’s Thunder HX range takes advantage of Intel Deep Learning Boost technology to enable faster machine learning application execution. The Thunder HX FT83-B7119 supports 3Tbyte of memory, 12 hot-swappable 3.5-inch drives in a 4U enclosure. The system is suitable for AI training and inference applications.

Another 4U server system is the Thunder HX FT77D-B7109. It supports dual second generation Intel Xeon Scalable processors with integrated Intel Omni-Path 100Gabit networking fabric that can reach a total of 200Gbits per second fabric connectivity for massively parallel workloads such as scientific computing and large-scale facial recognition.

Tyan’s Thunder HX FT48T-B7105 is a pedestal workstation platform with dual second generation Intel Xeon Scalable processors and support for up to five, high performance workstation graphic cards. The workstation gives maximum I/O to professional power users and can be used for 3D rendering and image processing.

Intel Optane DC persistent memory and increased DDR4 memory speed and capacity features in the Thunder SX FA100-B7118, which is Tyan’s flagship storage server supporting dual second generation Intel Xeon Scalable processors and 100 hot-swap 3.5-inch drives in a 4U rackmount form factor. The platform is designed for both cold storage and large-scale object storage applications such as Lustre.

Visitors will also see the Thunder SX GT90-B7113 which features support for dual Intel Xeon Scalable processors, 12 hot-swap 3.5-inch drives mounted to a service drawer and four hot-swap NVMe U.2 drive bays in a 35-inch deep 1U chassis. The platform matches CSP server-based storage applications by allowing serviceability and density within the same server.

The Thunder CX GT24E-B5556 is a 1U single-socket Intel Xeon E-2100 processor-based server platform for CSP deployment. It has support for up to four DDR4 DIMM slots, one double-width PCIe x16 slot, and dual 10GBase-T Ethernet ports. The platform is optimised for cost-effective cloud gaming applications, confirms Tyan.

Visit Tyan at ISC 2019 (17-19 June) Messe Frankfurt, booth C-1250.

http://www.tyan.com

> Read More

Miniature tact switches save PCB space for home automation

Two sub-miniature tact switches, the PTS636 and PTS526, save PCB space for design engineers developing home automation and IoT electronic devices, says C&K.

The space savings electronics designers to add other components on the PCB. The PTS636 switch has a 6.0 x 3.5mm footprint and an adjustable thickness of 2.5, 4.3 and 5.0mm; the PTS526 has a footprint of 5.2 x 5.2mm and an adjustable thickness of 0.8 to 2.0mm. Both offer a large actuation surface for easier integration, for applications such as home automation, IoT devices and e-cigarettes, as well as control systems for items such as drones, e-bikes and robot vacuum cleaners.

The PTS526 and PTS636 switches are manufactured in surface mount technology. The PTS636 is additionally made in through hole format. Both use a hard actuator, ensuring that it can be fully integrated into standard processes, with no need for an additional interface button. This functionality reduces time and costs for an end equipment manufacturer, points out C&K. The switches provide optimum performance and retains high reliability up to 500,000 cycles.

The PTS636 design includes a 2.5mm height version with integrated ground pin and the PTS 526 design includes a version with an integrated ground pin. There is ESD protection soldered to the PCB for routing unwanted electrical overload.   

The PTS 636 is available in four different operating-force levels 130, 180, 250 and 320gf. There are three actuation forces for the PTS 526: 160, 260 and 320gf. Both sets of switches are rated at 50mA, have a bounce time of less than 10ms and an operating temperature range of -20 to +70 degrees C.

C&K was founded in 1928 and manufactures electromechanical switches. It has more than 55,000 standard products and 8.5 million switch combinations for use in automotive, industrial, IoT, wearables, medical, telecomms, consumer products, aerospace and point of sale (POS) terminals.

C&K products include tactile, pushbutton, snap-acting, toggle, rocker, detect, DIP, keyswitch, navigation, rotary, slide, switchlock, thumbwheel, smart card readers, high-rel connectors and custom assemblies.

http://www.ckswitches.com

> Read More

Multi-spectral imaging and AI enhances intelligence gathering

Condor MS, a long range oblique photography (LOROP) system by Elbit Systems, combines enhanced sensing capabilities and high capacity analytics. According to Elbit, it is “a step change in generating strategic targets”.

The system introduces multi-spectral (MS) sensing capability and artificial intelligence (AI) analytics to stand-off strategic intelligence gathering missions.

Three high resolution electro optic (EO) sensors are integrated into the company’s certified Condor2 system. These are visible and near infra-red (VNIR), medium-wave infra red (MWIR), and short wave infra red (SWIR). The combination of multi-spectral sensing, high level of stabilisation and auto image enhancement enables this Condor2 system to dramatically extend coverage area in day, night and adverse weather conditions to improve the strategic reconnaissance output and increase the survivability of the platforms. Deep learning algorithms and precise geo-location enable the Condor MS to identify a large number of targets at extremely high rates, which “significantly” shortens the time frame needed to close sensor-to-shooter loops, claims Elbit.

Elbit Systems has been the prime contractor in numerous LOROP programs based on the Condor family of systems, claimed to feature one of the smallest and lightest pods in the category. The Condor family of systems is integrated on board a variety of platforms, including the F-16 A/B/C/D/I, F-4, SU-30 and B-737.

Technology company, Elbit Systems, is engaged in a range of defence, homeland security and commercial programs throughout the world. The company operates aerospace, land, and naval systems, command, control, communications, computers, intelligence surveillance and reconnaissance (C4ISR), unmanned aircraft systems, advanced electro-optics, EO space systems, EW suites, signal intelligence systems, data links and communications systems, radios and cyber-based systems and munitions. It also upgrades existing platforms, developing new technologies for defence, homeland security and commercial applications and providing a range of support services, including training and simulation systems.

http://www.elbitsystems.com

> Read More

Renesas announces memory technology for AI

Renesas Electronics has developed an AI accelerator that performs convolutional neural network (CNN) processing at high speeds and low power.  A test chip with this accelerator has achieved the power efficiency of 8.8Tera operations per second per W (TOPS/W), which is the industry’s highest class of power efficiency, reports Renesas. The accelerator is based on the processing-in-memory (PIM) architecture, in which multiply-and-accumulate (MAC) operations are performed in the memory circuit as data is read out from that memory.

To create the new AI accelerator, Renesas developed three technologies. The first is a ternary-valued (-1, 0, 1) SRAM structure PIM technology that can perform large-scale CNN computations. The second is an SRAM circuit to be applied with comparators that can read out memory data at low power. The third is a technology that prevents calculation errors due to process variations in the manufacturing. Together, these technologies achieve a reduction in the memory access time in deep learning processing and a reduction in the power required for the MAC operations. As a result, the accelerator achieves the industry’s highest class of power efficiency while maintaining an accuracy ratio more than 99 per cent when evaluated in a handwritten character recognition test (MNIST), claims Renesas.

Before this development, the PIM architecture was unable to achieve an adequate accuracy level for large-scale CNN computations with single-bit calculations because the binary (0,1) SRAM structure was only able to handle data with values 0 or 1. Additionally, process variations in the manufacturing reduced the reliability of these calculations. The technologies developed by Renesas resolve these issues and can be applied to implement AI chips of the future and e-AI solutions for applications such as wearable equipment and robots that require both performance and power efficiency, says Renesas.

Since introducing the embedded AI (e-AI) concept in 2015, Renesas has defined classes based on the effectiveness of e-AI and applications that are implemented and has been developing e-AI solutions based on four classes: judging the correctness or abnormality of signal waveform data; judging correctness or abnormality using real-time image processing; performing recognition in real time and enabling incremental learning at an endpoint.

https://www.renesas.com 

> Read More

About Smart Cities

This news story is brought to you by smartcitieselectronics.com, the specialist site dedicated to delivering information about what’s new in the Smart City Electronics industry, with daily news updates, new products and industry news. To stay up-to-date, register to receive our weekly newsletters and keep yourself informed on the latest technology news and new products from around the globe. Simply click this link to register here: Smart Cities Registration